Cloudflare fetched the original article and streamed it through HTMLRewriter to apply an entirely new visual system without rebuilding the source page.
// request.cf · coarse context
A page that knows where it met you.
Only coarse request metadata is shown. This demo does not display or persist visitor IP addresses.
The WMF section of the village pump is a community-managed page. Editors or Wikimedia Foundation staff may post and discuss information, proposals, feedback requests, or other matters of significance to both the community and the Foundation. It is intended to aid communication, understanding, and coordination between the community and the foundation, though Wikimedia Foundation currently does not consider this page to be a communication venue.
Discussions of proposals which do not require significant foundation attention or involvement belong at Village pump (proposals)
This page is not a place to appeal decisions about article content, which the WMF does not control (except in very rare cases); see Dispute resolution for that.
Issues that do not require project-wide attention should often be handled through Wikipedia:Contact us instead of here.
This board is not the place to report emergencies; go to Wikipedia:Emergency for that.
Threads may be automatically archived after 14 days of inactivity.
Behaviour on this page: This page is for engaging with and discussing the Wikimedia Foundation. Editors commenting here are required to act with appropriate decorum. While grievances, complaints, or criticism of the foundation are frequently posted here, you are expected to present them without being rude or hostile. Comments that are uncivil may be removed without warning. Personal attacks against other users, including employees of the Wikimedia Foundation, will be met with sanctions.
Latest comment: 1 day ago200 comments50 people in discussion
The extremely expensive, buggy and ill-thought out "Abstract Wikipedia" has gone live after many, many years, and already it hosts an "article" which not only go against the very purpose of it (putting a complete English article in html is not the way to get automated Wikidata-based translation to 300+ languages), but also creates an unattributed enwiki copy[1]. Not that I am able to get it to load completely, it only returns a few sentences and then nothing happens (which is better than the many errors other pages generate, usually either "Reached max retries. Try again later." or "Wikifunctions returned a failed response: Error in evaluation" or simply "Page not found" when going from "history" to "read" or "page"...). Why this pre-alpha thing has been released to the world is not clear, why the WMF would think it conceptually is a good or feasible idea even less so. Fram (talk) 10:05, 25 March 2026 (UTC)Reply
Also see [2], it looks like it’s just going to be another Anglo-project, how tf are non-English speakers supposed to develop the wiki's policies when everything is only done in English. It’s a project that’s specifically not meant to be Anglo and is irrelevant to en.wiki. It should be put on ice until people can discuss via the functions or there’s some kind of multilingual support Kowal2701 (talk, contribs) 11:12, 25 March 2026 (UTC)Reply
Abstract + wikifunctions is 2 sides of the same coin. And if wikifunctions has this many issues still, then abstract shouldn't have been launched as a "beta". The most basic things don't work yet: e.g. take a random page, click on "edit source" or "view history", then click on "page" or "read"... error, every single time. This is not some obscure thing, this is basic functionality for the whole site, and it doesn't work. Type a page you know exists in the search bar (e.g. Cape Verde), the first result in the dropdown is the Abstract Wikipedia page (see the AW at the end), click it, and again you get the "page not found".
This is probably a simple switch somewhere, but the total lack of care displayed by whoever decided that this was ready to go public is staggering.
As for the multi-language aspect, the core business of Abstract... Q143, Esperanto.
"Esperanto is the languages of internationality. An Esperanto is a languages." (sic!)
In French, this gives "espéranto langue international" plus an error (the title doesn't get translated..., international(e) should be female, the verb has disappeared, ...)
In Dutch, it becomes "taal internationaal". No error, but missing most of the poor original.
Q21, "England is a country in United Kingdom." (sic). In Spanish, this gives "Unable to render this fragment due to an unknown error. ", in Swedish "The rendering service is temporarily unavailable. Please, try again later. " in Italian "Wikifunctions returned a failed response: Reached time limit in orchestrator", in Portuguese "Wikifunctions returned a failed response: Error in evaluation", and in Thai "England is a country in United Kingdom." I'm sure it will work wonders in the small languages for which it is intended though!
Most Western companies, and the WMF is no exception, would work better with half the people on twice the salary. But we would have to have a different attitude to work for that to succeed. Phil Bridger (talk) 20:07, 25 March 2026 (UTC)Reply
Thanks for the link. The team bios seem somewhat telegraphic. Are any of them "hot shot" programmers? And they seem pretty happy. In the commercial world, best software is written by developers who are pushed to the limit. But not for me to dive into details of the team. Yesterday, all my dreams... (talk) 14:40, 26 March 2026 (UTC)Reply
best software is written by developers who are pushed to the limit
Your word was "best", not most widespread. I heavily disagree that this is the Microsoft development process, and in any case, the market share of all Windows systems combined had been precipitously declining for years now because of how bad it is. The attempt by the WMF to do the development process you propose is Knowledge Engine (search engine) and the surrounding turnover. Aaron Liu (talk) 12:36, 30 March 2026 (UTC)Reply
As a former programmer who's experienced severe, life-impeding burnout from overwork twice in her career, I can't disagree more strongly with the notion that "the best software is written by developers who are pushed to the limit". — Hex•talk11:37, 28 April 2026 (UTC)Reply
best software is written by developers who are pushed to the limit. Assuming the limit here is a tight deadline, I would argue that software death marches do not result in great software. Tight deadlines result in insufficient time to write quality code. Quality is sacrificed to (try to) meet the deadline. Not to mention the impact on team morale and work-life balance caused by pressure and overtime. –Novem Linguae (talk) 22:51, 28 April 2026 (UTC)Reply
I checked my browser console when viewing the OP's link and there were over 65 requests to the API! As of last week, the WMF (I'm presuming a different team which didn't talk to this one) has decided to limit unregistered users to 1000 requests per hour. Total, across all projects. That includes search suggestions, DiscussionTools previews, VisuaEditor, etc. So about 30 page views for you, and then every project breaks unless you have an account! Suffusion of Yellow (talk) 20:12, 25 March 2026 (UTC)Reply
What I don't understand is how we are meant to make an "abstract Wikipedia" that is automatically translated to every language when the very design of the site is so centered around English. In abstract:Help:How to create an article I see a reference to a function called "Article-less instantiating fragment" which creates sentences like "Paris is a city". However, in some languages (e.g. Greek) such a sentence needs an article so the result will be ungrammatical unless a different function is used. Thankfully it seems like someone noticed this issue because the actual page for Paris, abstract:Q90, uses a different function called "defining role sentence" which doesn't have the same problem. But if basic stuff like this is wrong in the documentation, I don't have much faith in the project. In fact, pages like abstract:Q667 (south) still use this "article-less instantiating fragment" function. Warudo (talk) 11:36, 27 March 2026 (UTC)Reply
Actually, I found abstract:Abstract Wikipedia:Useful functions for article composition which apparently says that "article-less instantiating fragment" is a good fit for sentences like "Nairobi is a city". It absolutely isn't. Again, it's article-less in English and several other languages but not all of them. While I'm not a linguist myself, it really feels like this system was designed without consulting experts. Warudo (talk) 11:40, 27 March 2026 (UTC)Reply
It seems to me to be referring to the fact that it doesn't require an article in English. It may or may not require an article in any other language you care to mention, so this seems to be a very anglocentric point of view. Phil Bridger (talk) 17:32, 28 March 2026 (UTC)Reply
No, that's definitely not it! Article-less and article-ful have different semantic meanings (the term article here seems to confuse a lot of people who do not understand the project, which I suppose means it needs a rename). All of these examples would use "article-less" (despite the fact that two of them have indefinite ones):
Golf is a sport
El golf es un deporte
The United States is a country
യുണൈറ്റഡ് സ്റ്റേറ്റ്സ് ഒരു രാജ്യമാണ്
And all of these examples would use "article-ful":
A bird is a dinosaur
Un ave es un dinosaurio
These are saying two fundamentally different things irrespective of language. Wikifunctions is already equiped to handle this distinction. Feeglgeef (talk) 18:28, 28 March 2026 (UTC)Reply
I don't see any great difference between "Golf is a sport" and "A bird is a dinosaur" apart from the presence/absence of an article in English. Please explain. Phil Bridger (talk) 19:18, 28 March 2026 (UTC)Reply
Then the name of the function should refer to proper and non-proper nouns, rather than articles. The point was that the name of the function is based on English language thinking. TietoTeekkari (talk) 07:58, 4 June 2026 (UTC)Reply
"Article-less and article-ful have different semantic meanings". But then why are these semantic meanings not mentioned in the documentation of f:Z26039? Instead the function's documentation says Makes a sentence of the form "X is a Y" e.g "Nairobi is a city.", i.e. it takes an entity (X) and its class (Y) and states that it is an entity of that class. What are the semantic differences with f:Z26095? In case you think this is a small problem, it really isn't. Look at the Japanese translation of this documentation (a language that does not use articles). The translator had to use an English example to explain what the function does because in their language the concept does not really exist. Indeed, as far as the users are concerned, these functions are in fact defined in terms of whether the sentences they generate require an article in English. The semantic difference behind that is hidden to them. Warudo (talk) 22:33, 28 March 2026 (UTC)Reply
Also, by the way Abstract:Q30 currently says "United States is a country. United States is a republic. Washington, D.C. is the capital of United States." So, I guess the article-less function was not the correct one to use in this case. Warudo (talk) 23:36, 28 March 2026 (UTC)Reply
I've already said this above, but the "article" being referred to here is a indefinite one (a/an) at the front. definite articles (the) have no semantic meaning and therefore are to be added on a language-by-language basis. Feeglgeef (talk) 00:40, 29 March 2026 (UTC)Reply
Definite article = "the", Indefinite article = "a/an". Also that's not true. The definite article has semantic meaning. It means that we are referring to a specific member of a group and not to the concept in general. In this case, "United States" means any states that are united, which would also include e.g. the Mexican United States. The United States on the other hand are a specific set of states that are united, in this case, the United States of America. Warudo (talk) 00:49, 29 March 2026 (UTC)Reply
We already have the distinction between the concept of a federation and the example in North America seperated by Wikidata. Some languages (like Malayalam, as in the example) do not use an article in front of the United States. Eventually, the function will be able to handle automatically adding a definite article. Feeglgeef (talk) 14:09, 29 March 2026 (UTC)Reply
Ok, to be fair, you're right. The generated sentence is wrong but that can be attributed to an incomplete implementation of f:Z26039 rather than a logic error in abstract:Q30 itself.
Still, there are several questions that remain. First, why are the semantics of the two functions defined in terms of English grammar? You said, Article-less is about one thing, article-ful is about a collection. but as I've said above, that's not how it's described in the documentation. But more importantly, if that's the difference between them then why are two different functions for "article-less" and "article-ful instantiating fragment" given to the user in the first place? Why don't you expose a single "instantiating fragment" function that checks if its input has a subclass of (P279) statement in Wikidata? If yes, use an indefinite article, if not, don't use one.
Just FYI, editing your own comments after someone has already answered without indicating the changes is considered bad practice in the English Wikipedia. Not a big deal in this case as you were just fixing a mistake but keep it in mind in the future. Warudo (talk) 14:47, 29 March 2026 (UTC)Reply
@Feeglgeef no, your statement is plain wrong. This is how 6 sentences above will be translated to Russian (browser machine translation, but it's a correct translation): As you can see, all 6 cases uses exactly the same grammar construction, because Slavic languages has no articles and they don't provide meanings, that is provided by articles in Germanic languages. We, native Russians, believe that this semantic meaning simply does not need to be conveyed. MBH (talk) 18:24, 21 May 2026 (UTC)Reply
You have to accommodate the languages that don't, though. The names of the function have been thankfully changed now to "subject is instance of" vs "class is subclass of". These are different things, even if they look the same in Russian. Feeglgeef (talk) 18:30, 21 May 2026 (UTC)Reply
All of other points other people mentioned aside, many languages have exceptions for articles. For example, in German proper nouns don't get articles (noone says "Der Deutschland") but Turkey for example is "die Türkei" (a similar example "The Netherlands" which also takes article in English too). I tried it and abstract:Q3640 produces "Ankara ist die Hauptstadt von Türkei." while the article in German Wikipedia (de:Ankara says "...die Hauptstadt der Türkei.". Ignoring the wrong "von" in the produced text, it is missing the article for Turkey as well (I copied the functions for Q3640 from abstract:Q61, so my apologies if I copied the wrong functions). I don't know how AW works in depth so my apologies if I'm misunderstanding anything. Ladsgroupoverleg16:44, 16 July 2026 (UTC)Reply
The copy of enwiki is attributed now, and the project is not doomed because one user fails to abide by copyright law or to make a quality article.
As for the beta release, it's impossible to debug or improve without community content, so I'm not sure what you'd have the WMF do. Feeglgeef (talk) 17:03, 28 March 2026 (UTC)Reply
"one user fails to abide by copyright law or to make a quality article." Not a single user has made a "quality article", which is hardly possible with the current setup. And the actual intention of the project, automatic translation to small languages, is just not happening.
Random "article"[4] translation to French: "Wikifunctions returned a failed response: No matching lexeme for item in language"
Random "article"[5], translation to Italian: "Wikifunctions returned a failed response: Number of arguments mismatch"
Random "article"[6] translation to Spanish: "(en) Bahrain is a country in Middle East." Hey, I can understand Spanish!
Random "article"[7] translation to Dutch: "Brussel is the hoofdstad of België." THE hoofdstad? Yep, clearly not a problem with the use of articles...
Random "article"[8] translation to French: "Paris est une ville. Wikifunctions returned a failed response: Error in evaluation" Translating two sentences was a challenge of course.
Random "article"[9], no translation: "Wikifunctions returned a failed response: Could not acquire WASI runner within time limit" and "Wikifunctions returned a failed response: No matching lexeme for item in language" and "Unable to render this fragment due to an unknown error."
You state "As for the beta release, it's impossible to debug or improve without community content, so I'm not sure what you'd have the WMF do." which is absolute nonsense. You don't do a public release of such extreleky buggy software, and you don't call it a beta either. The errors found so far are not edge cases where mass testing is necessary, but things a developers + QA team should easily have found. WMF should, for a $6 million + project, have some people on board who understand what this project is intended for and can test it before it is released as a beta. The release of a severely immature project where even the most basic fundamentals are being questioned is irresponsible. Then again, the decade-long development seems to have been irresponsible as well. Fram (talk) 10:19, 29 March 2026 (UTC)Reply
Fram, you have more than enough examples now. Item 17 is disastrous and tells me that the entire system needs a rewrite following a redesign of the architecture. But that would be throwing good money after bad. I predict that there will be no remedy for this project anytime soon. Once the underlying software architecture has problems 1 million bandaids placed on it will be no cure. Yesterday, all my dreams... (talk) 12:32, 29 March 2026 (UTC)Reply
None of these problems are architectural. All of these are the result of the work of a few (like less people than that participated in this topic) community contributors. Feeglgeef (talk) 14:17, 29 March 2026 (UTC)Reply
So there is no Wikidata item for "is a" in French, but this thing will be able to handle translations to languages with very few editors somehow? And "it knows how to say the words in Dutch", er, no: "the" (or "is the") is English, not Dutch. Again, it can't even translate that most basic building block to a language with a large editing base. As for 19, I have no idea how "unknown error" and a failed time limit are the responsibility of Wikidata contributors, nor how a translation tool was ever tested by the developers if the most basic aspects are missing in French, Spanish, Dutch, ...
Someone at the WMF was aware that translation (and specifically translation to languages with very small user bases) was the intention of this tool, right? Because it sure doesn't look that way. I don't see how they can have tested this at an alpha-level to give it the green light to go to beta, if if can't even handle these basic things. Blaming it on Wikidata contributors is rather rude, the developers/testers should have added things like "is a" or "the" or ... in major languages (both the ones I just tested, but also completely different ones with other scripts and grammar) to Wikidata. I assume these people have some fluency in Wikifunctions and Wikidata editing, and in languages and translation? Seems a prerequisite for such a project... Fram (talk) 15:30, 29 March 2026 (UTC)Reply
Question: How many people who work on Google Translate are linguists? Please answer to yourself before you read further. Answer: zero. We have all learned long ago that fiddling with linguistic constructs will end in one place: the shelf that holds the Aspirin bottle. So please do not assume that as a requirement. Their problems are much deeper and architectural in nature. They should have never used Wikifunctions, given that they are crowd sourced. Alas the key issue is that we can all huff and puff for ever but we have no control on what WMF does. So maybe we should all take an Aspirin and move on. Yesterday, all my dreams... (talk) 16:22, 29 March 2026 (UTC)Reply
Google Translate is not entirely human-made though, but is based on computer-learning (statistics, deep learning and brute force basically). Abstract is based on "humans will build it all", which, while admirable, then requires humans with very specific skills. And "we can all huff and puff for ever but we have no control on what WMF does" is false, we have forced them to shelve things like Flow and Gather. Fram (talk) 16:49, 29 March 2026 (UTC)Reply
I know exactly how G-translate works. Thanks. My point was/is that the crowd sourced paradigm works for text input but not for software. The world is moving towards automatic software generation now, so crowd sourcing will be inherently inefficient and error prone. But I think I have said enough now. No more comments from me here. You are right in objecting to the project but time will tell how much power you have over WFM. Cheers Yesterday, all my dreams... (talk) 22:57, 29 March 2026 (UTC)Reply
How many people who work on Google Translate are linguists? Why are you using Google Translate as an example? Do you know how many mistakes Google translate makes when translating to and from languages other than English? For some reason English homographs throw it off completely. I've seen it do stuff like this where I asked it to translate a verb that means "to bear" and it came up with a word for the mammal. Even ignoring the fact that English is clearly used as an intermediate language here even though it isn't suited for this purpose (probably not by design but because of the way Google translate was trained), why on Earth is Google translate translating a verb as a noun in the first place? I think stuff like this shows how Frederick Jelinek's quote is outdated and is leading us astray at this point. Warudo (talk) 16:53, 29 March 2026 (UTC)Reply
There is no implementation of the function in Dutch. The Abstract Wikipedia team has not added an implementation in Dutch because the community is responsible for creating the function in Dutch, and nobody has created the function in Dutch.
Essentially, how the current English implementation is to string together the first concept with "is the" with the second concept with "of" with the third concept. The function can get the Dutch terms for the concept, but it does not know how to string the words together because nobody who speaks Dutch has told it how. This is not something that the WMF can magically fix. Eventually, when the project is older than a week, somebody will implement it in that language, and in Malayalam, and Dagbani, and Massa, and Southern Altai, and Dusun. Feeglgeef (talk) 21:49, 29 March 2026 (UTC)Reply
Please see my last response to Fram. Anyway, time to cool off and move on before someone busts an artery here. You will be glad to know that I shall make no further comments here. Now, in what language shall I say goodbye? Yesterday, all my dreams... (talk) 23:02, 29 March 2026 (UTC)Reply
"when the project is older than a week"? It's a decade old or thereabouts, Wikifunctions specifically was created in 2020 and launched in 2023. But sure, some Southern Altai Wikipedian will go to Wikidata to translate everything that is needed, then go to Wikifunctions to translate all necessary functions into the grammatically correct version of their language (assuming naively that the used function can be one-on-one transformed to one in their language for every use of it), just so they can then autocreate stilted article stubs instead of either writing them directly, which would require a lot less effort and give a lot more satisfaction, or using an online translation tool to translate an existing Wikipedia article to give them much easier results. Totally realistic. Fram (talk) 07:32, 30 March 2026 (UTC)Reply
When I go to this totally not-alpha project, to the page we are discussing[10], and click on "defining role sentence in English as string" (which should apparently go to abstract.wikipedia.org/view/en/Z28109), I am taken to the Abstract Wikipedia Main page[11]. Please explain to me again how this has been sufficiently tested and was ready to be opened to the wider public? Fram (talk) 08:39, 30 March 2026 (UTC)Reply
"by putting several sentences into a single paragraph, the paragraph as a whole is being run, may cause time-outs, and will be cached. Instead, if, for now, you put one sentence into each fragment, caching and evaluation can be more spread out and should allow for more content. Eventually we want to fix that"[12] Gee, why would you fix the bug where putting more than one sentence into a paragraph makes it even more likely you will get a timeout error? Fram (talk) 15:55, 30 March 2026 (UTC)Reply
Probably a few years of pointing out both the immediate and the fundamental issues. The work of developers, testers and product managers. Once the grant and endowment money stops flowing, and some quarterly or yearly goals can be checked on some paperwork, the drive to continue this will stop. At best/worst to keep whatever exists at the time running and let some volunteers play with it for a few more years before completely stopping it (see the soon to be closed down Wikinews). At least with Gather, Flow, ... we could point out that it was actively, directly negatively impacting enwiki (and other wikis): here it is only money and developer time disappearing down the drain. Fram (talk) 14:16, 31 March 2026 (UTC)Reply
For what it's worth I would like to see them continue as the issues pointed out do not seem to be systematic (with the platform design) but an incredibly horrible implementation of it. I do agree that the state of the project seems at best alpha. Aaron Liu (talk) 14:37, 31 March 2026 (UTC)Reply
The systemic issue is the belief that English grammar rules can be copied one-on-one to all other languages. We have e.g. functions for "use this the" and " use with a/an" and "use without either", which even within English is problematic (e.g. sometimes you need X is the capital of Y, and sometimes X is the capital of the Y, like with United Kingdom): but in other languages half the cases of a certain English function may use one construction, and the other half uses another construction, and this needs somehow to be built into the simply English function. I'm simplifying things here, but I hope yo get my drift.
Purely on a word level this whole construction works somewhat theoretically, but requires a massive amount of work which is exactly the problem for the small languages where this is supposedly built for. On a sentence/paragraph level though, I don't believe this will ever work (for simple cases for related languages, yes, but not in general). If this is pushed through regardless, we will probably end with new Scots and Greenlandic version catastrophes, but then on a larger scale. The setup and performace issues are just the icing on the cake. Fram (talk) 15:05, 31 March 2026 (UTC)Reply
The ideal would be to store everything in context-free language, and then generate sentences in the target language by applying a set of rules. When I was a grad student in Linguistics lo these many years ago I wrote my dissertation on one aspect of Deep structure and surface structure. From my experience trying to figure out what some of the rules are in (my idiolect of) English for a limited subset of syntactical structure, it will take a very large and hard to maintain set of rules with extensive exceptions just for English. My mind boggles at the concept of doing that for all the currently spoken languages of the world. Donald Albury16:35, 31 March 2026 (UTC)Reply
Indeed. At the moment, they produce the article [13] "New Jersey is an U.S. state." I presume the "an" comes from an "an before a-e-i-o-u" rule, which doesn't deal with the many exceptions to that rule. And this is a very simple example, in the main development language. Fram (talk) 16:45, 31 March 2026 (UTC)Reply
The systemic issue is the belief that English grammar rules can be copied one-on-one to all other languages. You'd think that people would have learned from the failure of this approach when it was applied to Wikidata. Some editors wanted a property to express the relation "X is the mayor of Y". So they just made a property called "of". They then found out that the property was not only difficult to translate to certain languages but it also evolved into a monster that modeled many different, sometimes contradictory relations (which is kind of bad when the whole point of a database is to be machine readable) and it ultimately required a huge effort from the community to get rid of it.
