Skip to primary content

Statistical Modeling, Causal Inference, and Social Science

Main menu

  • Home
  • Authors
  • Blogs We Read
  • Books
  • Sponsors

Post navigation

The intervention and the checklist: two paradigms for improvement R graph gallery: click on the link below RIGHT AWAY and be awed

This one’s for Zacky

Posted on February 19, 2009 1:15 PM by Andrew

Los Angeles Tar Yields Mammoth’s Skeleton

This entry was posted in Miscellaneous Science by Andrew. Bookmark the permalink.
  • Art
  • Artificial Intelligence
  • Bayesian Statistics
  • Causal Inference
  • Decision Analysis
  • Economics
  • Jobs
  • Literature
  • Miscellaneous Science
  • Miscellaneous Statistics
  • Multilevel Modeling
  • Obituaries
  • Papers
  • Political Science
  • Public Health
  • Sociology
  • Sports
  • Stan
  • Statistical Computing
  • Statistical Graphics
  • Teaching
  • Zombies
  1. Joshua on If the stock market crash had happened three weeks ago . . .November 21, 2026 5:53 PM

    oncodoc - Indeed. I think it's a function of the fundamental pattern-finding nature of our cognitive processes. It's why conspiracy…

  2. Anoneuoid on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 5:16 PM

    Generating a document that when translated into prose results in 250,000 pages of content that is mostly drivel and that…

  3. K on If the stock market crash had happened three weeks ago . . .November 21, 2026 5:00 PM

    I miss those excuses. They were far better than "We lost because the election was rigged".

  4. Andrew on If the stock market crash had happened three weeks ago . . .November 21, 2026 4:36 PM

    Oncodoc: Yes, and also when people lose an election, they often want to delegitimize the loss in some way, for…

  5. Dale Lehman on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 4:24 PM

    Anonymous My experience is different than yours. AI has contributed to my human understanding about many things. I can't speak…

  6. oncodoc on If the stock market crash had happened three weeks ago . . .November 21, 2026 4:23 PM

    Here’s the conclusion that I have come to after these postings. Many social, economic, and political outcomes are the results…

  7. Anonymous on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 3:53 PM

    Generating a document that when translated into prose results in 250,000 pages of content that is mostly drivel and that…

  8. Priyantha Wijayatunga on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 21, 2026 2:35 PM

    My Proposed problem: How can we distinguish genuine psychological phenomena from statistical artefacts, measurement errors, misleading associations and spurious correlations…

  9. Andrew on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 1:47 PM

    Dale: The point of Campos's post was that, had this particular story been about Hunter Biden, it would've been promoted…

  10. Dale Lehman on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 12:41 PM

    I'm feeling particularly dense this morning - but since I pondered this when it first appeared, my density has been…

  11. Anon on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 12:39 PM

    I'm curious at what point a belief that LLM technology has a fundamental problem with facts has to be reconsidered.…

  12. Anon on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 12:28 PM

    Yes, that was the point.

  13. Andrew on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 11:04 AM

    Fixed; thanks.

  14. Gregory C. Mayer on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 10:49 AM

    The link to Campos' piece seems to be broken.

  15. Anoneuoid on If the stock market crash had happened three weeks ago . . .November 21, 2026 8:28 AM

    Stock market is back to where it was in April. Not a crash.

  16. Anoneuoid on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 21, 2026 8:17 AM

    Is it difficult for people to generate poorly specified questions of this quality? Like am I underestimating myself in thinking…

  17. David, a Bostonian in Tokyo on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 8:03 AM

    Yes, I just ignored that and went off on a tangent. But since you insist, I'm not convinced. It seems…

  18. AAAnonymous on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 7:54 AM

    A focus on good prose, I suppose, might also fit nicely with declining to review a mostly AI-written manuscript It…

  19. Jeff on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 7:02 AM

    The way I read Anon's comment was not that we'd be using a tool like Pangram to detect human slop,…

  20. Andrew on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 21, 2026 6:52 AM

    Jennifer: No, this post is not intended to mock psychology at all. This person sent me a description of their…

  21. David, a Bostonian in Tokyo on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 4:26 AM

    Yes. It's another Muphry's Law instance. My "writing that has errors in it" was intended to mean "writing that has…

  22. David, a Bostonian in Tokyo on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 4:20 AM

    Ha! I seem to have a typing tic that generates "it's" without thinking. I've corrected it 3 or 4 times…

  23. Jon S on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 21, 2026 3:56 AM

    All you point out in your other comment is that the question is heavily underspecified. That doesn’t mean the question…

  24. Anonymous on Major publisher releases AI-written history book that is full of glaring errorsNovember 21, 2026 3:13 AM

    Andrew's piece recounts factual errors in human writing -- how is this different from "truth and accuracy"?

