Edge Rewrite
// HTMLRewriter · presentation

This page was redesigned at the edge.

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.

Country
US
Cloudflare location
CMH
Connection
HTTP/2
Language
Not provided

Ray ID: a2652f51bd5764c2

Jump to content

Talk:Edward S. Herman

Page contents not supported in other languages.
Add topic
From Wikipedia, the free encyclopedia

Finally, a credible mainstream biography of Ed Herman

[edit]

I just saw this, this morning:

https://grokipedia.com/page/Edward_S._Herman

Take note, Wikipedia, on how to write credible and non-libelous biographies.

https://wadefrazier.substack.com/p/finally-a-credible-mainstream-biography Wadefrazier (talk) 19:08, 28 October 2025 (UTC)Reply

Wadefrazier: Is it credible? LLMs (and Grok is no exception) have a very difficult time accurately citing where they get information. This is because they ingest all of the sources in a big pile, break it down into "tokens" (about 4 characters of text) and the using statistical algorithms try to guess (infer) which token is most likely to come next. From this it builds words, sentences, paragraphs. The actual origin of any one sentence is many different sources that were mashed together in a blender. How do you cite that? The LLM doesn't even know itself, nobody does. Take for example from Grok: He resided in the Philadelphia area during his later years, where his health gradually declined amid advanced age. This is a very basic, simple information. It is probably even true. Grok cites this source. The source does not verify. Nowhere in that source does it say that his health "gradually" declined (versus "sudden" decline"). It doesn't even mention anything about his health. If this was on Wikipedia, it would be deleted as original research. Now imagine what happens with more controversial things. The sources frequently do not verify. And they are frequently riddled with original research - do you trust a machine to do original research? I've tried, and it often makes incorrect assumptions, because when faced with uncertainty the algorithms revert to the mean, the most "likely" outcome, even if that outcome is not actually true. So yeah, there are probably good things in the Grok article, and there are bad things, and I wish you luck spending weeks deconstructing that massive wall of LLM produced text, sentence by sentence, claim by claim. -- GreenC 17:31, 3 November 2025 (UTC)Reply
It isn't libelous... Wadefrazier (talk) 22:35, 10 March 2026 (UTC)Reply