Yet now, the abstract Wikipedia editors are doing the same things, defining their functions in terms of English grammar constructs and not the underlying logical relations those constructs represent. Warudo (talk) 17:21, 31 March 2026 (UTC)Reply
All of these rules assume that an English grammar rule is also 1 rule in another language. "Article-less instantiating fragment" will sometimes need to be translated with, and sometimes without an article (even if the remainder of the grammar is the same). If this can be done with one function, then there was hardly any need to have different functions with or without article in English surely? And this is a very simple and basic example. So how is this solved? Fram (talk) 20:28, 1 April 2026 (UTC)Reply
Article-less does not mean what you think it means here. The name is confusing and should probably be changed, but, for example "Golf is a sport" and "El golfo es un deporte" (both use article-less, despite the fact that the latter has an article) have the same meaning, but "A bird is a dinosaur" and "(The) bird is a dinosaur" have two very different meanings, even if, say, Bulgarian does not make the distinction. The distinction between article-less and article-ful is not actually articles. Feeglgeef (talk) 20:37, 1 April 2026 (UTC)Reply
So if you start from an article which doesn't make a distinction, and go to a language that does make a distinction, you're screwed? If in your example the base language would have been Bulgarian, and you went to English, sometimes the Bulgarian function should give "the" in English, and sometimes "a", which depends on context. And all of this is still between very comparable languages basically. If the base article is for some sentence / meaning "article-less" and the target language "article-full" (or vice versa), you have a problem, no matter how you call these functions. Fram (talk) 20:48, 1 April 2026 (UTC)Reply
I get the feeling that Feeglgeef is a little out of their depth here, but there's nobody with any decision-making authority at the WMF willing to rescue them. Phil Bridger (talk) 21:52, 1 April 2026 (UTC)Reply
Based on your replies in this thread, it seems you are unwilling to assume good faith of either your conversational partners or the functionaries (pun intended) of the wiki in question. Why not just ignore it, if you feel it is consigned to failure? Arlo James Barnes19:09, 11 April 2026 (UTC)Reply
You actually don't "start from an article which doesn't make a distinction", because all Abstract Wikipedia articles are supposed to be abstract. All articles are required to make the same distinction and all distinctions necessary for every single language (even if some languages ignore them) because they are all written in abstract language. There is no English or Bulgarian base, nor translating here. Feeglgeef (talk) 22:07, 1 April 2026 (UTC)Reply
That, again, makes no sense. Because in English, "a bird" and "the bird" have two different meanings (but refer to the same Wikidata item), you need a different function than for "golf" (the sport), which has in English one meaning for that Wikidata item. But still you claim that the functions, the whole approach, are language-independent, as if these issues in English are the same across all languages for the same words. If Abstract Wikipeda were truly language-independent, you wouldn't need the article-less and article-full functions. And that's still only at the word level, and doesn't go into sentence- and paragraph structures and countless other quirks, irregularities, ... Anyway, it looks so dumb that after all this time, apparently there isn't a function yet to start sentences/articles with an article; we get things like "Bible is a religious text." for an article specifically about the Judeo-Christian Bible (not about the general word). And finding out how things actually work for Abstract articles is very opaque as well, I have no idea where the "suns" instead of "stars" comes from in [14] "Stars are sources of light. Stars contain metals. Suns shine."
Oh well, the WMF team doesn't respond here, but they seem to read it, as some of the most stupid errors get fixed after they are reported here at least. Fram (talk) 07:44, 2 April 2026 (UTC)Reply
Wikipedia has articles about more than 20 million topics in more than 300 languages. But none of these languages alone allow access to the knowledge about these 20 million topics (...) Abstract Wikipedia does that without relying on AI. Each step of the way remains under human control, and is accessible and editable by the volunteers. I doubt volunteers will create articles about 20 million topics in "this thing" (and with 108 active editors, as there are now, it seems completely impossible to arrive at any place).
Looking at this article, it seems that a software-generated summary (yes, there is software other than AI) automatically combining some data from all language editions of Wikipedia, plus Wikidata and Commons, and using machine translation when needed (for machine translation, AI usage is usually not bad, and free and open source AI software also exists), would do a far better job.
In addition, if the point is to avoid AI, I don't understand why it says "Generated text" (who generated it?). Sorry if I am too critical (of course, I appretiate the efforts of the people who are working on the project), but I don't believe this adds anything new to the Wikimedia ecosystem: it looks as made for consumption by a machine, not by humans, so maybe it would be better if it was also mostly machine-generated.
The idea of joining the knowledge from all Wikipedia language editions (plus Wikidata and Commons, and, for some articles, also Wikisource, Wikivoyage, etc) is a very good one (and one I had thought about many times), but I think that a "global search" feature with integrated machine translation, and some way to access the combined data from multiple sources, would be the only way to achieve that. MGeog2022 (talk) 19:46, 8 June 2026 (UTC)Reply
I doubt volunteers will create articles about 20 million topics. Yes, exactly. There's no chance we ever surpass enwiki or other major language Wikipedias, as abstract articles take much more time to write than normal articles, and there's much less chance for onboarding new contributors.
In addition, if the point is to avoid AI, I don't understand why it says "Generated text" (who generated it?). Sorry if I am too critical (of course, I appretiate the efforts of the people who are working on the project), but I don't believe this adds anything new to the Wikimedia ecosystem: it looks as made for consumption by a machine, not by humans, so maybe it would be better if it was also mostly machine-generated. It was made for the consumption of humans who speak only small-to-mid sized languages.
Machine translation might be worth consideration for a future project, it's been used, for example, on Caesar DePaço to circumvent a court order, but writing content in an abstract (instead of concrete) language means we don't have to deal with information loss (see the whole "Google translate 100 times" thing).
If you click through the endless errors and finally try to see a translated article, you get monstrosities like "Ein Äpfel ist eine Frucht."[15] or even worse " Bisexuelles lieben Fraus. Bisexuelles lieben Manns."[16]
People from the WMF, could you please clarify: before releasing this as a supposedly beta product, which tests did you run? Which articles have you created, with which functions, and tested for which languages? Fram (talk) 13:16, 8 April 2026 (UTC)Reply
@Sannita (WMF): as someone who seems to be closely involved, but of course feel free to put this through to whoever is better placed to answer this. I also notice that most activity on Abstract seems to come from an editor who was indef blocked on enwiki for CIR / timewasting, and is now running some AI-generated tool to create non-working pages (like this 193K monstrosity[17]-) and to change working pages (no matter how bad they were) into non-working ones (e.g changing this into this ("Wikifunctions returned a failed response: Error in evaluation"), across a lot of pages. While I think the project should be abandoned as a waste of time and money, it shouldn't be done by mass-vandalizing the work of the editors there. Fram (talk) 16:30, 10 April 2026 (UTC)Reply
@Fram We did some limited tests in a controlled environment, but doing so can only help you so much in identifying potential problems. We are learning a lot by releasing the beta project (because this is still a beta), and we'll improve from there. Sannita (WMF) (talk) 10:28, 11 April 2026 (UTC)Reply
Thanks, but no, this is not a beta, this is barely an alpha, the thing is unworkable for all but the simplest sentences, terribly slow, had the most basic errors when released. Using volunteers as cheap/sheep testers on a $6 million+ project which hasn't been thought through, hasn't been tested, and has in its utopic, unrealistic ideals been overtaken by reality anyway, is old school WMF which I hoped had been left behind after previous such failures.
After the omnipresent "Reached max retries. Try again later.", you get a plethora of errors. A 2-sentence "article" gives "Wikifunctions returned a failed response: Reached time limit in orchestrator", basic English words give "Wikifunctions returned a failed response: No matching lexeme for item in language" (but this will work for languages with barely any editors somehow), other articles give "Wikifunctions returned a failed response: Error in evaluation", "Wikifunctions returned a failed response: Could not acquire WASI runner within time limit", "Wikifunctions returned a failed response: Reached rate limit in orchestrator"
And the things that do "work" give results like "Australian continent is a continent in the Earth. Australian continent contains Australia." (German translation of that last line: "Australien contains Australien.") Or the extremely basic issue that when you translate an article, you would expect the title of the article to be the first thing that gets translated, even in alpha-stage. No such luck. This thing is supposed to be used to create articles and translate them into manu languages, but not a single decent example has been produced so far. A "beta" product which simply can not produce an acceptable end product just isn't tested to even the most basic standards and should never have been released. And a project where the actual requirements don't seem to have been thought through, and where the results (if the wanted end result was ever reached) would probably make the Greenlandic and Scots disasters look like minor blips, should have been stopped much, much earlier, before so much money and time was wasted. Fram (talk) 17:44, 11 April 2026 (UTC)Reply
"両性愛者s 愛し 女性s.両性愛者s 愛し 男性s." If this is what it's producing for a language with over 100 million speakers, what's it doing for small languages? Sesquilinear (talk) 23:26, 9 July 2026 (UTC)Reply
That is hardly surprising. The function used here is English simple present collective sentence. Of course it does that. This, again shows how over their head some editors over on the Abstract Wikipedia are. When the whole point is to make a cross-language encyclopedia, you cannot directly call functions that only work for one language. Those should only be called internally. What makes this worse is that the language neutral function simple present collective sentence (creation date: 24 March) already existed when the bisexuality article (creation date: 4 April) was created. This means that the editor who made that article had a choice between "simple present collective sentence" and "English simple present collective sentence" and chose to use the latter in a multilingual project! For what it's worth, I fixed it. It now says "両性愛者は女性を愛する。両性愛者は男性を愛する。" which is at least grammatical. It took about 2 minutes. It would have taken 30 seconds if the abstract Wikipedia editor didn't overwrite the arguments of a function call when you change the function. Warudo (talk) 23:55, 9 July 2026 (UTC) (edited at 00:15, 10 July 2026 (UTC))Reply
If someone had to come up with a clever way to disrupt Wikipedia by wasting editors' time, I cannot think of a better way than creating Abstract Wikipedia. It feels like a clever sabotage operation Ita140188 (talk) 12:27, 10 July 2026 (UTC)Reply
According to f:Z32530, no, there isn't one. That is a separate problem to the one I'm trying to solve at the moment. When I tested on Wikifunctions I got this:
I can't really fix that because I don't speak German and can't write the gloss for the Sense entity. While I have come across Senses that don't have a gloss in the Lexeme's language like d:L:L1082283#S2, I don't think it's a good idea to make those, so I'll leave this one as is. Warudo (talk) 16:29, 10 July 2026 (UTC)Reply
I just don't get how they are going along and spending money on something so ill-thought out and poorly executed. Let's ignore for now the basic questions of "is this feasible, which competencies or expertises do we need to think this through, what is the best approach", the slight issue that reality (LLMs) has somewhat changed the whole environment this needs to be thought about, and the basic recurring problem that a completely untested, very buggy environment has been released as a beta for everyone to play with without any guidance. But why are they (WMF and editors) now going further with this in the most inefficient way possible? Everyone creates whatever they like, 99% of what is being made is absolute rubbish that serves no purpose at all. People are creating one- or two-sentence stubs for all countries which all have the same issues.
A logical, productive way of dealing with this project (apart from the most logical one of pulling the plug) would be to start with one article, take e.g. the lead from enwiki (or dewiki or whatever), and build the necessary structure (Abstract + Functions + Wikidata) to create this and translate this in 5 wildly different languages. Step by step, sentence by sentence, until you have a basic set of functions for this kind of article, and have an idea if it will work.
Second steps might then be either testing the same for a related article (say, test 1 was a country, take another country and see if the functions are all transferable or if there are things you missed), or doing the same "build from scratch" for a different topic (say, a biography).
That way, you build a structure, a set of reusable and needed blocks you can refine later on, and at the same time learn a lot of "don't do this" pitfalls and issues. And after you have done this for the 5 or 10 most common types of articles, you actually have a tested, usable, beta environment (or you have realised it won't work at all, or it will work in theory but the work to get things right for a more obscure language isn't worth the hassle).
Instead, we get articles[18] directly using the function "English plural"[19], really useful for an abstract, language-independent project. Or more commonly "articles" consisting of random repetitive sentences like "Reproduction is a biological process. Reproduction is a type of process. A reproduction is a biological process. A reproduction is a creation. Animal reproduction is the part of of reproduction. Plant reproduction is the part of of reproduction. Human reproduction is the part of of reproduction."[20]
Oh well, at least I learned that "An information is a knowledge."[21] or "biseksualiteit ∈ {seksuele oriëntatie}" in Dutch or still "Bisexuelles lieben Fraus. Bisexuelles lieben Manns." in German [22]. Note how here and on all other pages, the actual title doesn't get translated? This is not something the article creators can help, this is something which should have been included by the WMF as a basic element (assuming they realised what Abstract Wikipedia was intended for) but is missing.
Anyone minimally familiar with the history of Linguistics or Computer Science as disciplines should be aware that the core premise of the project was investigated and discarded decades ago (and anyone who has studied at least one language that isn't super closely related to their native language should be able to quickly come to similar conclusions). There is no universal deep structure across languages. Why was this ever greenlit? This is about as useful as funding new studies in Lysenkoismsigned, Rosguilltalk05:51, 3 June 2026 (UTC)Reply
When I first looked at the project, I just thought...oh, here's another attempt to make an interlingua. They've all failed before because it fundamentally misunderstands how meaning works, but ok, sure, do whatever you want, maybe you'll succeed this time... Erynamrod (talk) 12:42, 9 July 2026 (UTC)Reply
The WMF has launched a dashboard to follow the progress of Abstract Wikipedia [23]. It has e.g. a list of articles which work in any language! Well, most of them have no actual text translated through functions, just Wikidata items without any sentence-building, so yes, these work, they just aren't articles... The ones that do try to have actual sentences and supposedly work are e.g. Brussels[24], full text "Brussels is the capital of Belgium." This gets translated as "Brussel is the hoofdstad of België." in Dutch, which isn't correct Dutch. "Bruxelles is the capitale of Belgique." in French is equally wrong, as is the German "Brüssel is the Hauptstadt of Belgien." It doesn't work at all in e.g. Romanian, Moldavian. Anyway, I guess the function "defining role sentence in English as string" should perhaps not be used in Abstract Wikipedia, and items which use it should not be said to be working in any language, as the naturally don't. Something like "list of cities in Belgium"[25] gives a result in English, and keeps running endlessly when I try it in Dutch or French. So not really "working". It looks as if not a single actual article can be said to be working in all or even most languages, making the dashboard rather meaningless. Fram (talk) 11:49, 8 May 2026 (UTC)Reply
From reading the Abstract Wiki, it also seems like there is no way to utilize the past tense, which um… seems a bit critical in an encyclopedia. ExtantRotations (talk) 18:34, 8 May 2026 (UTC)Reply
On the plus side, they have introduced endless repetition, which makes for much better reading. At the moment, the article for New Year's Day reads in full: "New Year's Day is a public holiday. New Year's Day is part of the public holidays in Australia, public holidays in Australia, public holidays in Australia, public holidays in Australia, public holidays in Australia, public holidays in Australia, public holidays in Australia, public holidays in Australia, public holidays in Australia, public holidays in Australia, and public holidays in Australia." Fram (talk) 12:10, 22 May 2026 (UTC)Reply
Somehow, they have made this dashboard worse. It now has a section for "Articles ready to publish in English". What this means is "Wikidata Qnumbers we have created in Abstract but don't exist in enwiki", not that these pages are in any way either wanted in enwiki, nor that they are actually "articles" which are "ready". This includes things like [26] "Evaluating the biological validity of European river typology systems with least disturbed benthic macroinvertebrate communities", ne of the thousands (millions) of scholarly articles with a Wikidata entry, but also [27] Blake Lemoine, apparently a software engineer, where the complete "article" exists of three identical links to research.google, but the link gives a "page not found". Ih yes, this is so ready to be published to enwiki. Other "articles" we are apparently missing is [28]"interaction", full text "An interaction is an effect.An interaction is an effect.An interaction is an effect." Oh look, a Picasso painting we are missing[29]. Actually, no, we have it at Three Musicians (Picasso)
I don't even want to try to guess what the next section on the dashboard, "Quick wins in English", is supposed to be about. "Add a label to California"? "Pending fixes in English", explanation "All items where English is the blocker", e.g. "photon" needs a label.
Mind you, that's not even the most bizarre or optimistic thing on that dashboard. At the very bottom, there is a section for "Almost there — English is the only thing missing
Already usable in the most other languages. One English edit gives each near-universal coverage." Well, perhaps not. [30]Giza is not one English edit away from anything, Giza (or any Abstract "article") shouldn't use the function "defining role sentence in English as string" because "in English" is not really what Abstract is about.
While for English these things won't be published here and we see the problems from miles away, for small languages (the intended target) this is another dramatic Scots Wikipedia scenario waiting to happen, but this time pushed directly by the WMF.
According to the latest newsletter about Abstract, all that is left is improving Abstract and integrating it into language Wikipedias (the horror), but as of now, "Abstract Wikipedia can compose articles from functions on Wikifunctions". Bizarrely, despite this claim, not one actual article has been composed, only a few collections of loose, short, often ungrammatical sentences, which in a few cases can even be translated in even worse sentences in very few languages. But oh well, what can one expect from a newsletter which claims "Abstract article pages now show the label of the Wikidata item alongside the QID in the page title. " when in reality, for a week or so now, what is actually shown next to the Wikidata label is the letters "ltr" in grey. When you click on them, you get the text "copied!", and you have literally copied the text "ltr". Brilliant! Even things that barely worked before have since become worse, the page "Paul Cézanne"[31] now shows "<a href="https://abstract.wikipedia.org/wiki/Q17277950">The Card Players</a>" instead of the links to Wikidata it had. Also errors like "Reached max retries. Try again later. Retry" have reappeared.
Let's be clear: "Franz Schubert is a composer in Austria. Franz Schubert is a pianist. Franz Schubert is a teacher." is not an article. "Wikifunctions returned a failed response: Invalid key" is not an article. "Organism is a part of of population. Organism is a part of of group of living things." is not an article. "Roman Empire is an empire." is not an article. "2 is a small number." is not an article. "List of French artists is a Wikimedia list article." is not an article (why would anyone think that this is the kind of text wanted in any language wikipedia?) "August is a calendar month. August is part of the Swedish calendar, Swedish calendar, and Swedish calendar." is more than an article, it's a mantra. "A Citrus × limon is a useful plant." is the full article about the Lemon.
And the translations... "Organisme is an onderdeel van of populatie." is not a Dutch sentence. "2 is een ." is not a Dutch sentence. "Franz Schubert est un enseignant ou enseignante." is what you get when you start from English as the norm for all languages. "August is a calendar month. August is part of the schwedischer Kalender, schwedischer Kalender, and schwedischer Kalender." is supposedly a German article. "{citroen} ⊆ {nuttige plant}" is supposedly a Dutch article.
I clicked on random article, the first one I saw was Homer:
Homer is a poet. Homer is an author. Homer is a writer. Homer is a human whose existence is disputed. Homer is a conceptual character. Homer is the part of of Greek mythology.
What I don't get is, they don't have support for past tense yet. Whatever, they'll add it eventually, I hope. So, why are they making abstract articles that need it? Warudo (talk) 20:35, 1 June 2026 (UTC)Reply
Partially reminds me of caveman speak: If horse go walk, cart also go walk. You see, force from horse leg on ground push on rock, then normal force push horse leg in opposite direction, cause horse to move small amount, for each time horse leg push on rock. Rope tied to horse neck move with horse, and is connected to cart with wheels. Cart wheel spins if moved in same direction as wheel is facing, which glide across rock, but still exert normal force due to gravity. Therefore, if horse go walk, cart also go walk.
For what it's worth, the Homer abstract article was created by the editor who did this. As harsh as it may be to say, this is the level of quality we are used to seeing from this editor, both here and in the abstract Wikipedia. Warudo (talk) 19:04, 2 June 2026 (UTC)Reply
They were blocked for CIR/wasting community time and (how do I put this politely) I couldn't agree more with that description. Anyhow, most of the slop problems Fram is pointing to were caused by them, so I don't see how that justifies shutting down the whole project. Feeglgeef (talk) 03:51, 3 June 2026 (UTC)Reply
Not true though. Going over the issues in my above posts, the "ltr" instead of the Qnumber is obviously a WMF technical mistake (and a failure of whoever wrote the newsletter). The "articles" I randomly picked: Paul Cézanne "2", August, the Picasso painting, New Year's Day, Giza, ... all not edited by that blocked editor. And there are plenty of articles I could have picked, I don't think "Australian continent is a continent in the Earth.Australian continent is a component of Earth's surface.Australian continent contains Australia." or its Spanish translation "(en) Australian continent is a continent in the Earth.Australia is a componente of superficie terrestre." (note that the middle sentence simply disappeared, perhaps a blessing) is an indication that the issues are due to one bad editor and not to a terrible project. Bisexuality, in German, is first an error and then "Bisexuelles lieben Fraus. Bisexuelles lieben Manns." Not touched by the blocked editor. this is not created by the blocked editor. The "Italian" article "Holy Week is a liturgical season. Settimana Santa is part of the anno liturgico." is not edited by them. Obviously it would be best if you mass deleted the embarrassing creations by that editor (full article: "A current is a current.A current is an electricity."), but it wouldn't solve any of the fundamental issues. Fram (talk) 07:37, 3 June 2026 (UTC)Reply
And it's not like this is the only editor who writes articles about dead people. "William Shakespeare is a playwright from Kingdom of England." Has not been touched by that this editor either. Warudo (talk) 11:22, 3 June 2026 (UTC)Reply
@Fram Thank you for the continued updates, which are rather amusing, if in a morbid way... Would you be interested in starting an RfC on Meta on putting an end to this madness? If anyone has any better ideas, I'd be happy to hear them. Toadspike[Talk]11:49, 2 June 2026 (UTC)Reply
I normally don't edit Meta, and I fear that many people who hang out around there are more inclined to let such projects continue no matter what. Fram (talk) 12:17, 2 June 2026 (UTC)Reply
For what it's worth I agree with Toadspike and would like to see an RfC on meta. It might be true that this bias exists there, but Abstract is such a one-of-a-kind, unmitigated disaster that I believe it would not apply in this case. Choucas 🐦⬛12:29, 2 June 2026 (UTC)Reply
Not the right process; you're thinking of m:Proposals for closing projects.For what it's worth I think there's something worthwhile to come out of this; to resolve the fundamental problem the articles look to be moving towards using functions that express specific meanings rather than grammar. I definitely agree that the WMF is hopelessly overstating its progress, as Fram has meticulously documented.I would support some kind of discussion to make WMF be accurate when assessing the level of progress on the project. An RfC might be able to do that but I'm not sure if purely WMF communication matters can be subject to an RfC. In solidarity, Aaron Liu (talk) 12:40, 2 June 2026 (UTC)Reply
I have some strong opinions about the content created by a couple contributors. This one was the original unattributed enwiki cop[y] Fram had originally mentioned. In addition to Csisc, there's also User:Immanuelle, who was blocked here for WP:CIR. Generally I think they share a need to do something now, though at this stage these contributions are unhelpful and frankly actively disruptive. Feeglgeef (talk) 23:45, 8 June 2026 (UTC)Reply
What a poor thing if integrated into small language Wikipedias without any correction, especially for those Wikipedias without active sysops or rollbackers.--Jason2016426 (talk) 14:17, 12 July 2026 (UTC)Reply
People keep commenting about stub articles filled with grammatical errors. This seems to have changed. This is - or would be - a large article, but little text actually shows. The rest is just errors. ~2026-40174-24 (talk) 11:56, 17 July 2026 (UTC)Reply
Coming from huwiki, I've been trying to understand the scope and progress of Abstract Wikipedia, and I want to say first that Frem's updates in this thread have been more informative than the project's own pages on Meta. Thank you for that.
That itself points to my concern. This is an enormous and complex undertaking with many moving parts. A project of this size depends on community buy-in, so it needs to communicate its scope, progress, and decision-making clearly and regularly. Right now I don't think it does. If you read meta:Abstract Wikipedia and its subpages, much of it appears out of date, and the current scope and state are not stated anywhere in a way an ordinary editor can quickly grasp. You could say they're rather abstract. Additionally the original project proposal did not lay out the alternative solutions considered for the individual problem statements, nor any cost-benefit analyses. From what I've been able to piece together there were more readily available approaches, that could have delivered usable results by now (not including the recent advancements of LLMs). Checking the pages and the Wikimedia Board minutes it is also unclear what changes of direction have been discussed or approved since the initial 2020 resolution, or under what criteria the project would be judged unfeasible to continue. This is important, cause this is a highly ambitious project that, without clear guardrails, could go on indefinitely.