  25. Jennifer C on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 21, 2026 12:12 AM

    I'm not sure if this post is supposed to be funny! Millenium prize problems are high in the public consciousness…

  26. dsapsis on Major publisher releases AI-written history book that is full of glaring errorsNovember 20, 2026 11:52 PM

    Shouldn't that be "its linguistics is problematic."? Sorry, had to (S&W was drummed into me by a professional editor-as-Mom).

  27. David, a Bostonian in Tokyo on Major publisher releases AI-written history book that is full of glaring errorsNovember 20, 2026 8:14 PM

    Unfortunately, I'd guess that Pangram (an LLM-based program) is bad at detecting human slop. As defined here, human slop is…

  28. Boxo McFoxo on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 5:17 PM

    > I asked my contact in the AI industry and he says this is not true: using synthetic data too…

  29. Anonymous on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 4:35 PM

    Bob: What exactly does the "reasoning/thinking" part entail? Like, what is a high-level overview of how the new models are…

  30. Boxo McFoxo on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 4:34 PM

    I never said it was a myth that foundation models exist. I'm well aware of the pipeline described in those…

  31. Bob Carpenter on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 4:03 PM

    Boxo McFoxo: I'm agreeing with Alex Reinhart here. Fine-tuning a foundation model is not only not a myth, it's literally…

  32. Bob Carpenter on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 3:49 PM

    jcp: The post-training makes it a worse next-token predictor. What it does is turn it from a text completer into…

  33. Bob Carpenter on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 3:41 PM

    Andrew: The fine tuning of the foundation model produces another LLM, so it's still string prediction. What people aren't realizing…

  34. Bob Carpenter on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 3:30 PM

    Millions of documents at 1000 tokens each (i.e., billions of tokens) is 4+ orders of magnitude short of what the…

  35. Bob Carpenter on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 3:27 PM

    This is a real problem for just about any language analysis. The distribution of words changes quite a bit over…

  36. Bob Carpenter on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 3:24 PM

    Classifiers map problem instances to classes. Probabilistic classifiers give you a distribution over classes. What Pangram is doing is different.…

  37. Bob Carpenter on The 2026 Pangram social justice misinformation wars: Collective understanding of binary classification takes a hitNovember 20, 2026 3:16 PM

    I meant "anthropomorphic," because I think folks like Bender are objecting to the anthropomorphism inherent in people saying that machines…

  38. AAAnonymous on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 2:42 PM

    @Phil: Before the quote I posted Kun writes: "The success rate of the European Research Council’s (ERC) Consolidator Grant—the flagship…

  39. Anonymous on Major publisher releases AI-written history book that is full of glaring errorsNovember 20, 2026 1:23 PM

    Related to the last point, I saw something similar on X today where someone was making the point that AI…

  40. Anon on Major publisher releases AI-written history book that is full of glaring errorsNovember 20, 2026 1:21 PM

    Simon Jenkins is a well-known author of popular non-fiction aimed at the middle-classes, including extremely popular guides to churches, stately…

  41. Phil on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 1:17 PM

    “If grants are basically distributed only to a fraction of the qualified and worthy projects, then we need to admit…

  42. Paul R. Pudaite on On Spy FictionNovember 20, 2026 11:09 AM

    Since you've enjoyed Tim Powers, Poker and Spy Fiction, seems likely you'd enjoy Tim's 2000 novel _Declare_. It's my favorite…

  43. Anoneuoid on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 10:53 AM

    Interesting, because as explained below its difficult for me to take that question in good faith. It seems very low-effort…

  44. Andrew on Major publisher releases AI-written history book that is full of glaring errorsNovember 20, 2026 10:27 AM

    Fixed; thanks.

  45. Joshua on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 10:08 AM

    I see Anoneuoid clarified his point as I was drafting my comment. And I agree with his basic point. But…

  46. Joshua on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 10:04 AM

    Raphael - > However, I am not sure we will ever get a really good answer to that question. I…

  47. Anoneuoid on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 9:57 AM

    Well, what is a "psychological impact"? What is "social media", is this blog social media? Why does it only mention…

  48. vfm on Major publisher releases AI-written history book that is full of glaring errorsNovember 20, 2026 9:53 AM

    I think the author is Simon Jenkins, not Simon French

  49. Raphael K on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 9:14 AM

    I do not think many people would argue that social media is harmless. However, the question was not whether it…

  50. John S on Can you help to identify psychology’s most important unanswered questions? Each winning question receives $400.November 20, 2026 8:19 AM

    It's hard to take this comment seriously or in good faith. Yes, lots of people think that (some variant) of…

Proudly powered by WordPress