In this thread serious concerns were raised, and it's shocking that so far no one from the project has responded to these. I can only add to those, I agree that the project cannot really be described as an "early beta": after reading through pages of Abstract Wikipedia documentation, I still have no idea how I, as an editor, could begin supporting the project and test Hungarian implementations.
Without a clear understanding of what is happening and why, community discussion is effectively blocked. So that this does not sink without a reply, I'd like to ask the project leads to respond:
Current state and scope: What is the precise, current scope and phase of the project? What is the progress on the tasks, which of these are done from the initial 30 months plan? What is still prototype or research?
Supporting a language: Is there a documented, step-by-step path for a language community to add or improve support for its language? Concretely, what could a non-English editor do today to help with testing?
Documentation: Given that the Meta pages appear outdated, is there another current, plain-language page stating scope, status, and roadmap? Could Meta be updated?
Regular review: Since the Board's approval (the resolution of 22 May 2020), what changes of direction have been discussed or approved, and by whom? Are there defined milestones or off-ramps, and what criteria would lead the Foundation to conclude the project is not feasible to continue? How does the Board assure itself that the project remains on track and worth its cost, and what would have to be true for it to decide otherwise?
Resources and partial wins: Approximately what has been spent on the project to date, and how is the continued resource commitment justified against other Foundation and community priorities? What concrete, independently useful results has the project delivered so far (with examples)?
And @NGunasena (WMF): if you could comment on, or connect us with someone who can say more about the Board's thinking and its insight into the project's actual progress (questions 4–5), I'd be grateful.
To be clear, I'm raising this in good faith, I want the project to succeed, and clear communication from the stakeholders is a precondition for the community to participate. Boro (talk) 11:32, 7 June 2026 (UTC)Reply
Not WMF, but for #2, you could look at which functions are commonly used on abstractwiki (e.g. f:Z28016) and learn to contribute to them. I am aware this is a tall ask and if you want to, I could help you. In solidarity, Aaron Liu (talk) 14:17, 7 June 2026 (UTC)Reply
Thank you for the offer, but it's a taller ask than it looks. Hungarian is an agglutinative language that uses vowel harmony. The simple "defining" structure you linked (X is the Y) does exist in Hungarian, but the neutral, encyclopedic version doesn't use a definite article at all, while the variant that does carries extra contextual meaning. So it's unclear which needs to be used here: match the intention or the definition? The correct rendering needs helper functions to get the morphemes right, and research groups in MT and linguistics are still working on tools to do this reliably for all possible cases. I'm not an expert in this field, so I don't feel comfortable writing semi-correct functions to make this work. I've already seen students pick up incorrect uses from AI output, so getting this wrong at scale could affect the language itself.
From the Meta page this seems to fall under "Task P2.9: Renderer for a language from another family". But I can't find any notes on whether that was evaluated, and with what results and to what extent, before moving forward with the project. As I understand it, the majority of its benefit is supposed to come precisely from supporting non-Indo-European languages, so checking more than one language family seems essential. Boro (talk) 20:10, 7 June 2026 (UTC)Reply
My understanding of Hungarian isn't the best, but from what I remember, isn't just a rearranging of the words? So it'd be literally "Budapest Hungary's capital" (Budapest Magyarország fővárosa). Feeglgeef (talk) 21:09, 7 June 2026 (UTC)Reply
Not quite, főváros is the base word, +a/+e is a suffix, that can change form based on the word: capital of → fővárosa (főváros), sibling of → testvére (testvér), older brother of → bátyja (báty), younger brother of → öccse (öcs), mother of → anyja (anya). Also all these sentences are valid, but subtly different in their meaning: (1) Budapest Magyarország fővárosa, (2) Magyarország fővárosa Budapest (3) Budapest Magyarországnak a fővárosa (with definitive article). Boro (talk) 22:03, 7 June 2026 (UTC)Reply
If it doesn't use a definite article, then the output should not have a definite article. I think how this is supposed to work is that each function outputs the construct with the intended meaning no matter what the grammar is. In solidarity, Aaron Liu (talk) 18:15, 8 June 2026 (UTC)Reply
I'm not WMF staff, but as a volunteer I'm one of the most active users on abstractwiki and a functioneer on Wikifunctions.
Currently, on the Abstract Wikipedia wiki, which is in beta, users are able to create basic stub articles that render in English and often other languages. Wikifunctions is no longer in beta and is generally usable, P1.1 through P1.13 and P1.18 though P1.19 are things I'm able to do on the wiki right now. The plan listed for Abstract Wikipedia is very different from what it seems is actually being implemented, so I can't comment on progress there.
Somewhat. I do believe we need much better documentation, but f:Help:Contents has a few pages that are good starting points. I have helped and am willing to help others with the process of creating localized versions of functions.
I don't believe so, unfortunately.
I'm not sure, as again I'm not WMF staff.
On results, we've seen some use by smaller-language Wikipedias and small-and-mid-sized Wiktionaries of using Wikifunctions like templates. As mentioned before, Wikifunctions is also useful on its own and has a large catalogue of functions.
"users are able to create basic stub articles" for extremely low bars of "basic" and "stub" (and "create" for that matter, many editors don't seem capable of this, the learning curve is quite steep).
" that render in English and often other languages. " Very rarely, actually. But feel free to give us a few examples of what you consider basic Abstract stubs and which languages they render in. Take for example Australia[32], which probably comes the closest to an actual article. When I try to translate it to German (and after a very long wait), I get 5 errors, and a mixture of German sentences, English ones, and mixed ones like "The Einwohnerzahl of Australien was 27614411 in 2025." Same when I try it in French, 6 errors (4 different ones), mixed languages. But at least in English you get an article that reads as if some dumb robot compiled it, with the first five (obviously very short) sentences starting "Australia is", and ending with varied but hradly correct ones like "Melbourne is the most population urban area in Australia." Or let's look at the most recent one you created, Labrador Retriever[33]. Full text: "A Labrador retriever is a dog." Capitalization is hard apparently. Dutch translation: "{labrador retriever} ⊆ {hond}". Another one you created, 6-7[34], doesn't even work in English ("Wikifunctions returned a failed response: Reached recursion limit in orchestrator") Mount Everest[35]: "Mount Everest is a mountain of China–Nepal border.". Frenglish: "Everest is a montagne of frontière entre la Chine et le Népal." Swedish: "Mount Everest is a berg of ." Spanish: "Everest is a montaña of Frontera entre China y Nepal."
So no, both creating decent stubs and rendering them in other languages doesn't really work, even when one of the most experienced Abstract editors creates them. Fram (talk) 09:32, 8 June 2026 (UTC)Reply
It's crazy that the resources of the WMF are going into this hopeless and pointless black hole instead of fixing actual problems that the community have been highlighting for years. It would be fun if it wasn't tragic Ita140188 (talk) 09:42, 8 June 2026 (UTC)Reply
Yes. It's time to pause development on Abstract (to avoid the embarrassment of cancellation) and redeploy the engineers to process community tech requests. That's not as good a result as retaining the staff with relevant experience, but it's a start. Certes (talk) 13:01, 8 June 2026 (UTC)Reply
You're a one trick pony, Fram. We get it. Some of the articles on Abstract Wikipedia are not great. But if one billion useless articles were added to the English Wikipedia, I wouldn't call for it to be shut down. The fact that you can find articles that look bad doesn't mean the project needs to be shut down. As for your examples, they do not disprove (in fact, they prove) my intentionally conservative statement, users are able to create basic stub articles that render in English and often other languages (emphasis mine). Feeglgeef (talk) 15:07, 8 June 2026 (UTC)Reply
If you would look at article after article on enwiki that looked like this, you would be right to call for it to be shut down though. And it's not a case of "I can find articles", I picked recent ones created (or in the case of Australia edited) by you specifically, as last time you complained that some examples came from an editor who is blocked on enwiki (but not on Abstract).
But I see that you consider these rather dreadful examples as "basic stubs" and the abysmal translations as evidence that they "often render in other languages" and provide no better examples, so I presume that this "one trick pony" (WP:NPA) somehow found representative examples and you won't point to some better Abstract articles instead. Fram (talk) 15:42, 8 June 2026 (UTC)Reply
Firstly, I'm sorry for the "one trick pony" comment. I had meant it as a comment on the lack of variation in your evidence, and should have framed it as such.
Here's a decent article, abstract:Q1186. It's a basic stub (the information would be useful) and renders correctly in both English and Bangla, thus meeting all of the criteria I gave. Feeglgeef (talk) 16:25, 8 June 2026 (UTC)Reply
Thanks. Why does it give the population in 1961? It's less than half of the current population, so hardly good information to include here. Fram (talk) 16:31, 8 June 2026 (UTC)Reply
That's especially weird considering that not only does Wikidata have a figure from 2017 but it is also marked with preferred rank. Warudo (talk) 16:33, 8 June 2026 (UTC)Reply
I've fixed it such that it will use preferred statements first. Unfortunately, despite the fact that I've requested it multiple times, there's no way to fully clear the cache, meaning the Kerala article is stuck with the old value. But it will work for new instances, like Illinois, which I just created. Feeglgeef (talk) 16:48, 8 June 2026 (UTC)Reply
Seeing the working examples led me to think that, at the reader level, this is something that ArticlePlaceholder was already capable of: surfacing Wikidata facts about a topic with no local article. For topics that do have an article, inline Wikidata item calls can keep those facts current on its own (though the editing experience could be smoother). All in all, I don't see why sentence generation is worth all this effort, when easier paths were open to achieve essentially the same benefit. Boro (talk) 22:37, 8 June 2026 (UTC)Reply
There're certain things only articles can do, like document the history of how something happened. I don't really see any advantages of ArticlePlaceholder over just referencing Wikidata when you do need to switch wikis. In solidarity, Aaron Liu (talk) 01:51, 9 June 2026 (UTC)Reply
You described yourself as one of the most active users on abstractwiki and a functioneer on Wikifunctions, so I do not find it unreasonable for Fram to look at your articles and point at the glaring and massive issues there. Maybe being able to find articles that look bad means little, but are we even able to find articles that look good though? Or is the bar so low that "good" by Abstract Wiki standards just means "not broken in a thousand obvious ways"? You have to admit that from an exterior point of view, things really look dire, and the fact that there is not even more than one outdated meta page in terms of documentation does not bode well in terms of perspective. And that is even without getting into the immense issue that the basic premise of the project is based on disproved linguistics, as has been pointed out elsewhere. Choucas 🐦⬛16:04, 8 June 2026 (UTC)Reply
I do agree that things might look dire from an external point of view, but, as I've been trying to show, Abstract Wikipedia is a viable project. It just needs more time, and, ideally, better technical support. I do agree with those above that Dr. Vrandečić is overstating our progress. Feeglgeef (talk) 16:56, 8 June 2026 (UTC)Reply
Personally I think it's worthwhile to explore this avenue with the aim of making imperfect 'universal stubs' that serve as placeholders while small wikis are developing, but the design of abstract wiki is frankly amateurish, the team appear to be incredibly out of their depth. Why has wikifunctions been primarily based on English/Indo-European languages when there'll be practically no new wikis in languages closely related to English? Do we know if any professional linguists were consulted (the mastermind behind it sure as hell isn't one)? Why was this not done by experts? How on earth has $6 million been spent? On what? How has this gotten through so many checks and meetings? Kowal2701 (talk, contribs) 21:08, 8 June 2026 (UTC)Reply
I'm open to the possibility that there could be a way to write articles using a lingua franca that is amenable to translation to many different languages. But I have strong doubts about Abstract Wikipedia being viable as a crowdsourced project using Wikifunctions as the lingua franca. isaacl (talk) 00:17, 9 June 2026 (UTC)Reply
The lack of documentation means people have to (re)invent the wheel when they want to write things and, at least some of them are not good at it. Consider abstract:Q246463 where the editor who made it wanted to use a pronoun instead of repeating the name of the article subject, so they passed it (Q6091500) to the function that generates the second sentence. There are two problems with this. One, it leads to the awkward phrase "of it" instead of "its". But more importantly, it doesn't work for languages other that English because grammatical gender is different between languages. What is the equivalent of "it" in French, which doesn't have neuter? Or worse, what would happen in a language where neuter exists, but the word for "shrine" is not neuter? Warudo (talk) 17:08, 8 June 2026 (UTC)Reply
This is a problem as well. Wikifunctions (and, by extention, Abstract Wikipedia) has a very steep learning curve, so we have a few users who understand it deeply and everyone else, who barely understand it at all, which makes it very hard to find contributors, whereas the English Wikipedia has a remarkably flat learning curve and a very high ability to specialize. Feeglgeef (talk) 17:19, 8 June 2026 (UTC)Reply
This is the part that makes me ambivalent about the WMF's involvement. I think it's worthwhile doing research in understanding ways to capture information based on commonalities in language. It's not clear to me, though, that crowdsourcing is an effective way to generate an encyclopedia of articles written in a lingua franca with a programming-like syntax. Without the crowdsourcing aspect, I don't think the project is a good fit for the strengths of the Wikimedia Foundation. isaacl (talk) 17:42, 8 June 2026 (UTC)Reply
I'm reading through this again and that "Dutch translation" in particular scares me because of how wrong it is on multiple levels. First of all, is returning math notation really helpful? Wouldn't an error be better here so there's no illusion of support? But more importantly, no matter how smart the person who came up with this workaround felt, it is still wrong! "{labrador retriever} ⊆ {hond}" means that the singleton set {labrador retriever} is a subset of the singleton set {hond} which is obviously false. It should be "labrador retriever ⊆ hond" or, more explicitly, "{x|x ∈ labrador retriever} ⊆ {x|x ∈ hond}". If after all this time, people who work on this stuff cannot handle what are essentially instance of (P31) (∈) and subclass of (P279) (⊆) correctly how can we expect this project to ever succeed? Warudo (talk) 10:35, 13 June 2026 (UTC)Reply
Hello @Boro, and thank you for your good faith inquiry.
I agree with you that the project pages for Abstract Wikipedia could use an update. We will spend some time in the near future to do so, thank you for pointing that out. We have mostly focused our progress report through our weekly newsletter. We have been keeping the newsletter up for more than 250 editions now, and I am sure you would be able to find most of the answers to your questions in the updates. But I admit that sorting through so many newsletters to find the one answer you are looking for is not the best solution.
Here is the current status of the project, in a very short overview:
Wikifunctions launched almost three years ago (July 28, 2023). The project has a growing community, and there are now more than 4000 functions created. The speed of function creation is increasing.
Wikifunctions function calls can be embedded in almost all Wiktionaries, as well as Dagbani, Bengali, Igbo, Hausa and Malayalam Wikipedia. This allows editors to call a centralized store of functions to produce outputs, instead of having to create and maintain them locally in each wiki. An example is here, where the German declension table comes from Wikifunctions. If more Wikipedia language editions would like to have this functionality, please let us know. We are currently prioritizing article integrations and are planning to revisit embedded Wikifunctions at a later date, but let us know your feedback.
From Wikifunctions, we are able to access data from Wikidata. This includes most statements about items, and most lexicographic data. This allows the community to create functions using the rich lexicographic data in Wikidata, or using population numbers from Wikidata directly, etc.
Abstract Wikipedia is in early beta, and we made it available live to enable learning, testing, and iteration. We want to co-create this project with our communities, including on how NLG functions evolve. Here’s an example that was just working as I write this, a short article about the year 1234 in English and in German. We also want to see how the system performs with real traffic to make it more stable and responsive. We are listening to feedback and releasing improvements every week.
Now you might object that the Hungarian article about 1234 is better than our abstract article (for example, it doesn’t call 1234 a Gregorian year even though it’s a Julian one), and you’d be right. Our scope, however, isn't to replace existing articles, but to fill in gaps where they currently exist, particularly in language editions with a smaller community.
In a few months, this will be possible when language communities will be able to integrate Abstract Wikipedia articles into their Wikipedias. We hope by then such errors will be fixed, and more content will be possible. This will be the last major piece of the puzzle, and once in place, we will work closely with communities to discover missing capabilities, bottlenecks, and potential for improving the contributor interfaces. We will create feedback loops to ensure the product evolves to support community needs. We will assess how effective Abstract Wikipedia is at addressing multilingual content parity, and make further decisions on next steps from there.
We are fully aware that this is only addressing a small part of the questions here. Our suggestion is that we will summarize the constructive criticism and concerns to kick off a discussion page on Meta, and answer the questions there. This discussion here has surfaced great suggestions and criticisms which we want to address. We want to take this input seriously, but it will take some time, and we can then continue this conversation on Meta, where it is visible for all communities.
"making it much easier to start a local version of the article, instead of writing it from scratch or translating from another language. " That I doubt. It's not as if for these smaller languages, things will magically work if you have the right Abstract article and the right functions. Even completely ignoring the language structure issues (a pretty big issue in itself), people will need to navigate Abstract and Wikidata, add labels and lexemes, and who knows what else. Even a simple "article" like the rather useless 1234 linked here, gives 6 errors when translating to smaller languages like Nahuatl (and 2 errors for French), with the errors not really helping to knw what needs to be done.
Why you believe this can be integrated in any Wikipedia in a few months time is beyond me, sounds more like a KPI that needs to be met than an actual good plan. Publishing one or two symbolic "articles" from Abstract to some language and then celebrate this as a major milestone and a proof that it works, when in reality next to nothing has been achieved and questions about the fundamentals are completely ignored, does not give any confidence.
As for the Wikifunction, it looks to be wrong or at the very least oversimplified. A rather different table is given e.g. here or here or here.
Having basic information wrong - as in the 1234 example or several other examples listed here - is not filling in a gap on smaller projects it's spreading misinformation. I would not expect information from abstract to be 100% correct. Or even as correct as the average enwiki article or the average article on some other large project. But I would expect it to have basic information correct. And I would expect it to be reasonably intuitive for someone who spots an error to be able to correct it, which if I understand things correctly, would mean some kind of easy connection back to Wikidata. Best, Barkeep49 (talk) 15:59, 9 June 2026 (UTC)Reply
How is that a easier way to populate articles for smaller languages compared to automatic translation from a bigger language and basic copyediting? It seems to me this would require a lot more work for a much worse result compared to the basic way that worked until now Ita140188 (talk) 07:59, 10 June 2026 (UTC)Reply
They seem to have the wrong idea that when you have an abstract article and working functions, all you have to do is change the language and you have an article. Without even going into the grammar issues with many languages, it ignores the work needed in Wikidata to get everything you need for a certain language. Hence, in the 1234 example given, the 2 errors for France and the 6 for Nahuatl. It also ignores the complete instability of Abstract: if I go to 1234 (in English) now, instead of the text I saw yesterday I get "Wikifunctions returned a failed response: Invalid orchestrator result". Finally, it ignores that probably you will find more people interested in actually writing articles in their language, than people navigating 3 or 4 environments (Abstract, Functions, Wikidata, and Wikipedia) to copy someone else's creations, certainly when they can (for many, not all languages) much more easily copy creations to their language through online translation tools.
The only use case where I see this making sense is from a push-concept, where people from Abstract "push" translations (in languages they don't understand) to all Wikipedia languages where a certain topic doesn't exist yet. This obviously is a disaster waiting to happen, and should be opposed by all possible means. Fram (talk) 08:16, 10 June 2026 (UTC)Reply
Even for the example at wikt:hr:Wort this is painfully useless at scale, since the construction and further editing of {{#function:Z29055|L2206|}} requires knowing that there is a Wikifunctions page called f:Z29055 and a Wikidata page called d:Lexeme:L2206. It is abhorrent that instead of doing something along the lines of mw:Multilingual Templates and Modules, millions of dollars from the Endowment (source here) are being spent on this website that’s extremely hard and annoying to use, and is impossible to edit without JavaScript. stjn15:05, 10 June 2026 (UTC)Reply
@Sannita (WMF): can you give us clarity if there are future funds being planned to be sunk/invested into this project? Like many people above, I'm sceptical this can ever work at scale, and feel that small communities deserve better. In solidarity, —Femke (talk) 🐦 16:08, 10 June 2026 (UTC)Reply
To be fair, there's no good reason to call {{#function:Z29055|L2206|}} directly from mainspace. Functions that editors will use often should be wrapped in templates precisely so editors don't have to memorize the ZID. As for the LID, you can open Wikidata on another tab and look up the lexeme to get its ID. It doesn't sound that bad. Warudo (talk) 16:14, 10 June 2026 (UTC)Reply
I think if someone provides that as an example of the potential usefulness of Wikifunctions, I am going to judge it on its merits. And the merits here are entirely unconvincing (what you are describing in itself is extremely tedious), especially compared to investing in the less pie in the sky ideas that you can use the regular editing interface for (Wikifunctions and Abstract Wikipedia in themselves are a good example of how badly the website would be run if WMF developers ever forgot that to truly serve knowledge to everyone, we need to be HTML-first). stjn18:11, 10 June 2026 (UTC)Reply
How can it be an easier way to create articles for smaller languages compared to automatic translation? And the automatic translation is already given in the Special:ConstentTranslation.--Jason2016426 (talk) 14:32, 12 July 2026 (UTC)Reply
@Jason2016426: There are three issues with the automatic translation approach:
For many languages, automatic translation is still not good enough or even simply unavailable. Point in case is the English Wikipedia's reluctance to accept automatically translated articles -- so, even for English, the quality is not deemed good enough.
Automatic translation basically forks the content of an article. Updates are not propagated but manually.
Automatic translations are by design a one-way street. No knowledge added to a translated article propagates back to the source article.
There are also advantages to automatic translation. The good thing is that the Abstract Wikipedia project does not require exclusivity. Readers and contributors both can still use the benefits of automatic translation wherever it fits their needs. --DVrandecic (WMF) (talk) 09:02, 16 July 2026 (UTC)Reply
Point 1 is kind of a flawed argument. Articles created with machine translation are not great, but they are infinitely better than articles from Abstract Wikipedia. I don't see how this is going to change ever, given the continuous improvement of machine translation and the incredibly slow progress of Abstract Wiki. Ita140188 (talk) 10:44, 16 July 2026 (UTC)Reply
The Abstract Wikipedia team has drafted a response, which was reviewed and improved by the community. We want to express our gratitude to the community members who have shown their care and spent their valuable time contributing to the answer. All remaining shortcomings are strictly ours. The answer can be found here: meta:Abstract Wikipedia/Response to English Wikipedia criticism. You are welcome to comment and to pose further questions and criticism on the talk page. -- DVrandecic (WMF) (talk) 08:55, 16 July 2026 (UTC)Reply
Clearly not. I can't believe that there is nobody that can say anything useful here, so can only conclude that the non-reply is because they have been ordered not to communicate with us plebs. What do we know anyway? Phil Bridger (talk) 16:42, 18 June 2026 (UTC)Reply
@Phil Bridger, as of this message in this thread you have made nine comments. With them, you have managed to: - Imply every WMF employee answering about Abstract Wikipedia would be engaging in bad faith because their salary depends upon not fixing issues; - Suggest that the WMF workforce is so lazy that the place would run better after firing half of them; - Belittle (twice) an editor who made the error of trying to address your concerns in good faith; - Dare any of the hundreds of WMF employees to come spar with you; - Suggest their limited answers are actually due to a marching order because they despise the community. I want Abstract Wikipedia, a flawed project based on an erroneous assumption and no community to carry it, to be abandoned shortly as much as anyone here, including you presumably. I also think communicating with the community should be a bigger part of the responsibilities of WMF employees. Nevertheless, I do not believe dealing with your constant, pointless, and toxic abuse is part of that. Either engage in good faith to find solutions, or control yourself. If what you need is a place to vent, find another one, because aside from lashing out and poisoning this already heated conversation further, you have accomplishing nothing. Choucas 🐦⬛19:24, 18 June 2026 (UTC)Reply
Thank you. I'm operating in good faith here. Everyone here wants what's best, and, even without AGF, we wouldn't have spent countless hours contributing to Wikimedia projects if we didn't. Feeglgeef (talk) 21:24, 18 June 2026 (UTC)Reply
Thank you. However, I get the feeling that Feeglgeef is a little out of their depth here, but there's nobody with any decision-making authority at the WMF willing to rescue them wasn't particularly kind, nor did it have anything to do with the merit of my argument. I understand that this community has a poor relationship with the WMF (I don't particularly like them either!), but that doesn't mean that Abstract Wikipedia has zero potential in any form, and just because I believe in that potential doesn't make me a mindless WMF droid. Feeglgeef (talk) 21:59, 18 June 2026 (UTC)Reply
@Feeglgeef, @Phil Bridger Can both of y'all take it down a notch. Calling folks mindless WMF droids or insinuating that they are lazy or bad faith assumption because they have been ordered not to communicate with us plebs is definitely not okay. Sohom (talk) 07:21, 19 June 2026 (UTC)Reply
True, but the people in charge of the project posting halftruths at best and then ignoring all follow-up questions isn't really okay either, and causes the irritation that shines through in some ill-tempered remarks here. Fram (talk) 07:36, 19 June 2026 (UTC)Reply
I have my own misgivings that roughly align with yours (I've tried contributing to Abstract Wikipedia and found it extremely hard to do so -- much harder than the cost of machine translation or even writing something from scratch in my native language). However, I don't think Sannita is talking in half-truths, I think WMF (as an org) does believe that they can create a system that can: "make it much easier to start a local version of the article, instead of writing it from scratch or translating from another language." and the approach they are taking is to first build out the whole pipeline first and only then measure whether operating Abstract Wikipedia causes production outages, whether or not editors find it easier to write articles in Abstract Wiki-lang (for the lack of a better term) and whether the effort of learning to write Abstract Wiki-lang is worth the trade-off of being able to write a few dozen linguistic rules (or whether that is even possible at scale). cc @Femke's call out of whether future funds will be invested, my understanding is that the team is still on grant runway and my understanding is that they do plan on measuring whether or not Abstract Wikipedia is viable this year (actually maybe @DVrandecic (WMF) or @ATsay-WMF would be better placed to answer that specific question?) Sohom (talk) 08:20, 19 June 2026 (UTC)Reply
Comments like "We did some limited tests in a controlled environment, but doing so can only help you so much in identifying potential problems. We are learning a lot by releasing the beta project (because this is still a beta), and we'll improve from there. " are half-truths at best, with words like "limited" or "beta" doing an awful lot of heavy work here.
"Abstract Wikipedia is in early beta, and we made it available live to enable learning, testing, and iteration. " is also a dubious statement. Like has been said, it was an untested alpha, where the WMF really dropped the ball quite dramatically, with the most logical conclusion that some KPI or deadline had to be met. Even the examples given for the use of Functions and Abstract were and are very dubious. At the moment, 1234[36] gives me "Wikifunctions returned a failed response: Error in the function call API". In other major languages, for the same article, I get errors like "Wikifunctions returned a failed response: Invalid executor response", "Wikifunctions returned a failed response: No matching lexeme for item in language", and "Wikifunctions returned a failed response: Argument value error". Fram (talk) 09:00, 19 June 2026 (UTC)Reply
When AL2 was saying that we should not build publicly open tools that show the impact of WMF’s grant funding because of potential scrutiny and criticism of WMF by Wikimedia movement community members, she singled out and named a specific critical Wikimedia movement community member as an example. To me, the way AL2 portrayed the critical community member came across as very negative. It created the impression that the critical community member was a bad, hostile person with negative intentions towards WMF. My manager did not object to or question AL2 doing this to a community member. She appeared to be in agreement with how AL2 treated that person.Gnomingstuff (talk) 20:01, 27 July 2026 (UTC)Reply
Are you seriously responding to a comment from more than a month ago to just throw gasoline on the fire with something that is completely off-topic? I am aware people do not seem to think WP:FORUM applies the same way to WP:VPW, but this is simply ridiculous. What are you trying to accomplish here? Choucas 🐦⬛20:37, 27 July 2026 (UTC)Reply
Sorry, I thought this was part of the below thread, given that there is a lot going on right now on this large page and a lot of scolding of people to assume good faith toward the WMF (to which this is very much on topic) Gnomingstuff (talk) 21:23, 27 July 2026 (UTC)Reply
Two wrongs don't make a right? For what it's worth, while I cannot control (and have no desire to control) who says what in non-public spaces (both editors and Foundation staff), within public/onwiki/phabricator spaces you can be rest assured that if a foundation employee insinuated similar, I would escalate it and strongly demand a apology (and to my understanding Foundation staff are strongly cautioned against publicly castigating volunteers). Sohom (talk) 22:22, 27 July 2026 (UTC)Reply
No, they just ignore them completely, twist the facts, make incorrect or extremely dubious claims, and repeat this for years and years. But publicly castigating, no, not since WP:FRAM I think. We see the exact same pattern at the Image Carousel discussion, where they reply to the questions they think they can spin positively, and ignore the ones where this isn't possible. Fram (talk) 09:05, 30 July 2026 (UTC)Reply
But nobody called anyone a "mindless WMF droid". The only use of anything like that term was by Feeglgeef saying that they were not a "mindless WMF droid". But I see that my words here will be twisted beyond recognition. Phil Bridger (talk) 07:40, 19 June 2026 (UTC)Reply
I know it's popular in wiki-land to hate on anything related to AI, and in particular LLMs, but I look at what Abstract does and I look at what the LLM-driven translation tools do, and I just don't see the point of what we're doing. Our goal is to be able to take an article written in one language and allow somebody who does not read that language to read the article in a language they do know. The LLM tools do that today. They're not perfect, but they're good enough for most purposes, and they'll only get better over time.
I don't mind investing WMF money in research projects. There is value in adding to our understanding of how information is stored and processed, and exploring better ways we can deliver that information to our readers. Abstract and Functions is about that, so I'm happy that we've explored what's possible. But given the progress LLMs have made over the past few years, and the current state of what Abstract appears to be able to deliver, I'm just not seeing how this has a chance of becoming a viable product. That's not to say the experiment was a failure. A failure would be "we haven't learned anything". But we have learned something, which is that "this approach to automatic translations doesn't appear to be the way to go". RoySmith(talk)13:00, 19 June 2026 (UTC)Reply
AI translation has a tendency to be confidently incorrect, but AI is certainly more accurate than using buggy functions and blaming it on the community. ~2026-34629-56 (talk) 15:52, 19 June 2026 (UTC)Reply
It's a record of data and not AI. If WMF steers this in the right direction and invests the resources and work in the right places (which they are surprisingly failing to do for a project that they so champion) then the articles have the capacity to be much better than AI. In solidarity, Aaron Liu (talk) 17:31, 19 June 2026 (UTC)Reply
How? Why would WMF be able to translate into obscure languages without the help of one or two linguists of those languages? Someone will have to translate all lexemes for these languages, add all labels for these languages, check the output of Abstract, and then give the green light. We need to be certain that they are native speakers or very fluent in the language, to avoid a Scots scenario. And all that to translate an Abstract article which will probably be worse (language, completeness, ...) than enwiki, dewiki, jawiki... and not adapted to the target language (see e.g. the example of 1234, which doesn't take into account any other year system). LLMs won't do all of these things either, but for a fair number of languages it will already produce a result which is as good as can be expected from Abstract, but without all the additional effort described. How will Abstract ever be better at those things? Fram (talk) 18:15, 19 June 2026 (UTC)Reply
Exactly, that's what they would need to do (the advantage is deterministic outputs). It's worrying that the infrastructure for what they need to do is in such an alpha-at-most state. In solidarity, Aaron Liu (talk) 18:59, 19 June 2026 (UTC)Reply
"LLMs won't do all of these things either" is such an understatement by the way. They have no understanding of the things they translate at all, which is pretty bad for an encyclopedia because we use technical terms whose translation is not always straightforward. For example, in some languages a field is called a "body" (presumably the ones that borrowed the term from French). My native language, Greek, is one of them. Guess what Google translate does when I ask it to translate Field (mathematics) ([37]). It switches between the correct term (σώμα) and the literal translation of "field" (πεδίο) between paragraphs. To no one's surprise, it sticks to the wrong word most of the time and it also uses the wrong word in the title. "Field" is far from unique in this. While checking this and other articles, I saw words like function, matrix, ideal, differentiation, infinitesimal, domain etc. get mistranslated too. That is in addition to all the other things it butchers along the way ("general quintic equations" -> "general quantum equations"?? in the field article) as well as grammatical mistakes even a complete novice wouldn't make ("Εάν η συνάρτηση είναι διαφορίσιμο(??) στο a" in derivative[38]). It's not a google exclusive problem either. In my testing with DeepL it didn't do much better. Also, since I mentioned French earlier, here is a French example where Lie ring turns into "ring of lies" (I don't speak French so I verified this with Wiktonary): [39]. Warudo (talk) 20:04, 19 June 2026 (UTC)Reply
To be clear the point of my comment is that the only translation that is worth considering is one done by humans. I understand why people want to compare abstract Wikipedia to LLMs but I think that comparison misses the point. Even if abstract Wikipedia reaches a point where it is able to produce LLM level output, which I feel would be nothing short of a miracle, it would not be anywhere near good enough to consider using. Warudo (talk) 20:10, 19 June 2026 (UTC)Reply
Incorrect word choice is hardly limited to computers. People do it too. I used to work with a guy from Cuba who, while fluent in English, never quite mastered the difference between "in" and "on" (they're both "en" in Spanish). The phone would ring, he would answer it, and hand it to me: "Bob is in the phone for you". Oddly enough, "in the phone" actually makes more sense, but that's not the correct English idiom.
I read a lot of text related to checkuser actions in various languages. It didn't take long for me to figure out that "doll" was just Google Translate's way of saying "sockpuppet" in Spanish, and so on for other languages and other bits of wiki-jargon. In the end, it really isn't a big deal. It was obvious from the get-go that "doll" wasn't the right word but I quickly figured out what it meant and ultimately there was no real hit to my cross-wiki reading comprehension.
I expect the same is true of "field" for you. You recognized that πεδίο didn't make sense, but you figured it out and now that you know what it means, so what if it's not the correct word? Languages evolve. Hardly a day goes by on the internet without me being reminded that I'm no longer one of the cool kids because they're using words I've never heard of before and I have to resort to Urban Dictionary to figure out what they're talking about. Is it really fair that we hold the LLMs to a higher standard that we hold ourselves? RoySmith(talk)20:52, 19 June 2026 (UTC)Reply
Let's assume I didn't speak English and elwiki didn't have an article on fields. I'd read the translated enwiki article, see that the term is called "πεδίο" (this is the term that appears most, the other terms are probably translation errors) and move on. Except what if I then wanted to go look for more information outside Wikipedia? I'd then pick up my mathematics textbook, check the index and only find results about scalar fields and vector fields (which are called πεδία in Greek) but nothing about the algebraic structure. What I'm missing is that I'm looking in the wrong place. If only I knew what it was actually called. Then I want to learn what an ideal is. Google translate said it's called "ιδανικό" [40]. Huh, there's nothing here? If only I'd known the word is "ιδεώδες". And on and on it goes. So, ultimately, you can't look things up on your own because you don't know what they are called so you're stuck asking the AI about them. You also can't really discuss these concepts with others because your idiolect is completely different from theirs. And before you call this hyperbole, I'll let you know that google translate also mistranslates "associative" in the field article. So even if you try to tell your interlocutor the definition of the thing you're talking about, you're still out of luck unless they speak English and guess what happened.
You also underestimate just how confusing things can get. In the translation of ideal (ring theory), the term "right ideal" (which is supposed to be δεξιό ιδεώδες) turns into something that basically means "correct ideal" (σωστό ιδανικό) and "proper ideal" (which is supposed to be γνήσιο ιδεώδες) into something that means "suitable ideal" (κατάλληλο ιδανικό). So someone who doesn't speak English and does not know why these error happened will try to figure out why a right ideal is correct and a left one isn't and what a proper ideal is suitable for. Warudo (talk) 22:38, 19 June 2026 (UTC)Reply
I feel your pain. My particular peeve is symbols. If I read in an article: and don't know what that symbol is, how do I even begin to look it up? Claude actually does a good job if I copy-paste the markup, I get back "The notation: F \ {0} means "the set F, with the element 0 removed." which at least gets me pointed in the right direction. But to get back to my original point, "they're good enough for most purposes". Failing to correctly translate some terms in advanced math articles doesn't negate that. RoySmith(talk)23:01, 19 June 2026 (UTC)Reply
My favorite analogue phrase for "sockpuppet" is the one used in French, which is "faux-nez", e.g. "false nose". jp×g🗯️23:30, 18 July 2026 (UTC)Reply
If by "better" you mean produce better translations in theory, in essence this is pitting a rules-based translation system where the rules are explicitly programmed by humans versus a black box where a program adjusted its tuning parameters by itself that control how translation is done from input to output. One example is the world's strongest chess program, Stockfish, which originally had a human-crafted evaluation function, then introduced a trained neural network, and now has dropped the human-crafted evaluation function. However games have a very specific ultimate evaluation method – did the program win – which I suspect makes it easier for the network to evolve to a state where it wins more often. I think human language translation has more training challenges, and it's not as clear to me that a program self-training could in a theoretical perfect world do a well as an explicit set of rules.
From a practical standpoint, though, codifying rules to capture all nuances is very difficult. And even to the extent it's possible, it's not clear that there are enough people interested in working on it in a crowdsourced project. So while the black box may produce OK translations only some percent of the time, that might still offer a better cost-benefit ratio than getting Wikifunctions to meet the same level of quality. isaacl (talk) 02:51, 20 June 2026 (UTC)Reply
If I understand it correctly, Abstract Wikipedia is fundamentally different to LLM-generated translations. (1) Abstract runs on fully transparent source texts and algorithms instead of LLM sources and algorithms that are to a great degree opaque (despite the label "open") and whose results are not reproducible. (2) The carbon footprint accelerating the climate emergency is, in principle, auditable in the case of Abstract (within the overall Wikimedia carbon footprint) but is mostly hidden by corporate secrecy in the case of the big LLMs for which 23% of Ireland's 2025 electricity is used by data centres. There are similar problems for excess water usage. Boud (talk) 16:20, 16 July 2026 (UTC)Reply
Given what I’ve wrote here above, I feel like the community is owed a more detailed explanation on the funding of Wikifunctions and Abstract Wikipedia as of right now and overall. How much funds come from Wikimedia Endowment and/or Wikimedia Foundation? How much donor money was spent on making a JS-laden interface that’s incomprehensible to an average programmer, never mind a newcomer? Was that really deemed the best use of the funds by Wikimedia Endowment board, and if so, how (i.e., why was this project deemed more important compared to less frivolous ideas that would actually improve our software)? stjn20:30, 29 June 2026 (UTC)Reply
Since the draft response is linked at the very beginning of the RfC, I think it is fair to say it has been taken into consideration. Choucas 🐦⬛11:58, 14 July 2026 (UTC)Reply
@Sannita (WMF) and @DVrandecic (WMF): In addition to what Choucas mentioned, sweeping, handwaving statements like: Generated article quality: This is a concern that we share, but we have evidence that the results will get better with time, and with the expansion of the community. (with zero links to any kind of evidence) is well below the level of transparency and assurance I would expect in response to such detailed feedback and criticism about the problems of scaling this model. Please do better. Sohom (talk) 12:18, 14 July 2026 (UTC)Reply
Latest comment: 10 days ago51 comments23 people in discussion
The WMF is currently running banners with the text:
17 June: Wikipedia still can't be sold We're sorry we've made several attempts to reach you, but it's Wednesday, 17 June, and we still need some help. If Wikipedia is useful to you, please take a minute to protect its future with a $2.75 donation today. Most readers don't donate; only 2% do. Wikipedia is run by a nonprofit, not by a billionaire. So if Wikipedia has given you $2.75 worth of knowledge, please give. Any contribution helps, whether it's $2.75 or $25.
In the past, we produced a consensus that banners that state or imply any of the following are not considered appropriate on the English Wikipedia:
Wikipedia's existence or independence is under threat or dependent on donations
Donated funds are used primarily to support Wikipedia and/or its volunteer editors
Readers should feel obliged to donate regardless of their means ("guilt tripping")
I don't believe that these current banners comply with this consensus. I'm hopeful the WMF can resolve this quickly; I will post to the talk pages of a few relevant employees, asking them to comment. BilledMammal (talk) 07:25, 17 June 2026 (UTC)Reply
Could you expand on why you believe this is against community consensus? Compared to the ads of old, I think this passes muster but maybe I'm missing something. In solidarity, —Femke (talk) 🐦 07:56, 17 June 2026 (UTC)Reply
The most problematic text seems to be "protect its future", where "it" clearly means "Wikipedia". That could be read as suggesting firstly that a substantial part of any donation would go to Wikipedia, and secondly that Wikipedia is $2.75 short of being able to survive without it. However, the implication is indirect and less deceptive than previous banners. Certes (talk) 10:16, 17 June 2026 (UTC)Reply
@Certes: A substantial part does go towards Wikipedia, cc the Annual Plan breakup. Regarding the original question, I personally align with Femke and RoySmith in that I don't think the banner breaks consensus given the context of the guidelines (which were primarily made for a different category of much more egregious banners) Sohom (talk) 16:42, 17 June 2026 (UTC)Reply
Complaining about a banner is a different bar from "banner violates past consensus, needs to be resolved quickly" (which is the tone of this discussion). One is "lets have a dialog" another implies "do this now, or else". Sohom (talk) 19:41, 17 June 2026 (UTC)Reply
@Black Kite Could you explain what part of this is deceptive? (When you answer, one thing to keep explicitly keep in mind that the funding situation has significantly changed since 2022 when this consensus was obtained, WMF does spend 77% of it's budget on volunteers, a majority of which goes to English Wikipedia, and unlike in 2022, donations are actually on a downward trend and WMF is actually now increasingly turning to long-term donors to continue funding itself -- while it still has a fair bit of money around, it is not infinite and the lack of any donations will indeed lead to insecure future at some point, especially in a climate where we are competing with billionare led ventures like Grokipedia). Sohom (talk) 20:15, 17 June 2026 (UTC)Reply
"...take a minute to protect its future with a $2.75 donation today". Nothing is going to happen if you don't donate $2.75, and neither is it if no-one else does either. That was the reason the first clause was agreed the last time we had this discussion. Black Kite (talk)20:26, 17 June 2026 (UTC)Reply
Collectively if no-one else does either is a problem. I remember doing some back of the napkin math, at our current spend rate, I think we have roughly 1.5 to 2 years assuming everyone immediately stops donating. This is obviously not a dooms-day, we pack up tomorrow scenario, but it's not "if nobody donates we will be around for 10 more years" scenario either and the "nobody really has to worry about donating" scenario that you are laying out. Sohom (talk) 20:49, 17 June 2026 (UTC)Reply
Nothing is going to happen if you don't donate $2.75, and neither is it if no-one else does either. How exactly do you think WMF pays its employees and its bills for running its data centres? – SD0001 (talk) 16:47, 18 June 2026 (UTC)Reply
I'm sorry, SD, you're just plain misinformed about where the money comes from. The way it works is the good wiki fairy comes along and waves her magic wand. That keeps the disks spinning and the bits flowing and enlightens repressive regimes around the world about freedom of information so we don't have to waste any effort defending our editors in court. It also makes hotel rooms and airline tickets drop from the sky so volunteers can get scholarships to wikimania. And as the wiki fairy walks along the beach, gold coins grow out of her footprints in the sand; these go to funding local affiliates to keep their education and outreach programs afloat.
All we need to do is join hands and think happy thoughts in a big group hug so the wiki fairy will come and save us. Until then, I'm willing to put up with a few banners. RoySmith(talk)17:26, 18 June 2026 (UTC)Reply
Do we have any detail on this 77% figure? It's one I've not seen before, and I am really struggling to see where $150 million is being spent on volunteers this year. It feels much, much less. Certes (talk) 20:41, 17 June 2026 (UTC)Reply
Thank you. That's interesting. I'm particularly heartened to see that the Annual Plan actually mentions Wikipedia – an improvement on the last one I read. It's always difficult to categorise spending. "Investment in Technology" (48%) includes issues such as management of scraper bots, which is a very welcome function but not one directly aimed at volunteers. Presumably it also covers developments such as Abstract Wikipedia which fewer volunteers might prioritise. We certainly appreciate the work of the community tech team which this item has funded in previous years. Certes (talk) 21:05, 17 June 2026 (UTC)Reply
Presumably it also covers developments such as Abstract Wikipedia which fewer volunteers might prioritise., it does, I had written up a guide about the current annual plan to contextualize what is planned for next year. It's a bit out of date, but it should be relatively accurate and Abstract Wikipedia is a fairly small portion of the pie. A lot of WMF's current spending is on making it easier for intermediate editors learn to become competent editors, make readers stay around for longer and a chunk to help admins fight vandalism effectively (outside fighting scrapers itself and making sure we have high availability) Alongside that you have work being done to make Toolforge more user friendly and making existing APIs better so that folks can reuse content without taking down Wikipedia itself. Sohom (talk) 04:18, 18 June 2026 (UTC)Reply
Most readers don't donate; only 2% do was previously rejected as "guilt tripping"
I also consider Wikipedia still can't be sold to be suggesting that Wikipedia's independence is under threat, although it isn't as blatant as previous examples.
I think people would also be frustrated by the fact they're seeing something several times in the first place. I can't guarantee how other people would perceive it, but more than three times feels like a lot, especially if the answer has been "sorry, no, don't have the money to spare" the first time. If I recall correctly (please tell me if I'm incorrect!) the banners also have a "maybe" option if you want to be reminded, so I don't think it's simply a matter of people that might actually want to be reminded. When I worked at McDonald's, we had lots of people who would donate to the Ronald McDonald house the first day or two of fundraising and then got quickly annoyed later on as we kept going throughout the week ("I already donated"; "no, I just want my coffee"). Clovermoss🍀(talk)01:08, 24 June 2026 (UTC)Reply
No one is going to donate when prompted the first time, regardless of how much they have in the bank. The fundraising team gets it. Once someone has donated, I believe cookies are used to track it and they stop seeing the banners. – SD0001 (talk) 07:53, 24 June 2026 (UTC)Reply
As I recall, it's been previously said that no association is made between donors and on-wiki user names or any session information for a given browser (linked via cookies or I suppose local browser storage). From a privacy perspective, personally I think this is sensible. isaacl (talk) 17:26, 24 June 2026 (UTC)Reply
I support the banner ad program and the current language. It is a reasonable, unobtrusive, and voluntary option for meeting the organization's financial needs... ensuring that Wikipedia can persist long into the future as an independent source of information about the world. They tried letting the community run the ship on the fundraising a few years back and it imploded. Clearly this is not something that the "consensus" is equipped to handle. Ice Vest (talk) 22:57, 17 June 2026 (UTC)Reply
Is no one else irked by We're sorry we've made several attempts to reach you? That's the kind of nagging tone that turns me off any appeal from a charity or non-profit if I get it in a phone call/voicemail or in the post. It sure feels like "guilt-tripping" to me. —In solidarity with Wiki Workers United · ClaudineChionh (she/her · talk · email) 23:00, 17 June 2026 (UTC)Reply
But are they truly sorry? Wouldn't someone who was genuinely apologetic about having contacted me refrain from doing so again? Certes (talk) 11:11, 18 June 2026 (UTC)Reply
I want to give some context about the phrase "protect Wikipedia's future;" the intent is not to suggest that Wikipedia's existence is in immediate danger or that any individual donation determines whether Wikipedia survives. In an era of falloffs in referral traffic and readership decline, protecting Wikipedia's future means ensuring that Wikipedia remains accessible, visible, and sustainable in a changing world.
I also recognize that people may interpret fundraising language differently – which this conversation shows. We value your input and will think of other ways to convey this point, and we'll share them back with you. Feedback on different interpretations is something we look at closely and frequently incorporate into banner language. Every year we workshop fundraising language with communities. You can see Fundraising/2025 banners for community-WMF banner discussions about last year's campaign. Next week, we will be publishing our regular collaboration on fundraising banner content at the Fundraising Hub. We welcome feedback on particular messages there, where we can get more in depth around wording and the considerations behind it.
@TurboSuperA+ Please clarify what you mean here. If you mean our executives, I don't think it is a unknown fact that some of our C-level executives might be millionaires through non-Wikipedia ventures/their former jobs, what matters is that they don't "run" the website. The community "runs" the website and makes content decisions. This is in contrast with Grokipedia where it's trillionare founder makes content decisions and "runs" the website. If you are talking about WMF the corporate entity, XAI, the corporate overlords of Grokipedia have assets that dwarf Wikimedia Foundation's corporate income by multiple billions. In fact, I'd say that the WMF is much much smaller than almost every org that serves "big-tech" numbers of users daily such that the WMF corporate budget in it's entirety would be considered a drop in the bucket for most big tech companies operating at our scale. I don't think it is out of touch to point either of these out and pitch this as a strength. Sohom (talk) 19:45, 18 July 2026 (UTC)Reply
To be fair, the sentence says "run by a nonprofit", so it's clear what definition of "run" they're using. I agree with the rest, and in fact I'm skeptical whether any executive is a millionaire since the amount saved after you deduct everything from the $400k salaries is probably $100k and I'm on the "optimistic" side about their accumulation. (Jimbo probably does but even if we count him, he's the exception, and IIRC doesn't give that much input on WMF's operations. I could be wrong on this.) A lot would probably have a million by retirement but I don't think that counts because you need a million to retire. In solidarity, Aaron Liu (talk) 23:28, 18 July 2026 (UTC)Reply
The CEO of the WMF, Bernadette Meehan, is a millionaire.
I don't think it is a unknown fact that some of our C-level executives might be millionaires through non-Wikipedia ventures/their former jobs
Well, there you go.
The community "runs" the website
Did I imagine all the threads where editors complain about WMF decisions? Aren't we in one such thread now? WMF can make decisions that affect the website without consulting editors.
and makes content decisions
The content is not the website, the technical infrastructue, or the organisation. Is the community ever consulted on what kind of servers Wikipedia should run, where it should be hosted, who to hire to do it...? TurboSuperA+[talk]04:52, 21 July 2026 (UTC)Reply
I would imagine most people in the US who have held down a well-paying job for 10+ years are millionaires.And yes, lots of work is done by volunteers on the technical infrastructure. Phabricator and Gerrit are open forums where you can propose feature requests and implement them. – SD0001 (talk) 06:14, 21 July 2026 (UTC)Reply
These banners that are asking for donations and making false claims about the website going bankrupt is "misleading and unacceptable" and is being used as a scare tactic to force editors to donate. These banners represent nothing but corporate greed and lies.
The platform's fundraising strategy faces considerable public and internal critique:
Substantial Reserves: Financial statements reveal the Wikimedia Foundation holds hundreds of millions of dollars in cash and investments, meaning it is not in danger of shutting down.
Aggressive Language: Critics, and even some veteran Wikipedia editors, argue that the banners misleadingly imply the site is on the verge of going offline if readers do not donate immediately.
Organizational Growth: A large portion of the budget goes toward expanding the Foundation’s bureaucratic footprint and funding adjacent non-profit grants rather than just maintaining the core website.
Hello @~2026-35699-89: could you share what you see in those banners? They should not imply bankruptcy is near.
Just so you know, the Foundation has stopped its grant programme for knowledge equity. Having reserves is a good thing given how tumultuous the world is now I'd say. In solidarity, —Femke (talk) 🐦 18:26, 18 June 2026 (UTC)Reply
This is exactly what i saw: 18 June: Wikipedia still can't be sold
We're sorry we've made several attempts to reach you, but it's Thursday, 18 June, and we still need some help. If Wikipedia is useful to you, please take a minute to protect its future with a $2.75 donation today. Most readers don't donate; only 2% do. Wikipedia is run by a nonprofit, not by a billionaire. So if Wikipedia has given you $2.75 worth of knowledge, please give. Any contribution helps, whether it's $2.75 or $25. ~2026-35699-89 (talk) 05:43, 19 June 2026 (UTC)Reply
Anyone who knows me will be well aware I won't hesitate to criticize the WMF when warranted, but I don't see anything there that states or implies imminent bankruptcy. SeraphimbladeTalk to me14:02, 25 June 2026 (UTC)Reply
2nd Warning to the Wikimedia Foundation
These banners that are asking for donations and making false claims about the website going bankrupt is "misleading and unacceptable" and is being used as a scare tactic to force editors to donate. These banners represent nothing but corporate greed and lies.
The platform's fundraising strategy faces considerable public and internal critique:
Substantial Reserves: Financial statements reveal the Wikimedia Foundation holds hundreds of millions of dollars in cash and investments, meaning it is not in danger of shutting down.
Aggressive Language: Critics, and even some veteran Wikipedia editors, argue that the banners misleadingly imply the site is on the verge of going offline if readers do not donate immediately.
Organizational Growth: A large portion of the budget goes toward expanding the Foundation’s bureaucratic footprint and funding adjacent non-profit grants rather than just maintaining the core website.
another example: 1 July: Wikipedia still can't be sold
You deserve an explanation, so please don't skip this 1-minute read. It's Wednesday, July 1st and you only had a small chance of seeing my message. This year, Wikipedia has had fewer visitors and fewer new supporters, so your visit today means a lot. I hope that Wikipedia's given you at least $2.75 of knowledge recently. If everyone who found Wikipedia useful gave $2.75, we'd hit our goal in a few hours.
After nearly 25 years, Wikipedia is still the internet we were promised-created by people, not by machines. It's not perfect, but it's not here to push a point of view. It's run by a nonprofit, not a giant technology company or a billionare. (Lies)
Less than 2% of our readers donate. The rare few who give do so because Wikipedia provides them with useful knowledge. If that sounds like you, please donate $2.75. Any contribution you make today helps.
This is unacceptable. These banners are highly misleading, financially manipulative, and damaging to Wikipedia's reputation among the community. Thank you. ~2026-37878-87 (talk) 22:46, 1 July 2026 (UTC)Reply
The truth: Substantial Reserves: Financial statements reveal the Wikimedia Foundation holds hundreds of millions of dollars in cash and investments, meaning it is not in danger of shutting down. Aggressive Language: Critics, and even some veteran Wikipedia editors, argue that the banners misleadingly imply the site is on the verge of going offline if readers do not donate immediately.
The Lie: 1 July: Wikipedia still can't be sold You deserve an explanation, so please don't skip this 1-minute read. It's Wednesday, July 1st and you only had a small chance of seeing my message. This year, Wikipedia has had fewer visitors and fewer new supporters, so your visit today means a lot. I hope that Wikipedia's given you at least $2.75 of knowledge recently. If everyone who found Wikipedia useful gave $2.75, we'd hit our goal in a few hours.
After nearly 25 years, Wikipedia is still the internet we were promised-created by people, not by machines. It's not perfect, but it's not here to push a point of view. It's run by a nonprofit, not a giant technology company or a billionare.
Less than 2% of our readers donate. The rare few who give do so because Wikipedia provides them with useful knowledge. If that sounds like you, please donate $2.75. Any contribution you make today helps. <(Corporate greed and lies) ~2026-37878-87 (talk) 22:59, 1 July 2026 (UTC)Reply
I'm still not seeing where the lie is – can you be more specific instead of pasting in a LLM summation? If we break down every claim in what you call "The Lie":
Wikipedia still can't be sold – Y true
It's Wednesday, July 1st – Y true
This year, Wikipedia has had fewer visitors and fewer new supporters – I know readership is down, and wouldn't be surprised if the correlates to supporters, so gonna say Y true
After nearly 25 years, Wikipedia is still the internet we were promised-created by people, not by machines. – Y true
It's not perfect, – Y true – but it's not here to push a point of view. – Y true, though some people may subjectively disagree with that last point
It's run by a nonprofit, not a giant technology company or a billionare. – Y all true, WMF is a non-profit.
Less than 2% of our readers donate. – Y true
Any contribution you make today helps – Y I'm sure it does.
Again, where is the lie? Read the banner. And then read it again. Don't use an LLM to draft your response. You keep seeing something in the banner that is quite simple not there. In solidarity, nilnz04:04, 5 July 2026 (UTC)Reply
Yesterday, I was planning to boldly close this thread as the OP had been indeffed, but then I found that there was a new recent comment, so I left this sub-section to be. In the case these TAs are just trolling us, shouldn't this thread be closed/hatted? - BlueEleephant (talk · contribs) 18:43, 5 July 2026 (UTC)Reply
Just a passing comment. I do not object to the banners, for I think they will generate revenue. But they did make me laugh and shake my head because of some item I saw on the news, decades ago, before I threw away my TV. They were interviewing the wife of televangelist Jim Bakker, who at the time was a guest of the US government and could not be interviewed. They asked her why they kept saying that unless people donated to them their program would fail soon. She said: because that would generate more donations than anything else. So I laughed today, and will have no further comments here. Yesterday, all my dreams... (talk) 02:56, 19 July 2026 (UTC)Reply
Does the WMF plan to voluntarily recognize Wiki Workers United in the UK?
Latest comment: 11 days ago22 comments10 people in discussion
Two days ago, Wiki Workers United, the global WMF staff union, formally requested recognition as a bargaining unit in the UK. Bernadette Meehan, the CEO of the WMF, stated a month ago that the Wikimedia Foundation respects the rights of all eligible employees to determine for themselves whether to pursue union representation. We are committed to ensuring that any process related to union conversations respects the rights and perspectives of all staff. We aren’t only doing this because the law requires it; I want the Foundation to operate this way because I think it’s the right approach.
I have a simple pair of questions for Bernadette, or Jimbo, or anyone else who can speak for the WMF. And I don't want these questions to sound adversarial, and would ask anyone who also wants them answered to not turn this into something adversarial. The WMF has repeatedly reassured both the community and WMF staff that it is not anti-union and that it supports whatever outcome may come. At the same time, a lot of the WMF's messaging has given off the impression that fixing community–WMF relations is a lost cause, that some of us can't be pleased. So, Bernadette, Jimbo, anybody, please understand: I and—I can confidently say—the vast majority of the other 1,123 editors willing to go on strike in solidarity with WWU would love to see this be a turning point where the WMF makes good on what it's committed to and starts to rebuild the trust it's lost in the community and its workforce alike. We're a month out from Wikimania and it would be great for us all to be able to have conversations there in a spirit of mutual trust.
Hi Tamzin, I am responding on behalf of the Foundation as VP, People Experience in the People (HR) Department.
You’re right that the Wikimedia Foundation respects the right of staff to unionize. Every country has different rules around how union membership works. In the UK, employees decide individually whether to join a union or not. Also, Wikimedia Foundation staff in the UK are hired and employed by a third-party employer of record, Safeguard Global. The request for recognition as a formal bargaining unit was sent to them, with the Foundation on copy for awareness. We have confirmed that Safeguard has received the request, and we are working with them; they will respond by next week. We will keep Foundation staff updated as we continue to work through this process with them. -- PThombley-WMF (talk) 20:04, 1 July 2026 (UTC)Reply
Thank you very much for your response, @PThombley-WMF. That complication sounds like an understandable reason for a delay in response; I look forward to hearing Safeguard's (and by extension the WMF's) full response sometime next week, and I'm sure the members of WWU in the UK look forward to it as well. -- Tamzin[cetacean needed] (they|xe|🤷) 20:08, 1 July 2026 (UTC)Reply
Thanks @PThombley-WMF. Looking forward to a rapid resolution to recognise the bargaining unit. I imagine as the 'effective management', it is completely up to WMF to honour this request, even if Safeguard Global would deal with the administrative side of this? In solidarity, —Femke (talk) 🐦 20:18, 5 July 2026 (UTC)Reply
I appreciate the care for Foundation staff that is driving these questions from members of the community. At the same time, I’d like to emphasize that the Foundation is in direct contact with staff on these issues, and, as I shared in my note last week, Safeguard is engaging with CWU. This week Safeguard accepted the suggestion from CWU to engage in discussions about pathways to recognition through one of the avenues suggested by CWU.
While our primary focus is on staff, we also want to prevent misunderstandings about who is involved in decision-making. It is correct that the Foundation must work with the employer of record in a given country, as this is not simply an administrative relationship. We currently have staff in 39 countries, and our relationships with employers of record (not all of which are Safeguard) are not as straightforward as they may seem. Choosing and working with employers of record involves many factors, including local laws and the services they can provide to support staff, such as benefits. At the same time, our mission is rooted in collaboration, and we are committed to making sure our values show up in how we treat one another in every situation, regardless of the formal structures that shape our workplace. While we want to move quickly, we and Safeguard also have a responsibility to follow the required legal and administrative steps and to ensure that any unionization process respects the rights and opinions of all affected staff, while complying with local laws. How this looks may vary depending on the laws and rules in each country. We’ll continue to update affected staff as more information becomes available. Thanks, PThombley-WMF (talk) 18:24, 9 July 2026 (UTC)Reply
After having talked to a few former colleagues -- I get the impression their are several mistruths and misrepresentations in this statement that would not stand up to the Wikimedia movement values centered in principles like verifiability and assume good faith, Sadads (talk) 23:03, 9 July 2026 (UTC)Reply
@PThombley-WMF, you say that How this looks may vary depending on the laws and rules in each country. But we are quite specifically talking about the UK here, and this is the first group of WMF employees who are unionizing. So what, exactly, is varying? And compared to what? You say you want to prevent misunderstandings (emphasis mine); perhaps your statement has done so. What it has not done, however, is "promote understandings". In solidarity, asilvering (talk) 06:06, 10 July 2026 (UTC)Reply
@PThombley-WMF: Your expression accepted the suggestion from CWU to engage in discussions about pathways to recognition is disrespectful towards the WMF staff represented by CWU. WWU says that it requested recognition: you have described that request as a "suggestion". Moreover, on 1 July, you stated that Wikimedia Foundation respects the right of staff to unionize. A request for a right to be respected is not a suggestion for that right to be respected. The right to not be imprisoned for one's opinions is not just a "suggestion" to not be imprisoned. Moreover, engage in discussions about pathways to is a longwinded way of saying that Safeguard Global is not even willing to say that it aims to recognise WWU. Your use of the term "any" in any unionization process further sounds like you are considering the request of WWU recognition as a hypothetical possibility, and that WMF does not acknowledge that this is a unionization process. Boud (talk) 15:16, 10 July 2026 (UTC)Reply@PThombley-WMF: If Safeguard Global objects to unionisation, better state that in plain English. Is there any evidence of Safeguard Global having ever recognised a trade union, anywhere on Earth? If yes, please provide, ideally, w:WP:RS that could potentially be used for w:Safeguard Global, which currently is not encyclopedically notable, although since 2010 has apparently "partnered" with Workday, Inc., which is notable. It sounds like Safeguard Global's main activity is managing legal requirements in writing contracts and getting them signed, and has been unreliable in a few 2026 cases. (As an aside, the monthly fee of a minimum of 500 USD/month/employee sounds steep: if 400 of the WMF employees outside of the US have Safeguard Global as the EOR, that makes 2.5 million USD per year in WMF fees to Safeguard Global if WMF is only paying at the minimum rate.) Boud (talk) 15:16, 10 July 2026 (UTC)Reply
@PThombley-WMF The community is tired of corporate speak that says a lot but means little. If Safeguard Global wants to keep its contract with WMF, it needs to make you happy. This means that they would expand or contract the workforce at your request, and recognise or decline to recognise a union based on your request. If you want to recognise a union but they do not, simply say so and we will gladly support you in finding a new employer of record. I hope your communication with Safeguard is less convoluted than it is with us. ~ In solidarity 🦝 Shushugah (talk) 09:49, 11 July 2026 (UTC)Reply
@PThombley-WMF. I don't know what this is trying to convey. I believe what is written between the lines is that WMF/Safeguard has declined to go for the easiest route under British law (voluntary recognition), given we're past 10 days and we still have not heard any good news related to union recognition. So we're in the slow lane now, which is not the route I had hoped would be taken. In solidarity, —Femke (talk) 🐦 18:35, 11 July 2026 (UTC)Reply
@PThombley-WMF: Per the rules stated in plain English by UK authorities, and given that 24 June + 10 working days = 8 July, then, unless Safeguard Global has violated British law, it has already informed WWU UK that it either (1) agree[d] to recognise the union voluntarily - and beg[a]n collective bargaining; or (2) reject[ed] the request - the union may then apply for statutory recognition; or (3) refuse[d] the initial request but agree[d] to negotiate. You earlier stated that Wikimedia Foundation respects the right of staff to unionize. Thus, assuming good faith, your statement implies that WMF/Safeguard Global chose option (1). Could you please confirm that option (1) was chosen? Thanks! Boud (talk) 20:44, 11 July 2026 (UTC)Reply
I wanted to respond to the request to explain the process more clearly. The main update that I wanted to convey is: Safeguard will have talks to explore recognition through one of the proposals that CWU had suggested. These talks are a common approach in the UK that allows both parties to discuss the proposed agreement and better understand important details, under mutually agreed-upon timelines.
Candidly, there are limits on what more we can say in an open, public forum because this involves an employment relationship managed through a third party. Communications of this nature are typically handled between the employer of record and the union. Any employer of record will have their own considerations in what to share publicly. At the same time, we recognize that volunteers have a stake in Foundation issues, and so I hope you all can appreciate that we are an unusual organization in this respect. This situation directly affects staff at the Foundation, and we want to make sure they have the information they need: their voice in the process and understanding is a priority over updates on-wiki. I appreciate that volunteers may also want more details, and ask for your patience. Processes like this take time, require coordination among several groups, and involve all parties, including staff, having a good understanding of the details. We’re committed to a thoughtful, fair process for all involved, and I will update as we are able. PThombley-WMF (talk) 01:10, 15 July 2026 (UTC)Reply
@PThombley-WMF: You seem to have withdrawn your earlier statement that Wikimedia Foundation respects the right of staff to unionize and replaced it by the weaker statement "Wikimedia Foundation respects the right of staff requests to unionize to be considered as suggestions and to be explored"; you have effectively acknowledged that by the 8 July deadline, WMF/Safeguard Global chose option (3): refuse[d] the initial request but agree[d] to negotiate; and you have failed to apologise for describing the unionisation request as a suggestion (failure to apologise for an insult is not a great values-alignedpeople practice). As for a fair process for all involved, you seem to be denying the power imbalance between management and workers that obstructs fairness: if the union is recognised, then the expression fair process will start to become more credible. Boud (talk) 15:48, 16 July 2026 (UTC)Reply
Maybe we can all take a rest from nitpicking and playing "gotcha" games with WMF public statements? If you're a random person on the internet -- i.e., not an employee of WMF -- then they really don't answer to you, so perhaps we can stop playing the role of "Chief Inquisitor." You are not conducting a congressional hearing here. They are limited in what they can say publicly, as they have explained. Levivich (talk) 16:07, 16 July 2026 (UTC)Reply
There's no need for PThombley-WMF to answer me personally; if my points are valid, they stand independently of who wrote them. A WMF leader responsible for values-alignedpeople practice is not absolved from responsibility for her words. By accepting to work for WMF, she accepted that she is serving the Wikimedia community. Boud (talk) 17:24, 16 July 2026 (UTC)Reply
Again, I hope that the WMF will demonstrate their respect for its employees by voluntarily recognizing the union. This would be a particularly good time for it, as several hundred members of the editing community, staff, and affiliates are presently attending Wikimania. It would be wonderful if the WWU lunch meetup on Saturday could be a celebration. In solidarity, asilvering (talk) 23:36, 20 July 2026 (UTC)Reply
From The New York Times: Wikipedia Is Battling for the Soul of the Internet
Latest comment: 13 days ago21 comments8 people in discussion
Wikipedia Is Battling for the Soul of the Internet[42][43]
"The organization that supports the site, the Wikimedia Foundation, is increasing its lobbying budget and advertising in Times Square. It is charging companies like Google and Meta that gobble up the encyclopedia’s 65 million articles, and throttling access for certain scrapers. And it is expanding its human rights team to better protect volunteers against rising harassment, surveillance and retaliation."
“We are not seeking to be part of culture wars, of political discourse,” Ms. Meehan said. “These are people who want to put knowledge into the world, for people to do with it what they want.
To those convinced of bias on Wikipedia, her recommendation is simple: Become an editor.
“More dialogue, frankly, is what’s needed in this moment,” she said. “So the door is open.”
The NYT and CAMERA keep confusing Wikipedia with the WMF This should come as no surprise to anybody. Even being aware of the existence of the WMF is inside baseball. I would imagine that the number of people who know that the WMF exists is about on a par with the number of people who know that The New York Times Company is a distinct entity from The New York Times. RoySmith(talk)15:08, 8 July 2026 (UTC)Reply
I think the NYT story does a good job of not conflating the two. It makes clear that WMF doesn't handle content and is separate from the community:
To de-escalate the antagonism, Ms. Meehan and her team are addressing misconceptions about the site. Wikipedia is largely decentralized, with independent contributors distilling facts through open debate and agreement. Elected arbitration committees handle dispute resolution. The foundation, with a budget of $208.6 million and about 600 employees and contractors, provides servers and funding but steers clear of content decisions.
I agree that only a very small percentage of internet users understand what WMF is, how it operates and that it attempts not to be involved in content. But why? Because WMF does not clarify this. I think any new user should receive a message explaining this, and banners on pages should do so once I a while. If we are here to "provide knowledge" let us start with this. Yesterday, all my dreams... (talk) 16:26, 18 July 2026 (UTC)Reply
Newcomer editors are already overwhelmed by the huge amount of information, policies etc. they are encountering when trying to participate in our projects. And they don’t need to know anything about the WMF in order to contribute. Why should we bother them with even more notifications? Also anyone who really wants to know can read Wikipedia where Wikimedia Foundation is mentioned a lot. Johannnes89 (talk) 17:54, 18 July 2026 (UTC)Reply
That proves that most people with accounts aren't editors (they don't actually edit, or they edit very little). But that doesn't prove that most editors -- like regular editors who have made more than 10 edits -- don't know about their own talk page, noticeboards, or the WMF. Levivich (talk) 19:09, 18 July 2026 (UTC)Reply
I'm assuming that most editors with such a low edit count don't ever get called to their talk page or a noticeboard, and that most people who randomly create an account and then don't edit or make one or two edits don't really know or care about what the WMF is or does. I think those are safe assumptions to make, but they are assumptions. voorts (talk/contributions) 19:17, 18 July 2026 (UTC)Reply
"Most editors never see their talk page, let alone a noticeboard" is not the same as "most people who randomly create an account and then don't edit or make one or two edits don't really know or care about what the WMF is or does." The second one is obviously true. Not sure about the first one. People who randomly create an account and then don't edit or make one or two edits are not "editors." (You seemed to have missed or ignored the part of my message where I said "regular editors who have made more than 10 edits," who are "editors," not "most people who create an account and then don't edit," who are not "editors.") Levivich (talk) 21:17, 18 July 2026 (UTC)Reply
Yesterday's proposal was that there should be banners and any new user should receive a message explaining that the WMF does not influence content, not just regular editors. voorts (talk/contributions) 21:56, 18 July 2026 (UTC)Reply
I will do my best not to give a long lecture here on the pitfalls of drawing rapid statistical conclusions from incomplete data. But to begin with the data in the table lacks time, so it is incomplete in 2 senses, the time between edits and the years involved. In 2005 data in that table? The talk page there has some broohaha on the issues. Secondly the data is about edits, not views of talk pages. So we can draw no conclusions there. I would never draw any conclusions from a dataset so incomplete. Yesterday, all my dreams... (talk) 20:13, 18 July 2026 (UTC)Reply
I'm not drawing my conclusion from the data. I made an assumption about editors with low edit counts and then pointed to the data to indicate that I beleive an exceedingly small percentage of editors know about and/or give a shit about the WMF. If you believe my assumption is incorrect, you are the one proposing giving all new editors a notice that the WMF does not control editorial content, so it is your burden to prove that this notification is justified. voorts (talk/contributions) 20:22, 18 July 2026 (UTC)Reply
The WMF clarified this in the article that is the subject of this thread. It seems to me that the WMF clarifies this in most of not all of their relevant public communications. Levivich (talk) 19:10, 18 July 2026 (UTC)Reply
UK Wikimedia Foundation staff unionising update - July 2026
"A quick update from the UK Branch of Wiki Workers United on developments since our voluntary union recognition request was filed last month.
As noted in the previous update, on Wednesday 24th June UTAW-CWU sent a letter to the Wikimedia Foundation (WMF) and Safeguard Global (the WMF’s Employer of Record in the UK) requesting recognition of a bargaining unit that would represent all the WMF’s UK staff. They had 10 working days to respond and choose one of three standard options. Either to: voluntarily recognise this bargaining unit/union; reject recognition; or request negotiations.
On 7 July, UTAW-CWU were notified that Safeguard and the WMF had chosen the third option, and as such, we are now entering negotiations with mediation from ACAS (the Advisory, Conciliation and Arbitration Service) – a UK government body that assists with such matters. These negotiations aim to reach an agreement on the proposed bargaining unit, and will proceed over the next few weeks in accordance with standard practice. If voluntary recognition is not reached during these negotiations, we may then file for statutory recognition via CAC (the Central Arbitration Committee) – as the application requirements are already met.
We want to repeat our thanks to the volunteer community for your support. WMF staff will be attending Wikimania in Paris next week in person, including from the WWU UK Branch. Of note, there is a workers’ union meetup on Friday. [This appears to be an error and should be Saturday. -Guy] If you are also attending, come and say hello. We have stickers and badges to share!"
Latest comment: 6 days ago9 comments8 people in discussion
Bernadette Meehan joined the Wikimedia Foundation on January 20, 2026. As part of a community meet and greet (Around the Puzzle Globe), she asked volunteers these three questions:
A few things stood out to me. My overall impression is a little underwhelming: there are some interesting signals but nothing particularly concrete. She has clearly heard some of the community's most common complaints and concerns but there are no actionable solutions other than vague unmeasured platitudes.
Moving fast (and breaking things?)
Meehan writes that standing still is not an option, later saying nostalgia is not a strategy, and calls for bold thinking, faster experimentation and iteration, and difficult decisions. This seems to establish that the WMF is going to start moving quicker and pushing for change, while maintaining the underlying principles of the movement. There is, of course, often quite a gap between what the community thinks should change and what the WMF wants to change.
Declining readership
Meehan highlights an 8% year-on-year decline in pageviews across Wikimedia projects with declining referral traffic from search engines. She argues that Wikipedia must meet people where they are – on other platforms, in new formats, and through new partnerships. I'll be interested to see how Wikipedia adapts to this and how editors can be part of that adaptation. I do have concerns about how far the push towards apps, personalised feeds, video, and off-platform distribution might go and if the heart and soul of Wikipedia might be lost in a race to the bottom.
Community relations and changes
Meehan says that many volunteers asked for greater transparency and stronger communication, and that the Foundation needs to improve how it listens, communicates, collaborates. I don't reckon anyone will disagree with that.
The changes to the Community Wishlist are mentioned specifically:
Recent changes to the Foundation’s approach to the community wishlist prompted strong reactions and generated significant discussion. I take those conversations seriously. While the intent of the changes is to better meet volunteer needs by ensuring that multiple Foundation teams – not just one – share responsibility for addressing wishes, community feedback shows clearly that the Foundation must build a shared understanding of internal structural changes when they directly affect our communities.
The last sentence struck me as particularly unhelpful corporate speak. I think it means something like "when we reorganise something that affects volunteers, we need to explain what we have changed and why". The problem identified here seems to be that the Foundation did not establish a sufficient shared understanding of its internal changes, rather than that we had substantial objections to the changes themselves. I read this as "we need to communicate organisational changes better", which is not necessarily the same as "we need to reconsider changes when the community objects to them".
There is a similar point a bit later. Meehan says that disagreement is inevitable and the goal is not unanimous agreement, before asking how we can harness divergent ideas to move forward. I think that's reasonable as a general principle, but in the context of Foundation/community relations it raises the governance question of what happens when the Foundation believes a change is necessary and the community disagrees.
Artificial intelligence
Meehan says that 65% of the most resource-consuming traffic Wikimedia receives is coming from bots, with automated traffic placing increasing strain on Wikimedia's infrastructure and sometimes degrading the experience of human users.
There is also the problem of the contributor pipeline. People who no longer come to Wikipedia because they get our content through "zero-click" experiences are also less likely to stumble into editing and become contributors themselves. The Foundation appears to regard preserving this reader-to-contributor loop as a strategic issue. qcne(talk)11:23, 20 July 2026 (UTC)Reply
Ever-increasing readership or even stable readership cannot be the goal. The goal should be to build the best possible encyclopedia. Anything else is a distraction to the project. What does standing still is not an option mean? Where exactly are we moving? Ita140188 (talk) 12:27, 20 July 2026 (UTC)Reply
What does standing still is not an option mean? Where exactly are we moving? It means that standing still and fading into irrelevance is not an option. People contribute to Wikipedia because it is widely read. If it stops being widely read, the incentives to contribute fade away. Why would anyone want to build a knowledge base that isn't actively serving anyone? – SD0001 (talk) 09:22, 21 July 2026 (UTC)Reply
I don't think that's true, or at least not fully true. Open culture generally works because contributors are scratching their own itch, and the contribution system makes it easy for that work to benefit other people. I'm biased because I tend to work on more esoteric topics, but when I write an article about something, I benefit by doing the work and clarifying my own thinking and understanding on the subject, and by having my work in a permanent repository, nicely hyperlinked, where I can look it up again with ease when my knowledge has faded a bit. That's not to say declining readership is harmless, but I don't think most of our editors are working with one eye on the page views meter. Choess (talk) 04:55, 25 July 2026 (UTC)Reply
It seems no matter who occupies that job, they always sound the same, right down to the "if you disagree with a decision, it must mean we didn't explain it well enough." Levivich (talk) 14:48, 20 July 2026 (UTC)Reply
Re: "Meehan says that many volunteers asked for greater transparency and stronger communication", picking one Wikimania and publishing exactly how much it cost and what was bought with that money would go a long way toward convincing me that she is serious about transparency.
(Please, people, don't reply with a link to a document that lists total expenditures for "Wikimanias, air conditioning upgrades, office supplies, consulting services and private jets")
Latest comment: 2 days ago15 comments9 people in discussion
For the many people who are not subscribed to the mailing list, important information was shared about board election "reform". From my perspective, it's incredibly concerning. [44]
Seems like WMF is trying to strengthen the empirical evidence in favour of the Iron law of oligarchy conjecture: According to Michels, all organizations eventually come to be run by a leadership class who often function as paid administrators, executives, spokespersons, or political strategists for the organization. Far from being servants of the masses, Michels argues, this leadership class, rather than the organization's membership, will inevitably grow to dominate the organization's power structures. The conjecture has the word "law" in it, but that does not make it a sociological certainty.Or to put it more bluntly: the proposed set of criteria sounds like an oligarchical self-coup.There are already several excellent comments at m:Talk:Wikimedia Foundation elections/2027/Proposed eligibility criteria objecting to the self-coup. Boud (talk) 15:22, 21 July 2026 (UTC)Reply
AFAIK this is mainly from one/two board members, it doesn't seem to have been socialised much, and they seemed receptive/good-faith in the Meta channel on WP:DISCORD. Anxiety about it is understandable though given how often community concerns get relegated/ignored re this type of thing Kowal2701 (talk, contribs) 20:12, 21 July 2026 (UTC)Reply
I get the impression I also get chalked off and ignored as a 'WMF hater', which is frustrating and I'll probably leave the server. But this seemed to be a mistake based on one person's perspective (they mostly just edit plwiki), but at the very least they should have realised this'd poke the hornet's nest, "community-selected" members are meant to be in touch with the communities fgs, not up in ivory towers Kowal2701 (talk, contribs) 21:09, 21 July 2026 (UTC)Reply
I have to be honest, at this point, I've given up on caring too much about the Board of Trustees, since I don't believe there's any real path to the community having a real voice anymore. The only two times the WMF appears to really care about what the community thinks about things is, in my view, when the community's view happens to correspond exactly with what the WMF board wants anyway and secondly, when the community is so mad that there's a real danger to day-to-day Wikipedia that might affect fundraising. Fretting about the WMF co-opting the movement are 2010 actions; it's already happened. CoffeeCrumbs (talk) 14:31, 22 July 2026 (UTC)Reply
Sadly you're right. I've lost count of how many WMF-owned committees now claim control of our encyclopedia. However, there may be hope. Although the BoT was laid as yet more astroturf, it has somehow acquired a majority of community members. Although they were all WMF-approved, with candidates likely to effect actual change being carefully weeded out, this worm may still be able to turn. Certes (talk) 08:48, 23 July 2026 (UTC)Reply
We've never tried developing a consensus platform (a list of board resolutions) and voting for candidates that pledge to vote for the platform. Levivich (talk) 14:58, 23 July 2026 (UTC)Reply
Sohom Datta'smessage on the mailing list summarises the entire crux of the issue. One dosen't need to be an experienced 20-year veteran to see where the WMF is headed with its criteria for Board membership. It's therefore not surprising they've kept their token 'open' discussion very much inside baseball. IMO someone should at least ping all the signatories of the petition begun byuser:Clovermoss and the participants of all the other discussions about it. Kudpung กุดผึ้ง (talk) 08:03, 29 July 2026 (UTC)Reply
Latest comment: 6 days ago2 comments2 people in discussion
Here is a quick overview of highlights from the Wikimedia Foundation since our last issue on July 3. Please help translate. The work follows annual plan goals for this fiscal year (July 2026–June 2027).
Wikimania 2026: Wikimania is happening this week! After the event, all streamed sessions will be linked in the program on Eventyay and later uploaded to Commons.
Grantmaking: The Global Resource Distribution Committee has published a Grantmaking Strategy draft that sets out a renewed approach to how the Wikimedia Foundation's Community Fund is distributed across the Wikimedia Movement. The GRDC is requesting feedback from volunteers and affiliates, regardless of whether they are grantees or not.
Movement Ecosystem: A proposal that would update movement affiliate recognition and establish new, connected criteria for eligibility to receive Community Fund grants is now available for community review. You are invited to read the proposal and participate in the discussion until August 7.
Women+ contributions in Wikimedia Tech: A guide based on lived experience on how to address some of the invisible barriers for women+ in more technical Wikimedia spaces and recommendations to become more inclusive. Help further by filling out this survey to better understand technical contributions by women+ across Wikimedia projects until July 20.
Structured Experimentation: A reflection on the first year of structured experimentation highlights successful experiments such as Paste Check, Reference Check, and Tone Check, which improved editing outcomes and have been rolled out to more users, as well as experiments that did not lead to product changes.
Revise Tone test ended: The A/B test of Revise Tone ended on July 9. It showed that newcomer task completion rates increased by 38.7% compared to the default Copyedit task, with no decrease in edit quality. The feature is now available for everyone on the Arabic, English, French, and Portuguese Wikipedias. The plan is to release Revise Tone to more wikis.
Wikidata: The latest Wikidata Platform newsletter (July edition) shares how to identify and rewrite queries that rely on Blazegraph-specific extensions and affected by the migration off Blazegraph.
Discussion Tools: On English Wikipedia, DiscussionTools' Usability Improvements has now become default for talk pages. You can opt-out of these changes at any time in user preferences. With this, Discussion Tools are now fully available at all wikis.
Tech News: The latest highlights from Tech News week 28 and 29 include the new Parsoid parser continues to be deployed to additional wikis, making it easier to introduce new reading and editing features. See also the 72 community submitted tasks that were resolved over the last two weeks. Overall, from April – June 2026 about 337 community tasks were resolved by the Wikimedia Foundation.
Language inclusivity at Wikimania 2026: New approaches to translation and interpretation will be tested at Wikimania this year.
Call for submissions open for Wikimedia Latin America Conference 2026: The Wikimedia community in Latin America has opened the call for session proposals for the Wikimedia Latin America Conference 2026. Community members are invited to submit proposals for the conference program by August 10.
UK Online Safety Act: Ofcom, the United Kingdom's Office of Communications, announced that Wikipedia is not designated as a Category 1 service under the Online Safety Act (OSA). This is an important and welcomed outcome as a Category 1 designation could have included privacy and safety risks to our global community of volunteers.
UN Open Source Week edit-a-thon: Volunteers created 60 new Wikipedia articles and made nearly 700 updates to improve Wikipedia's coverage of UN and open source topics at the second UN Open Source Week edit-a-thon co-hosted by Wikimedia Foundation.
"Don’t Blink": The latest developments from around the world about protecting the Wikimedia model, its people and its values.
"A Wiki Minute" videos: New videos are added to the series answering some of the most common questions such as "Do you still need Wikipedia when AI can answer anything?" and "Does Wikipedia push a political agenda?".
Wikipedia 25 brand collaboration in Indonesia: On 4 July, the Jakarta-based street wear company Ageless Galaxy launched a Wikipedia 25 collection, the first ever Wikipedia Brand Collaboration in Asia. The apparel collection featured hats, t-shirts, and a jigsaw puzzle cardigan. Wikimedians in Indonesia joined Ageless Galaxy for a launch party.
Board Elections Eligibility: There are new proposed eligibility criteria for standing as a candidate in Wikimedia Foundation Board of Trustees election now available for feedback. The requirements are more specific and detailed than in years past, to both inform the community of what the Board needs and to create multiple pathways to the Board for Wikimedians.
For information about the Bulletin and to read previous editions, see the project page on Meta-Wiki. If you have feedback or suggestions about the bulletin, let us know at foundationbulletin@wikimedia.org. For questions about the Wikimedia Foundation's work, contact us!
Undermining WMF's relationship with the Wikimedia movement community
Abuse of power
Undermining the workplace culture at WMF
Avoiding accountability
Incompetence
Potential financial waste
Moreover, Wikifired states that s/he reached out to many of my colleagues at WMF and learned of many other instances of potentially problematic conduct by WMF management and HR. It was very disappointing to learn how widespread this is, and how there's apparently no accountability for it. ... And also stories that started with something like 'you think that's bad, here's what happened when I dissented or tried raising concerns about WMF management'. Stories where, in response to employees raising concerns in good faith, what they often heard back was 'if you don't like it, then leave' or 'you need to practice curiosity'.
To put it bluntly, Wikifired is asserting that something is rotten in the state of WMF leadership. Corporate waffling and "we can't talk about this for legal/privacy reasons" are unlikely to be accepted by the community as an adequate response. We don't need to know the identity(ies) of AL2 or any other WMF managers who have violated Wikimedia norms; we need their unethical behaviour to cease. Boud (talk) 17:14, 24 July 2026 (UTC)Reply
They're not anonymous, the about me section makes it clear that this was an actual verified former employee of the foundation. This story is new to me, but unfortunately the underlying subject matter isn't. I've heard multiple horror stories with similar substance. The internal leadership at the WMF needs to change and be much healthier than it is. Clovermoss🍀(talk)19:48, 24 July 2026 (UTC)Reply
Side point on anonymity: I agree that Wikifired seems to intend to be not too difficult to identify, but the LinkedIn link is closed access, and it's up to Wikifired or someone else to state his/her name explicitly if desired. I'm assuming that "Wikifired" is an acceptable pseudonym until then. Boud (talk) 20:28, 24 July 2026 (UTC)Reply
I have a regular LinkedIn account and I could see it. I'm not saying go rake them over the coals, I just wanted to preemptively stop people from questioning whether they were actually a former employee. I'd hope the details are specific enough that people wouldn't but yeah. I'm fine with the pseudonym. Clovermoss🍀(talk)20:56, 24 July 2026 (UTC)Reply
I think it is important to not identify Wikifired, even if they say it is OK. Knowing Wikifired's identity, it would be trivial to figure out the identity of "AL2", which would not only violate Wikipedia policy but would be morally and ethically wrong. --Guy Macon (talk) 21:30, 24 July 2026 (UTC)Reply
I've heard severance in exchange for these non-dispargement clauses are quite common, and it's partly why I made such a big deal out of it in the sprawling discussion about the dissolution of CommTech weeks ago. Who knows how much money has been spent keeping people quiet. And when part of the WWU's stated aims is preventing inconsistent firing, one also wonders how many former employees truly believed the WMF lived up to its values and wouldn't suddenly find themselves without a job and in a financially precarious situation for trying to do something about this working environment. Clovermoss🍀(talk)21:43, 24 July 2026 (UTC)Reply
I note there's no mention in that about the non-disparagement attached to the severance. Also, I've heard they recently silently changed it from "one month per year" to "up to one month per year", just before things like the Community Tech firing and the slashing of the states and countries they'll hire from. Anomie⚔18:20, 25 July 2026 (UTC)Reply
"The Wikimedia Foundation is a remote-first organization with staff members including contractors based in 40+ countries.
Please note that we are currently able to hire in the following:
US States: Arizona, California, Colorado, Connecticut, District of Columbia*, Florida, Georgia, Idaho, Illinois, Indiana, Iowa, Maryland, Massachusetts, Michigan, Minnesota, Missouri, New Jersey, New Mexico, New York, North Carolina, Ohio, Oklahoma, Oregon, Pennsylvania, Puerto Rico*, Rhode Island, Tennessee, Texas, Utah, Vermont, Virginia, Washington, West Virginia, Wisconsin and Wyoming (*US Territory or Federal District)
Countries: Brazil, Canada, Colombia, France, Germany, Ghana, India, Indonesia, Italy, Kenya*, Mexico, Morocco, Netherlands, Poland, Singapore*, South Africa, Spain, Switzerland and the United Kingdom.
Thanks, Guy! Anybody got any guesses what these states have in common that the omitted states do not have? I can't discern a pattern. Montana doesn't have at-will employment but that's the only thing sticking out for me. The states listed include some of the states with the strictest employment laws, so it doesn't seem like that's the difference. Why would they hire in NC but not SC, for example? In Utah but not Mississippi? How odd. Levivich (talk) 22:12, 25 July 2026 (UTC)Reply
the fact that only some EU countries are listed seems strange to me. Surely the rules are pretty similar between eu members. Bawolff (talk) 22:28, 25 July 2026 (UTC)Reply
Also strange: The WMF will hire you if you simply live in Ghana, but in Kenya you have to be a citizen or permanent resident. Why? And why no country in Central America, Eastern Europe, or the Middle East? --Guy Macon (talk) 23:38, 25 July 2026 (UTC)Reply
If I'm not misremembering, I recall hearing that Singapore had a law change recently cracking down on non-permanent-residents working remotely. Kenya may have similar restrictions. Anomie⚔00:59, 26 July 2026 (UTC)Reply
Yes, there was indeed a recent change of that sort in Singapore. I can't speak to how that has affected Wikimedia staffing as I'm uncertain what is and isn't covered by my NDA. brooke (talk) 05:56, 26 July 2026 (UTC)Reply
It may be as simple as "WMF currently has things set up to withhold state taxes in these states, and they're refusing to add any more to that list". Anomie⚔00:54, 26 July 2026 (UTC)Reply
When this was discussed before, it was explained with the fact that WMF needs to provide healthcare benefits across the state lines and that's not easy or cheap if the company has to do it separately in all 50 states. Don't know how valid this explanation is, just what I've heard from WMF-related people. stjn05:17, 26 July 2026 (UTC)Reply
Note that while I was working in Oregon, I received my WMF health insurance as a Blue Shield of California "out-of-state" plan. I don't know the details of how this works on the internal/HR end and whether there's anything particularly difficult to adding more states. brooke (talk) 05:59, 26 July 2026 (UTC)Reply
I have a problem with like 98% of things the WMF does, but this isn't really one of them. This is bog standard in situation where there's no requirement for severance (outside of a labor agreement, a contract, or a company policy). Now, there are exceptions when it comes to enforcement, in both federal and California law, (for example, abuse and harassment under federal law and illegal company actions in California). But non-disparagement clauses themselves are not unusual when the a company pays in severance agreements. CoffeeCrumbs (talk) 11:42, 25 July 2026 (UTC)Reply
Just because it's normal for American companies doesn't make it right. Where I live, you cannot make severance conditional upon signing an NDA. If you qualify for it, you get it, and that's that. I think there's some serious human rights issues to this norm. Clovermoss🍀(talk)12:07, 25 July 2026 (UTC)Reply
When NDA stifles criticism of the WMF afterwards, as I know is the case, it is not normal or acceptable, even if it is a standard practice elsewhere. We should be better than for-profit companies on this. It should be imperative for the WMF to disavow the parts of the NDA that currently prohibit employees to tell their stories like the one above or criticise various wasteful initiatives. stjn12:54, 25 July 2026 (UTC)Reply
Coffee is right in that non-disparagement clauses are "bog standard" for severance agreements, in the US but not only, also in other countries (but not everywhere), and not just in the for-profit sector, but also in the non-profit sector.
At the same time, I'm quite persuaded by Clover and stjn's point that just because something is standard doesn't mean it's right, or moral, or that the WMF should follow the crowd.
It is very possible for the WMF to not require non-disparagement clauses in severance agreements (while still receiving consideration that makes the severance agreement legally enforceable).
It is also very possible for us, the community, to force the WMF to stop using non-disparagement clauses. It's the thing I keep harping about that very few of us seem to care about: we control a majority of the Board! That means we can set WMF policy!
All it takes is one Board resolution to get rid of non-disparagement clauses in severance agreements:
Board Resolution No. 1: RESOLVED, that the Wikimedia Foundation shall not include non-disparagement clauses in severance agreements.
That's it, that's all it takes. If the Board passed that, the WMF would be required to comply.
There are currently 6 elected Trustees and 4 appointed Trustees (the max under the bylaws is 8-7). The elected Trustees, by design, are a majority of the Board.
If the Trustees we elected won't support that Resolution, then we should elect Trustees who will, and keep electing them until that Resolution is passed. It might take two three-year cycles to get the right Trustees on the Board.
We could repeat this process for everything we want to change.
The current Trustees and Officers know this, and that's why they're "vetting" candidates and now trying to limit candidate "qualifications."
If they take away the community's control of the Board before the community uses that control to push through reforms, well, then we will have lost control, forever.
Given this public record by several different people of misbehaviour and attempts to hide that misbehaviour by WMF management, this raises the credible hypothesis that the real reason for the removal, just two days before the Oct 2025 WMF Board election, of Bluerasberry as a candidate in the election was that WMF leadership refused to accept having a Board member who is opposed to misconduct: excerpt from Bluerasberry's June 2021 statement: "There is an unacceptably high rate of Wikimedia Foundation staff and consultant misconduct in the Wikimedia Movement. Where money is involved, there are paid Wikimedia Foundation representatives who violate the meta:Universal Code of Conduct and stifle free discussion about the regularity of this problem."Boud (talk) 08:54, 25 July 2026 (UTC)Reply
To me, the way AL2 portrayed the critical community member came across as very negative. It created the impression that the critical community member was a bad, hostile person with negative intentions towards WMF.
Likely what happened is that WMF was offering compensation for the amendment of the NDA to include such a clause and it's just being described strangely. Jerod Lycett (talk) 01:06, 26 July 2026 (UTC)Reply
Wikimedia has a boilerplate separation agreement, into which they drop the specific amount of severance offered for individual cases. The agreement includes both a non-disparagement clause and a non-disclosure clause.
I can't speak about the actual details of the agreement, however, due to the agreement.
Note that Wikimedia considers severance to be "optional" in exchange for agreeing to the conditions of the separation agreement; in some countries severance is in fact a legal right and may have a minimum amount, which has lead to occasionally spicy times when someone is let go outside the US. brooke (talk) 06:07, 26 July 2026 (UTC)Reply
I'll note that I also brought this up at Jimmy's talk page if anyone wishes to read that discussion. He hasn't said anything yet, but he's probably the most reachable board member when it comes to likelihood of responding to queries on-wiki. Clovermoss🍀(talk)08:15, 26 July 2026 (UTC)Reply
If there are similar stories from former (or even current) WMF employees, hopefully some news outlets will pick up on this. Some1 (talk) 15:59, 26 July 2026 (UTC)Reply
Noteworthy quote from the article: And don’t even get me started on WMF leadership’s hilariously frequent use of “practice curiosity” in response to employees questioning the ethics of WMF leadership’s conduct.Some1 (talk) 16:36, 26 July 2026 (UTC)Reply
Recent questions about the Foundation and US unionization request
Latest comment: 17 hours ago65 comments29 people in discussion
Hi All, I am posting this on behalf of The Wikimedia Foundation as the Chief People Officer. During Wikimania last week, we received several questions about Wikimedia Foundation staff unionization. Most questions focused on the recent request from the Communication Workers of America (CWA), affiliated with the American Federation of Labor and Congress of Industrial Organizations (AFL-CIO), to voluntarily recognize it as the exclusive collective bargaining representative for eligible US-based staff. We sent a message to staff and posted a statement on the Foundation website to openly answer questions we have been receiving. You can read more in our statement and the FAQ that was posted here.
What a shameful decision to tie the process to the willingness of Trump-controlled NLRB. Everyone involved in denying the voluntary recognition should resign in disgrace or be driven out. I have absolutely zero faith that @BMeehan-WMF didn't just tell a bald-faced lie when she refused to address the question at Wikimania by saying that they're just too focused on the conference. We are ruled by lying politicians, and not even the ones that are particularly good at lying. A sad day for the Wikimedia movement. stjn16:19, 27 July 2026 (UTC)Reply
CBasssherizen-WMF, be assured that the community of editors -- even those of us who have deep disagreements with some of the things that the WMF does -- will not tolerate abuse of individual WMF staff. I am so sorry that you had to read that. My personal view is that there are no individuals that we can point to and blame but rather that we are looking at an institutional priorities that have persisted through multiple changes of WMF leadership. --Guy Macon (talk) 16:59, 27 July 2026 (UTC)Reply
People are referring to what executives said at wikimania during the fireside chat, that they wouldn't say anything until Monday because they were busy. I think it's perfectly reasonable to suggest that they made this decision to avoid backlash at the conference itself. Clovermoss🍀(talk)17:08, 27 July 2026 (UTC)Reply
Wow, calm down. To write such a big comment (and post a warning on my talk page) you should be aware of the context, but I see you're not, go read my comment again. Nemoralis (talk) 17:16, 27 July 2026 (UTC)Reply
They were explicitly asked about voluntary recognition and said they didn't have anything to say until Monday because of Wikimania. I refuse to believe that this post was agreed upon in less than a day. I think Bernadette should've bravely said they won't agree to it and got some well-deserved boos. Instead, she didn't tell the truth. The exchange happens here.stjn17:21, 27 July 2026 (UTC)Reply
If there's gonna be sanctions over stjn's -- or Nemoralis -- words, then you better be ready to sanction me too. I agree with every word they both said. ⇒SWATJesterShoot Blues, Tell VileRat!17:31, 27 July 2026 (UTC)Reply
I don't owe you, or anyone else, "evidence" that I agree with an opinion *someone else* expressed. If you want to pick a fight with someone, Guy, I'm not the one -- WP:AGF is not optional, and that's *twice* in this thread that you've gone off half-cocked at something while being wrong about who wrote it. Do it a third time, and you're not gonna like the consequences. ⇒SWATJesterShoot Blues, Tell VileRat! — Preceding undated comment added 17:55, 27 July 2026 (UTC)Reply
[Redacted to avoid further drama] I repeat; WMF employees are not your personal punching bag. Argue with what is said. Don't insult the one who said it. --Guy Macon (talk) 14:41, 28 July 2026 (UTC)Reply
I think there's a difference between treating employees as a punching bag (clearly wrong) and suggesting that the chief executive might have mislead her audience (or [told] a bald-faced lie) when she was talking on a contentious issue.
I completely understand your position about defending WMF staff from unwarranted attacks — and I broadly agree. But criticising the chief exec of an organisation for public comments she made that, in hindsight, could appear not to have been in good faith is perfectly legitimate commentary and describing that as a lie is not unusual. I would suggest that you might want to back away from dying on this hill. — OwenBlacker (he/him; Talk). I support Wiki Workers United15:15, 28 July 2026 (UTC)Reply
But it's OK for community members to be their own personal punching bags? I look forward to you keeping the same energy for this: When AL2 was saying that we should not build publicly open tools that show the impact of WMF’s grant funding because of potential scrutiny and criticism of WMF by Wikimedia movement community members, she singled out and named a specific critical Wikimedia movement community member as an example. To me, the way AL2 portrayed the critical community member came across as very negative. It created the impression that the critical community member was a bad, hostile person with negative intentions towards WMF. My manager did not object to or question AL2 doing this to a community member. She appeared to be in agreement with how AL2 treated that person.Gnomingstuff (talk) 18:46, 28 July 2026 (UTC)Reply
I'm not sure whether you're talking to me or not, but I'm quoting a blog post from a former Wikimedia employee. I don't know what I could possibly be accusing you of. Gnomingstuff (talk) 05:40, 30 July 2026 (UTC)Reply
(Off-topic) Thank you for clarifying to Gnomingstuff what happened. (I will say that the confusion ended up being helpful to me as I didn't notice Guy Macon's changes until now.) --Super Goku V (talk) 09:28, 31 July 2026 (UTC)Reply
This is an extremely disappointing statement. It says that Foundation leadership respects the right of staff to unionize, if they choose to do so. That decision rests with them. But that decision has been made - by a supermajority of US-based staff. The only thing standing between the US WWU and recognition is WMF executive leadership, who could recognize the union today if they cared to.
Editors less familiar with union organizing may be wondering what is wrong with holding a secret ballot on the question. After all, that certainly sounds perfectly fair and right, and, if a supermajority has signed union cards, surely a supermajority will vote yes to the union? Certainly, I would expect them to, if that ballot was held today. But the ballot is not today. It will be later - months later - even if the NLRB does not itself add more roadblocks to the process. And the WMF can spend the intervening months pressuring and propagandizing employees. You can already spot some examples of the standard anti-union playbook in their FAQ. I particularly like Our goal is to ensure that everyone can make their own decision in an informed and civil environment. This is an old standard ("informed" is code for "'educated' on the 'dangers' and 'risks' of joining a union", and I expect you can figure out what that in turn is code for), and it rings even more hollow than usual, given that a significant driver of WMF staff dissatisfaction is related to information issues: employees feel that a) they are insufficiently informed, b) they are instructed not to inform us of things they believe we should be informed of, and c) when they inform upper management of issues, they face retaliation.
You can see another example of how this rhetoric works in the section, "How has leadership historically responded to concerns raised directly to leaders by staff?" I'll quote it in full:
We have a record of listening carefully to the concerns raised by staff and have implemented several changes based on staff feedback to strengthen their experience. We have taken steps to improve communication and transparency, including more opportunities for staff to engage directly with leadership and more regular updates about the organization’s direction. We are committed to improving and are continuing to strengthen how we gather feedback and how we act on it. That said, we recognize that if the CWA becomes staffs’ exclusive representative for purposes of collective bargaining, we will have a legal responsibility to avoid dealing directly with staff on their pay, benefits, and other terms and conditions of employment.
I have not added "citation needed" templates to various points in the first two sentences, but that hurt my Wikipedian soul a bit. For the rest: if they're committed to improving, what's wrong with doing so alongside the unionization efforts? And notice how the final sentence - which expresses a core purpose of a union - has been framed to suggest this is a bad thing? Dealing with staff directly on their conditions of employment means the employer can take a divide-and-conquer stance and keep employees in the dark about what other employees are receiving. It means a state of "every employee for themselves". It means employment can be terminated, as we saw with commtech, with little forewarning and and seriously poor planning. This is not good for the employees, it is not good for the volunteer communities they serve, and, frankly, it's not good for the WMF either. But now they have plenty of time to try to let this kind of misdirection marinate.
I don't have anything to add that asilvering hasn't already said. This is an incredibly disappointing statement and that FAQ at the bottom might as well have been pulled directly from a Union-Busting 101 handbook, it's all the same rhetoric I've seen 100 times before. --Grnrchst (talk) 17:21, 27 July 2026 (UTC)Reply
As I said way back when the CommTech dissolution happened, they sound exactly like Amazon does here. I'm tremendously disappointed that Wikimedia Foundation executives are not acting within the spirit of the movement itself. Clovermoss🍀(talk)17:24, 27 July 2026 (UTC)Reply
+1 to this, as I stated to Bernadette Meehan at the fireside chat, their actions are not matching their words, and actions are important. I'd have respected Bernadette more if she directly said she opposed unionization. I'd have been very disappointed in WMF, but I'd have given her credit for not using corporate doublespeak to obscure their anti-union views. Abzeronow (talk) 00:38, 28 July 2026 (UTC)Reply
Major own goal here by WMF management. Shameful indeed!
As a U.S. trained labor lawyer and past organizer with WWU, this statement, alongside the WMF’s public facing community post reek of union-busting disinformation. What’s critical for the community to know is that a supermajority of the workers already did vote by signing a card. This “extra step” is just a delay tactic that employers are advised to do when they don’t want a union. They are doing the same thing in the UK as well.
These WMF workers have already been extremely brave by expressly affirming their commitment to a union. Years of efforts to get this accomplished. This is not some rash decision by them. Moreover, the community supports their union too! It literally makes no sense as to why WMF thought this was the right decision here in this critical moment.
Finally, as a general reminder of the lack of good faith that was demonstrated by WMF here, there was a legal process for WMF, offered by the union, that intentionally avoided delay and going through the Trump administration. It is a very well-trodden path which includes a third-party arbitration process to verify the integrity of the signed members. Again, I believe this demonstrates a willful attempt to ignore the supermajority of their workers’ desires and, worse, causes a further erosion of trust and goodwill between WMF management and their staff going forward. Jbernick98 (talk) 17:55, 27 July 2026 (UTC)Reply
As stated, the decision to unionize belongs to staff. – Correct, and a supermajority of staff have already made the decision to unionize. It's simple, any unnecessary delays in recognizing this fact are nakedly anti-union. fifteen thousand two hundred twenty four (talk) 21:08, 27 July 2026 (UTC)Reply
I agree with the essentially unanimous criticism here. I look forward to hearing next steps from the WWU and to doing whatever in my power I can to help oust the WMF’s existing leadership. signed, Rosguilltalk21:15, 27 July 2026 (UTC)Reply
The WMF and its CEO, BMeehan-WMF, (who served under Biden and Trump administrations), are well aware that Trump cut funding for the National Labor Relations Board, which causes several months of undue bureaucratic delay pending a supervised vote.
At best, the WMF believes that a majority of its US staff are lying when they sign union cards and wants additional verification. At worst, the WMF knows that a majority of its staff have signed union cards and hopes that, over the upcoming months, pro-union staff will vote against in a desperate bid to make this stressful situation go away. Communication Workers of America, the union representing US WWU staff, has a union-election success rate of 80%, above average for private-sector unions. They are serious about winning and have dealt with employers far worse than Wikimedia, so even with NLRB elections, there is good reason to be optimistic, but it is clear we cannot be complacent. Source
This is a common tactic in union-busting campaigns. According to labour sociologist Joshua Mayor, unionization efforts sometimes fail not because workers are against unions, but because they do not believe they can win. Source.
Between now and the yet-to-be-scheduled NLRB-supervised election, there is time for WMF to do the right thing and recognize the will of the majority of its staff voluntarily. WMF is supposed to be better than the average tech company, and largely is, but right now it is behaving worse than Microsoft when it comes to labour rights.
As a long time donor and editor, this an outrage. The work that these folks do is vital to the world. They deserve the opportunity to collectively bargain for a fairer workplace.
This is incredibly disappointing to see, especially when conditions that may seem "generous" in the US are below legal minimums in most of the Global North.
Good employers who value their employees have nothing to fear from unionisation. The way the Foundation is responding to a good-faith unionisation that is supported by a supermajority of US staff is not how a good employer behaves — and when the Foundation publishes this statement the day after Wikimania, it is hard to see how @BMeehan-WMF's comments at Wikimania could be anything other than misleading delaying tactics. (And the sophistry around UK unionisation efforts is not much better, given that the WMF is the "effective employer", even if it is handled through a "company of record".) Assuming good faith is all good and well, but it does come to a point where repeated assumption of good faith is merely naïveté.
This is profoundly disappointing and yet another example of the Foundation leadership being out of touch with the Movement. We should be able to expect better from the Wikimedia Foundation and its leadership. Frankly, it is not good enough. — OwenBlacker (he/him; Talk). I support Wiki Workers United14:04, 28 July 2026 (UTC)Reply
I fully agree with Asilvering. I hope that nobody mistakes my objection to personally attacking the WMF staffer who wrote the message with in any way agreeing with the message itself. In my opinion, this is the same old story that has been going on for decades: The WMF long ago gathered enough money to keep Wikipedia running on the interest without any further fundraising banners or accepting large donations from corporations as if there were no strings attached. They continue moneygrubbing for the purposes of -- well, we don't really know, do we? It's a huge secret where, exactly, the money is being spent other than the minimum disclosure required by law. --Guy Macon (talk) 17:56, 27 July 2026 (UTC)Reply
Today was a day to show trust in staff. Because trust is what we will need in the years ahead, which are likely going to be as chaotic as the last couple of years if not more so. The framing in this letter does not convince me the Foundation leadership is neutral in this matter. A small example: rather than acknowledging the WWU effort to build a global union (or network of bargaining units), the text focuses on the CWA, which cannot represent global staff, seemingly to diminish international support: It is important to note that the CWA is only seeking to represent a portion of US-based staff, and there is no global union option due to different labor laws across countries and jurisdictions. Also note how the text emphasises the 'portion of', which is technically correct as US labour law exclude managers, but downplays the role of the union in representing most workers. The FAQ emphases what leadership calls 'good working conditions'. The listing of benefits tries to convince staff that all is well and that unions might not be necessary. What the text fails to mention is that these are not contractual and can be changed unilaterally. In tumultuous times, we need staff that has the security of actual 'secondary labour conditions' as we call them in the Netherlands. The 'benefits' are especially questionable when the "generous" 15-day holiday allowance would be illegal in every single EU country. Today was a day to show trust in staff. Instead, we've started a rough road of persuasion and spin. Fortunately, as a movement, we are exceptionally well trained to recognise this. I'm confident staff will prevail. In solidarity, —Femke (talk) 🐦 21:02, 27 July 2026 (UTC)Reply
So where's the petition to have the Board pass a resolution requiring the WMF to voluntarily recognize unions that receive supermajority employee support? Levivich (talk) 23:12, 27 July 2026 (UTC)Reply
How likely is it that this third (!) petition will accomplish anything? Hard to imagine that the Board of Trustees will (publicly or not) disagree with their CEO and "instruct the CEO to voluntarily recognize the union"... but who knows, I guess. Some1 (talk) 01:47, 28 July 2026 (UTC)Reply
Petition fatigue is a real concern. This 3rd petition is newly directed at the BoT who till now did not feel obligated to "interfere" or state anything till now. Prior petitions were directed at editor community and the Foundation. ~ In solidarity 🦝 Shushugah (talk) 10:47, 28 July 2026 (UTC)Reply
The Wikimedia Foundation’s statement and its frequently asked questions section is full of very carefully-worded language that is common among companies and organizations that have fought against unionization. For example, the FAQ includes a long section about the benefits that Wikimedia Foundation already offers its staff, and the statement suggests that there is a “wide range of views on unionization” among employees.
This statement is very disappointing. You try to portray the union as some outside force separate from the foundation workers, when the union is made up of WMF workers, and already has the support of a supermajority of eligible workers. You are not a corporation. Your responsibility is to the community, not to shareholders. 1something (talk) 01:41, 28 July 2026 (UTC)Reply
Of course the foundation is a corporation. They are a 501(c)(3). Note that our article on the subject starts with A 501(c)(3) organization is a United States corporation. Their corporate structure and goals may be different from a for-profit publicly traded company, but they're still a corporation. As for "Your responsibility is to the community", that's not true either. The community and the foundation are both responsible to the project. Neither is subservient to the other. Both are equal partners working to advance a common cause. We really need to get away from the us-vs-them mentality. RoySmith(talk)12:02, 28 July 2026 (UTC)Reply
I think the us vs them mentality is a reaction to an already adversial relationship. Consistently taking actions that anger the community is not a neutral stance and that's not the way to treat people you consider to be partners in your shared endeavour. Clovermoss🍀(talk)12:08, 28 July 2026 (UTC)Reply
WP:AGF applies. Lots of decisions get made every day. You're not going to agree with every one of them (I certainly don't). But I can say the same about Arbcom. People like me and you have the advantage that we're not legally obligated to uphold our fiduciary responsibility to manage a lot of money, or to comply with labor laws in multiple countries, or sign our names to corporate filings, so it's easy to take pot-shots at the people who do. If we screw up, the worst that can happen is we get trouted or banned. If WMF folks (at least at a high-enough level) screw up badly enough, they can go to jail. RoySmith(talk)13:50, 28 July 2026 (UTC)Reply
Fidicuary responsibility does not require people to take these actions. I'm not taking potshots at people, I'm telling them to do the bare minimum. Clovermoss🍀(talk)14:04, 28 July 2026 (UTC)Reply
WMF has twice decided to take up the slowest option to delay having to deal with a union, they could have voluntarily recognized in the UK and in the US, and chose not to. WMF cannot claim to support labor rights and take actions that we'd expect Amazon to do. WMF is choosing this path, their relationship with a union does not have to be antagonistic but yet again as we've seen in the past few years, WMF acts like it's their way or the highway. Abzeronow (talk) 14:27, 28 July 2026 (UTC)Reply
There is nothing about voluntary recognition option that could make anyone go to jail. See Labor unions in the United States#Labor unions in the 21st century. Presenting it as such is exactly why the disingenuous comments are being made by the executives. They refused to agree to a union because fundamentally they are anti-Union. Even if the CEO says that she was a union member for 13 years. stjn17:31, 28 July 2026 (UTC)Reply
That's a rather ... unique ... interpretation of the (true) statement "If we screw up, the worst that can happen is we get trouted or banned. If WMF folks (at least at a high-enough level) screw up badly enough, they can go to jail." --Guy Macon (talk) 17:48, 28 July 2026 (UTC)Reply
In addition to all the good arguments given above, I’m concerned and a bit disappointed by the way this has been communicated: just few days after Wikimania, where the world convened into Paris to discuss and share about the Wikimedia project*s* (and not only the English-speaking ones), and where some talks were had about WWU and its recognition, with a lot of attendees showing WWU badges and pins, showing that the international communities do care about this matter… and yet it seems that the only official notice has been given here… Let’s say that I’ll try to continue to assume good faith also about the messaging, but I also hope that the next developments would be shared with a large part of the Wikimedia communities! Sukkoria (talk) 21:40, 29 July 2026 (UTC)Reply
+1. En-wp is not the only Wikimedia project; WMF should communicate this information on other projects where WWU has been discussed as well. — Jules*talk08:56, 30 July 2026 (UTC)Reply
Latest comment: 1 day ago24 comments13 people in discussion
Pardon the new section; I wasn't sure where to put this, and it's been weighing on me.
I'm going to make an argument because I haven't seen anyone make it.
Let me first say that, given a supermajority of eligible US workers have supported unionization, I hope they're successful and agree it should be recognized. I've had union jobs myself and have heard enough stories from then-current or former staff over the years to buy that there are issues with management that a union would help to resolve. But we should be angry, or sad, that it got to this point, because what I'm not persuaded of is that a WMF union is definitely good for Wikipedia.
There's no great data about the relationship between unionization and the quality of outputs. Some studies with decidedly mixed results and a ton of anecdata. My sense is that most of the time, stable unions don't have much of an effect on outputs and in some cases may help. Strikes, however, are obviously very bad for outputs, as are pronounced episodes of labor unrest/conflict (another argument for voluntary recognition IMO). Elsewhere, that might mean more defective tires or mislabeled cans that harm consumers and eventually harm profits, which harms the executives. Here, executives might lose a labor conflict in the end, too, but Wikipedia is the product in the balance and the stakeholders are readers and editors. It's for these reasons that I'm ambivalent about a union in an organization like this one, because the source of the union's power is less its ability to hurt executives or shareholders and more its ability to hurt a public good.
A union often means more rules for who can do what, with unclear implications for boundaries between staff and community as concerns our technical infrastructure. What work protections might a hypothetical future bargaining unit put in place that restricts what volunteers can have access to? Is it hard to imagine some future grievance about volunteers working on gadgets that were the focus of laid-off staff members? Is it hard to imagine a worker strike that renders a range of volunteer maintenance tasks effectively struck, too?
Having a union means adding yet another stakeholder to decision-making processes. The union's ambition is to have a say not just in wages, hours, and working conditions, but also in strategy/annual planning, and indeed I've heard supporters mark that as a major reason for their support. Putting aside that formal say in strategy feels like an unrealistic ask, why would we assume that a union run by some future unknown staff would be acting in the way we want (sufficient for a large portion of this community to pledge its own labor to an organization that has no reciprocal obligation)? For those of us whose Wikipedia priorities are independent of the WMF, is there nobody else who finds this scary?
To read these threads about the union (to the exclusion of many, many other past threads on this board) one might get the impression the community believes every WMF staff member is a Brooke Vibber or a Community Tech team member, with long connections to the community and in tune with our needs/interests. I broadly have a favorable view of WMF staff, who are mostly good, mission-driven people, and agree with a likely response to this: that there have been several times when staff members have spoken up when a bad idea was being implemented (Superprotect, Knowledge Engine, etc.). I also would concede that my work protections hypothetical above is unrealistic among current staff because they get the mission. But who's to say who future staff will be? "In tune with the editing community" is already not descriptive of most staff, well-intentioned though they may be. We have also heard from many staff that the enwiki community in particular is rather more like a thorn in their side (again, see most past threads in the archives of this board for examples). Why would we assume that, in a decision-making conflict, the community-minded staff would win the day over those who aren't, or that the union would always take the community's side? Why would we assume that the currently most vocal proponents among staff, who indeed are aligned with us, are representative of who will be running the union 10-20 years down the line? And what can be done to codify these obligations, if they're so baked into broad community support? (I don't know if that would be a smart thing, to be clear, as it feels like a desire to govern indirectly through on-wiki !votes, but if that's the promise people are banking on, how will it be enforced?)
If others have these worries, I haven't heard them expressed. It would make me feel better (and possibly others) to see that others are thinking about this, too, and that the outpouring of support for a big, blunt instrument isn't just a cascade in response to a recent instance of injustice. I feel like it would be helpful to articulate that it's fundamentally a calculated gamble -- that a union might in fact be a bad thing for Wikipedia in the long run, but that the median prediction for a WMF future with a union looks better than a median prediction for a WMF future in which management continues as-is. And that for those who aren't scared at all, it's less a belief that unions are always good for everybody and more that WMF management has, over the years, convinced so many people of its disconnect from this community that hundreds are desperate enough to cling onto any prospective entity that holds more power with the WMF than we've ever had. That feels more sad than exciting to me, regardless of the outcome, and if it's accurate it's worth sitting with that sadness and acknowledging the gamble. But here we are. — Rhododendritestalk \\ 17:46, 28 July 2026 (UTC)Reply
Thanks for the thoughtful post. To my way of thinking, the main difference is this: If the employees at Goodyear stop working no tires get made. If the employees at America Red Cross stop working a lot of people won't get the help they need. If the employees at the Wikimedia foundation stop working, the effect on Wikipedia will be pretty much zero. There are a few exceptions -- core tasks that must be done. Examples include running backups on the servers, paying bills, having lawyers respond if someone sues, and even checking that the fire extinguishers at HQ follow the legally mandated service schedule. I would be very surprised if any strike didn't have exceptions for those few vital functions, and even if they didn't, the WMF would be perfectly justified to hire temporary replacement workers.
(@Guy) the effect on Wikipedia will be pretty much zero Two responses to this: first, if they matter so little to Wikipedia, what difference does it make to Wikipedia if they unionize? Solely operating in the domain of "nice to haves" rather than "need to haves" doesn't seem like it would foment such parallel collective action on-wiki. But second (and much more importantly), I'd just strongly disagree with that premise. To be clear, yes, they need us more than we need them -- not disagreeing with that. The product that brings in donations is primarily the product of volunteer labor. But staff maintain toolforge and cloud services, they review community-requested config changes and patches, they respond to outages and mitigate DDoSes, they respond to trust & safety threats of harm and child safety, they handle non-emergency T&S business like harassment, they handle DMCA takedowns and subpoenas on fixed deadlines, they have to follow a litigation calendar, they disburse grant funds to affiliates and event organizers who have their own non-unionized staff not to mention bills to pay, the stream of data dumps and database updates and analytics get outdated, we fall behind in security patches, there's all the boring hardware maintenance... and all that is putting aside addressing any issues with the fundraising pipeline, which I suspect would not be persuasive. :) — Rhododendritestalk \\ 18:33, 28 July 2026 (UTC)Reply
This question pivots between the disruptiveness of a potential strike or not. Even in the cases of schools or hospitals, where a strike would be incredibly disruptive AND harmful to communities, they still happen, with responsible caution. For example, nurses will still show up for life-saving emergency shifts. In the case of schools, it is less dramatic, but still detrimental for youth education. A framework to bridge this gap is the bargaining for the common-good, which is alluded to in Wiki Worker's United listed priorities. ~ In solidarity 🦝 Shushugah (talk) 18:23, 28 July 2026 (UTC)Reply
Thanks for the link. And yes, that's relevant to what I'm talking about. I'll be curious to see what kind of ~"emergency provisions" WWU would codify that ensure Wikipedia and its non-executive stakeholders aren't harmed during labor conflicts. — Rhododendritestalk \\ 18:33, 28 July 2026 (UTC)Reply
I don't think that difference gets to the heart of the concern. Unions are responsible to their members, so they may strike over issues that aren't supported by the community. Ensuring the servers are running safely (if they remain operational) is an essential task for safety, but making Wikimedia sites accessible to the public is not. I agree that employees have the right to unionize and a union ensures employees get a seat at the table. But I also agree that it doesn't necessarily have a beneficial effect on an organization's suppliers – in the case of Wikimedia, for example, those who supply content. isaacl (talk) 18:29, 28 July 2026 (UTC)Reply
Sorry, but I'm not particularly impressed by the argument of 'if we give people the right to negotiate as a group, they might side against us and that's bad'. Don't get me wrong - I've seen unions not work. I've probably seen them work a lot worse than 99% of people in the discussion. I can't talk about my experience with a particular union without getting horrible anxiety. Still think people should have the right not to get dicked around by their bosses. Even if it personally benefits me or makes my life more convinient. GreenLipstickLesbian💌🧸18:23, 28 July 2026 (UTC)Reply
Good point. This relate to the question above "if they matter so little to Wikipedia, what difference does it make to Wikipedia if they unionize?" I don't support unionization because it will be good for Wikipedia. I support it because supporting it is the right thing to do. --Guy Macon (talk) 18:41, 28 July 2026 (UTC)Reply
Thanks. It feels useful (or at least clarifying) to have articulated that there are several motivating factors here. It has felt like the biggest factor was "it's in our interest" and I'm just not at all convinced. It's heartening to see "it's just the right thing to do, even if it's against our interests". Good vibes there. — Rhododendritestalk \\ 18:47, 28 July 2026 (UTC)Reply
@Rhododendrites, I do strongly believe that it is in our interests that WMF employees are unionized, but I'm confused by how you've conflated the idea of "this is generally in our best interests" with "the union will agree with the community and act in line with what the community wants". I can believe that it's in my kid's best interest that their teachers are unionized without concluding something like "because my kid's teachers are unionized, they will give my kid an A+". In solidarity, asilvering (talk) 02:32, 29 July 2026 (UTC)Reply
The conflation isn't mine? "Generally in our best interests" is the gamble that may well pan out, but that (IMO) is worth acknowledging as a gamble. The union advocating for and/or acting in the interests of the community is a more specific claim that's part of the union's own pitch (e.g. "empowering our union to advocate for you") and something echoed by many supporters as part of the reason for their support. To me, the conflation appears in a lot of the discourse around the union, and in a way I'm trying to challenge the taken-for-grantedness of the latter to better consider the former. As for the analogy, I'll put aside the A+ bit at the end, which is an illegitimate favor that isn't in anyone's interest and doesn't correspond to anything here, and can't really be adapted into something that fits because there's no additional class of volunteer stakeholders who do most of the educating of kids. I wrote out a list of differences (the scope of people affected by labor action, the agency of affected people, the kinds of accountability, the driving motivations for unionization, the difference in the role of the members in the final output, the absence of any major volunteer group of stakeholders, etc.), but replaced them with this vague parenthetical because I don't really want to go on a tangent. At the end of my message above, I was trying to say that what I'm looking for is from the folks who aren't in the "unions are just good" camp -- those who find the prospect of a union a gamble that could turn out badly but is better than the alternative. If we're getting into analogies with very different kinds of unions, I'm guessing we're talking in the realm of broad support for unions in general? I respect that position -- I'm just trying to see how much of the support isn't that, I guess. — Rhododendritestalk \\ 03:52, 29 July 2026 (UTC)Reply
I mean, I am in the "unions are just good" camp, as opposed to the camp of "unions are good when they benefit me personally." I don't care about the community wishlist in the slightest, and focusing on it feels like a bit of a trap because it presents an easy way out of just prioritizing the wishlist and keeping work conditions otherwise exactly the same. Gnomingstuff (talk) 13:06, 29 July 2026 (UTC)Reply
I think it's odd (to say the least) to use someone who joined the WWU in part because of management disparaging the community to claim that WWU members will probably have that contempt. (The latter is not unreasonable as a reading of that plank, though I'll note that interact directly with movement communities almost certainly means, in large part, dealing with the stuff that gets put under Oversight) Sesquilinear (talk) 20:20, 28 July 2026 (UTC)Reply
I think mental health support being available for roles that deal with incredibly large online communities is actually a pretty reasonable ask, and doesn't necessarily say anything bad about us on the whole. Interacting with people on that scale can be exhausting and distressing, especially on online forums where dogpiling can happen. --Grnrchst (talk) 21:35, 28 July 2026 (UTC)Reply
I'm one of the people who had these worries but hadn't really expressed them. Truth is, I don't have a ton of sympathy for WMF workers wanting to unionize, particularly given my impression that most if not all of these workers are being paid over $100k per year. (I chuckled when I read that wikifired's two months' severance was $25k, meaning they were paid $150k/yr.)
Anyone making $100k+/year is in the top 10-20% in the US, top 5-10% in Europe, top 2% globally. WMF employees are, objectively, among the richest people in the world. Plus they get benefits -- fully paid health care, life insurance, retirement plan, etc. Plus they work from home. So a bunch of the richest people with full benefits working from home want to unionize so they can get a little bit richer? Not exactly what I think of as "oppressed workers."
And what are their gripes? Are their paychecks bouncing? Are they forced to work double shifts? Is there rampant sexual harassment? No, AFAIK, none of these things. Their gripes are that a department was restructured and some of them were offered reassignment to other high-paying, comfortable jobs. The other gripe is that their bosses are jerks. Ohh nooooes, how will they ever manage?
So I have little patience for any comparisons between the richest people in the world with comfortable work at home jobs, and actual oppressed workers, like people who work in mines or are forced to live in company towns. That's why I think the criticism of "cosplaying a labor movement" is, while uncharitable, not far off the mark.
Still, what's the alternative? Say no to unionization? Idk about anyone else, but I wouldn't be willing to tell other people I'm against their unionization, even if the unionized workers are among the richest, most comfortable workers in the world. Let anyone who wants to join a union, join a union. I certainly don't think the WMF should take any kind of anti-union stance, it'd be against the values of the mission and the project. And anyway, it's really their business: the business of the employees. Not for me to say "no," and I don't support the WMF saying "no" either.
So while I might roll my eyes at people claiming "solidarity" with some of the richest, most comfortable people in the world who are complaining about crappy bosses, I still support the WMF voluntarily recognizing a union that a supermajority of WMF employees want to join. Levivich (talk) 18:42, 28 July 2026 (UTC)Reply
Gonna be the bitch here and point out that acting like The other gripe is that their bosses are jerks. and sexual harassment are completely unconnected is... well, it shows a naivety I am jealous of. The WWU listed one of their goals as better mental health support, which, despite years of sitcom -esuq Ohh nooooes, how will they ever manage? jokes to the contrary, is important I'm also not sure that 'well, they make a lot of money, so who cares about Occupational stress or workplace bullying, at least they aren't slaves or not being paid' is a winning argument, but, well, I'll let an editor who you respect have that conversation with you. GreenLipstickLesbian💌🧸19:04, 28 July 2026 (UTC)Reply
@Levivich, they are not unionizing so they can get a little bit richer. Certainly, for many unionized workers, increasing pay or benefits is a significant priority. But the immediate issues the WWU is more concerned with are things like "I want to be able to voice concerns at work without being afraid I will be fired for doing so" and "I do not want my benefits to be changed without notice". These are pretty basic things that any worker should be able to expect, and it is shameful that WMF employees cannot presently do so. In solidarity, asilvering (talk) 02:22, 29 July 2026 (UTC)Reply
"the rich" == people who are in the top 2% globally, top 5-20% in their country depending on country. "Comfortable" == working from home with benefits. And pay and benefits equity that makes sense to the staff is a plank on the WWU platform, and that's corpo-speak (just like the WMF) for "more money and better benefits for some if not all". So, yeah, the rich and comfortable are organizing so they can be a little bit richer and little bit more comfortable. Put more charitably, so that their slightly-less-rich and slightly-less-comfortable colleagues can also be equally rich and comfortable. That framing of course doesn't sound quite as noble as "labor rights," but that's what "pay and benefits equity that makes sense to the staff" means: more pay and benefits.
The way you phrased the first example is not really on point, but I'm not going to waste your time arguing semantics, I think we'd both agree that whistleblower protection is important, and the WMF should be not only complying with the minimums required by law but serving as a leading example in the world of how large organizations should work. But whistleblowing is more than just "voicing concerns at work," it's more than just griping about a bad boss. I haven't seen the union anywhere talk about a lack of whistleblower protection, or even an example of a WMF whistleblower who wasn't protected. (Not a disgruntled employee, but a whistleblower.)
The second example, I'm not familiar with... I didn't see that on the union's website or in the blog post, maybe I missed it... and there are laws against that sort of thing, at least with respect to important benefits like health insurance (fringe benefits, maybe not, but who cares about those). Can you point me to where WMF workers are complaining about benefits being changed without notice? Levivich (talk) 02:49, 29 July 2026 (UTC)Reply
Labour rights are human rights. We should support WMF workers' unionisation rights as a human right.Respecting others' human rights does not always bring personal benefit. There's no guarantee that in a decade or so (or even sooner), WWU will not have flaws as serious as (but complementary to) those of the WMF leadership. Institutional safeguards and structures that tend to reduce those flaws are needed in both WMF and WWU. Anyone talking constructively with WWU organisers right now might have a chance of convincing them to make the right institutional choices now that reduce the chance of WWU becoming evil by 2036. Boud (talk) 20:35, 28 July 2026 (UTC)Reply
Are WMF employees laid off when projects are shuttered? For example, Knowledge Engine or potentially Abstract Wikipedia. If so, it could be possible that if the community attempts to get a project closed, the WWU would be opposed in an effort to keep people's jobs (which is understandable, but could also be a waste resources). ARandomName123 (talk)Ping me!03:49, 30 July 2026 (UTC)Reply
I assume that part of the idea of having a union is to bring about a workplace that offers more than the two options of "do this questionable project and keep our jobs" vs. "shutter this questionable project and lose our jobs." Gnomingstuff (talk) 05:37, 30 July 2026 (UTC)Reply
Wikimedia Foundation banner fundraising campaign in Malaysia
Latest comment: 2 days ago10 comments6 people in discussion
Dear all,
I would like to take the opportunity to inform you all about the upcoming Wikimedia Foundation banner fundraising campaign in Malaysia.
We will run banners for non-logged in users in Malaysia on English Wikipedia itself. The banners will run from the 2nd to the 30th of September 2026.
A brief note on the timing. We are moving the fundraising campaign in our planning from June to September which means that logged out readers will see fundraising banners twice this year. Going forward the campaign will again only run once each year.
Prior to this, we are planning to run some tests, so you might see banners for 3-5 hours a couple of times before the campaign starts. This activity will ensure that our technical infrastructure works.
Generally, before and during the campaign, you can contact us:
@JBrungs (WMF): Isn't that rather risky of a backlash in the current context? Malaysiahas manybigtrade unions. In fact, per the current lead, National Union of Plantation Workers is one of the largest [trade unions] in Asia. A month from now, if WMF has not yet voluntarily recognised WWU UK and WWU US, then readers in Malaysia risk getting angry that they're being asked to donate to a foundation that doesunion busting. The damage to WMF's reputation risks extending to the wider Wikimedia community itself, since much of the public doesn't see the difference between "Wikipedia" (the Wikimedia community) and "Wikipedia" (WMF that runs the infrastructure). You're playing with a tinderbox. WMF's union-busting will not remain secret. Ordinary Wikimedians do not want to discourage donations: we know that WMF needs to pay salaries and server costs. But we will not hide the evidence. Boud (talk) 08:31, 29 July 2026 (UTC)Reply
As a sign of good faith by WMF, I would recommendyou might want to considerdelaying the banner campaign until WMF has recognised both WWU UK and WWU US. "We hate trade unions, please give us some money" is not a great way of convincing people to donate. (@JBrungs (WMF): It would also be a good sign of respect for the community to answer the questions at the discussion pointed to below by Chipmunkdavis.) Boud (talk) 12:03, 29 July 2026 (UTC) (edited Boud (talk) 13:55, 29 July 2026 (UTC))Reply
In the spirit of WWU's social media post stating that it is not calling for donations to be delayed, I've weakened my recommendation to a "maybe consider". Even if WWU is not calling for a delay, the risk of backlash remains. I've seen plenty of social media (both decentralised and centralised) posts by people declaring that they would stop donating because of the union-busting. WWU cannot force those people to keep donating, nor convince English speakers in Malaysia that they should donate despite the union-busting (unless the banner includes a reference to WWU's social media statement that it is not calling for donations to stop, which would risk being interpreted like an out-of-the-blue statement "I am not a thief", which makes people wonder why it was stated at all). Boud (talk) 13:55, 29 July 2026 (UTC)Reply
Regardless of what my views on the WMF union situation is, I do think they should still run the fundraiser and get funds. Not getting funds is a ecological risk for our movement. Sohom (talk) 14:45, 29 July 2026 (UTC)Reply
Note that server costs are not a major expense item for the WMF. Internet hosting costs are less than $3.5 million p.a., while the Foundation has close to $500 million in assets (if you include the Endowment).
On the plus side, the banners look so much better these days than they used to. The emails however I still find borderline wrong, because they arguably include subtle suggestions that free access to Wikipedia might be under threat.
"My hope is that you'll think about how useful Wikipedia has been to you this past year, and how valuable it is to have access to a source of knowledge that doesn't charge you anything, doesn't sell you anything, and doesn't interrupt you with ads." (E-Mail 1)
"We owe it to them, in a world that is always changing, to keep Wikipedia free for everyone. Like it always was and always should be." (E-Mail 2)
"This might be my last chance to ask, so I want to make sure this email reaches everyone who might give. Right now, we're at a critical stage of our fundraiser in Brazil." (E-Mail 3)
"We need our community of donors to help us reach our goal, and time will soon run out in this fundraiser. ... We want to make sure we all have equal access to high-quality information – something that is really hard to find online these days. We still have a lot of work to do to make knowledge free for everyone, and you can help us achieve it. With your donations, The Wikimedia Foundation keeps Wikipedia operational ..." (E-Mail 4)
@Jayen466We need our community of donors to help us reach our goal, and time will soon run out in this fundraiser. (what is actually being said) and need our community of donors to help us reach our goal, and time will soon run out (what you tried to highlight in your framing) are two very different sentences espousing very different meanings. For what it's worth, we spend approx >200 million a year, most of which is spent on the movement, 500 million buys us a year and a few in change (unless you want to only keep the lights on, in which case 3.5 million is still woefully insufficient since it actually discounts the salary of the people maintaining the hardware/software). Sohom (talk) 14:30, 29 July 2026 (UTC)Reply
Of the $190 million in 2024/2025 operating expenses, $114 million were spent on salaries and benefits for just under 700 employees (averaging about $165,000 per head), and another $16 million on professional services. [49] This may be money spent "on the movement", but indirectly so. Regardless, keeping Wikipedia operational can be done for far less. It's not what the money is needed for, or primarily spent on. To my mind the emails highlight that aspect a little too much. AndreasJN46614:55, 29 July 2026 (UTC)Reply
To be fair, not answering is standard procedure. Try asking them to tell us how much just one Wikimania cost and what was bought with that money. Try to get any answer including "we have decided not to tell you that". You will get various responses from other volunteers who don't have the information and don't speak for the WMF, but the WMF itself will stonewall you. --Guy Macon (talk) 13:21, 29 July 2026 (UTC)Reply
Petition to the WMF board of trustees to voluntarily recognise the US staff union
Latest comment: 17 hours ago7 comments5 people in discussion
Hi all
As some of you may have seen WMF announced the day after Wikimania they have chosen not to recognise the WMF staff unionisation efforts in the US and UK voluntarily. I'm sharing this as someone who has never worked for the Foundation, but if I did I would not like to be treated like this.
I hope you will consider signing this petition to ask the board to instruct the CEO to voluntarily recognise the US union and begin bargaining in good faith.
Where was it "discredited" that the WMF has the right to ask for a secret ballot vote? Please quote the exact part of applicable labor law that says that WMF management is required to submit to third-party arbitration. --Guy Macon (talk) 00:42, 31 July 2026 (UTC)Reply
That's an unnecessarily combative way to say "the WMF has the right to secret ballot and is not required to arbitrate." Anyway, who cares about rights and requirements? The WMF could, and should, choose to voluntarily recognize or arbitrate. (That's why the thing everyone is calling for is called "voluntary" recognition.) Levivich (talk) 01:15, 31 July 2026 (UTC)Reply
I don't believe that's what Stjn is saying. I believe they mean that they used discredited anti-union talking points in their response. For example, calling the vote a commitment to ensuring that every eligible staff member can make their own decision freely, privately, and democratically flies in the face of how staff did make a decision, freely, privately, and democratically, when ~70% of eligible staff signed a union card. Best, HouseBlaster (talk • he/they)01:26, 31 July 2026 (UTC)Reply
OK, I can see that. I also asked the WMF to voluntarily recognize the union, but I expect WMF management to do what management does, which is to make anyone trying to start a union follow every step that they are legally required to do.
I predict that, once there is a union in place, a time will come when the union and management disagree on something. Will I see the same reaction when the union fails to voluntarily give management everything they want without a fight? When some manager who really doesn't like unions accuses the pro-union side of "using discredited anti-management talking points"? I suspect that the reaction would be quite different. --Guy Macon (talk) 01:52, 31 July 2026 (UTC)Reply
They are not legally required to do a secret ballot election under Trump-controlled NLRB. They would only have to do it because the WMF rejected a good faith settlement with the union for reasons ranging from flimsy to downright improbable. stjn09:09, 31 July 2026 (UTC)Reply
"It is completely legal for the WMF to insist on a paper election conducted by the underfunded Trump-appointed NLRB. I find it ironic that half of the BoT trustees are unelected in the first place. For the remaining trustees, online voting was considered secure enough to oversee the foundation's $200 million annual budget, yet it is deemed too risky for the 200 US staff."
So when everyone is talking about "signing a union card" are they talking about doing that online?
If so, I would be very much interested in the security of the method the WMF uses to manage the budget over the internet and the security of the method the union uses to gather signatures over the internet. --Guy Macon (talk) 02:17, 31 July 2026 (UTC)Reply
So who, outside of the union or WMF management, has confirmed the claim that a supermajority of eligible workes have "signed a union card" (falsely so called)? If, as I suspect, management is doing what management does and trying to stop some union from reducing their power, it seems reasonable to have a secret vote that neither side can cheat on. --Guy Macon (talk) 03:34, 31 July 2026 (UTC)Reply
The WWU page that links to the form says By law, your answers on it are confidential between you, the Wiki Workers United union, the CWA and the National Labor Relations Board (NLRB). Your answers are not allowed to be shared with WMF management.ARandomName123 (talk)Ping me!04:05, 31 July 2026 (UTC)Reply
No one has confirmed it yet, because no one has had the chance to. Were the WMF to do the right thing here and go for voluntary recognition, the next step would be a process in which a neutral third-party arbitrator reviews the cards that have been signed and responds to any concerns the WMF may have about potentially invalid cards; in the very unlikely event that the arbitrator strikes enough cards to shrink the share to below 50%, voluntary recognition would not happen. Voluntary recognition does not mean the WMF forfeiting its right and responsibility to do due diligence before taking a legally binding action. It means agreeing to recognize the will of the majority assuming it actually is the will of the majority—the same thing the WMF is claiming it needs a drawn-out, government-run secret-ballot election to accomplish. -- Tamzin[cetacean needed] (they|xe|🤷) 10:37, 31 July 2026 (UTC)Reply
Again, I agree that management should go the voluntary recognition route. I have signed the petition asking them to do that. But they clearly don't have to. And, given their past performance in other areas, I fully expected them to do everything they can to maximize their own power and wealth. Nor do I have any expectation that any petition or discussion at the Village Pump will change the situation.
So here we have an actual, genuine, bona fide example of assuming bad faith. There's absolutely no reason to believe that WWU is "cheating" on the ballot. Gnomingstuff (talk) 15:04, 31 July 2026 (UTC)Reply
Show me a diff where I or anyone else implied that anyone is cheating. Assuming good faith does not require letting the WMF count the ballots. Assuming good faith does not require letting the union count the ballots. Is meta:SecurePoll an AGF violation? --Guy Macon (talk) 17:11, 31 July 2026 (UTC)Reply
Here: So who, outside of the union or WMF management, has confirmed the claim that a supermajority of eligible workes have "signed a union card" [...] it seems reasonable to have a secret vote that neither side can cheat on.
Unless there is actual evidence suggesting that the union is falsely claiming that a supermajority of eligible workers signed the card, there is no reason to suggest it might be false, as you have above. Gnomingstuff (talk) 18:18, 31 July 2026 (UTC)Reply
.png)
15:15, 28 July 2026 (UTC)
Let’s say that I’ll try to continue to assume good faith also about the messaging, but I also hope that the next developments would be shared with a large part of the Wikimedia communities! Sukkoria (talk) 21:40, 29 July 2026 (UTC)