







This post https://x.com/CJHandmer/status/2122609626588061816 has been making the rounds on X. The post claims to use Claude Opus 5.5 to identify 872 major errors in published astrophysics papers, described at https://science.caseyhandmer.com/physics/
Casey Handmer on Twitter / X
In my ongoing quest to find the limits of Opus 5.5, I directed it to audit my old scientific field, theoretical astrophysics. We indexed about 150,000 papers, audited 50,000 for issues, and did progressively more thorough checks on thousands more. We found nearly 1000 major… pic.twitter.com/VarvnTXohF— Casey Handmer (@CJHandmer) November 17, 2026
Perspectives on Scientific Error 2027 · PSE9
A field of physics, audited by machine
158,059 papers, 554,670 citations, 872 verified errors followed through the literature. Half of the copied errors were still being copied 7 years later.
Ferhat Tura on Twitter / X
I have recently detected minor mistakes in the codes of a paper published in a high-quality journal and contacted the lead author first!, who acknowledged the mistake and contacted the journal. I thanked the author for doing open science and now we are collaborating on a paper💪— Ferhat Tura (@FerhatTura19) May 15, 2024
Errors in the research of Emmanuel Saez and Gabriel Zucman - Marginal REVOLUTION
AI issues its judgment: 138 papers audited 2,035 problems found 147 serious enough to change a headline number 69% of papers have at least one 97% of major and moderate errors with a clear direction overstate the paper’s claim And on the method: Two independent audits of each paper. One by Claude Opus 5.5, one […]
Common Errors in College Math
This page describes the errors that I have seen most frequently in undergraduate mathematics, the likely causes of those errors, and their remedies.
More than half of recent astronomy papers are written with language-model assistance
Language models leave a distinctive vocabulary in the prose they help write, and we measure how much of the astronomy literature now carries it. From the full text of 207,111 astro-ph papers spanning 2015 to mid-2026, we count those words in each paper and model the counts, in proportion to paper length, as a mixture of assisted and unassisted writing in a hierarchical Bayesian model. Papers from before 2020 calibrate the unassisted rate, and the 392 papers that disclose model use calibrate the assisted one. Our answer depends on how often these words would appear today if nobody used a model, a rate that must be modeled rather than observed, so we extend it past 2020 under three assumptions and report all three. For 2025 that gives $54^{+8}_{-8}\,(\mathrm{stat},\,95\%)\,^{+26}_{-0}\,(\mathrm{sys,\ background})$% of papers, the second error being the spread across the three. The estimate stays at or above 36% when we vary that choice, the calibration, and the requirement that adoption only rises. A word list built from the astro-ph corpus, keeping only words that rose across every subfield, leaves 2025 in the same range. Assisted writing is also getting harder to see, since authors adapt to the words that reveal it and the marker excess more than halves between 2023 and 2026. Our model allows for that fading, so it can separate a fainter trace from reduced use. More than half of recent astro-ph papers therefore carry a language-model trace, while only 0.81% of 2025 papers disclose it, one declaration for every $\sim$66 papers with a trace.

Home
ERROR is a bug bounty program for science to systematically detect and report errors in academic publications

Opus 4.7 isn't dumb, it's just lazy — Shimin Zhang
Some follow up experiments with Claude Opus 4.7 based on Simon Willison's Pelican Benchmark Shocker.

Michael C. Frank on Twitter / X
New blog post: I used Claude to audit the reproducibility of every journal paper I've been an author on in the last five years. Our commitment to openness meant that the code and data were available, but there were some significant issues. 🧵https://t.co/O5pkxUM4Jx— Michael C. Frank (@mcxfrank) November 10, 2026
Michael C. Frank on Twitter / X
New blog post: I used Claude to audit the reproducibility of every journal paper I've been an author on in the last five years. Our commitment to openness meant that the code and data were available, but there were some significant issues. 🧵https://t.co/O5pkxUM4Jx— Michael C. Frank (@mcxfrank) November 10, 2026
Jianyang Gao on Twitter / X
The TurboQuant paper (ICLR 2026) contains serious issues in how it describes RaBitQ, including incorrect technical claims and misleading theory/experiment comparisons.We flagged these issues to the authors before submission. They acknowledged them, but chose not to fix them.… https://t.co/wzMsDcWOtF— Jianyang Gao (@gaoj0017) March 27, 2026
Brandon Yates on Twitter / X
I wrote a cosmology paper that's under peer review.I can't post it to arXiv because I don't have a university affiliation.If any of you happen to have published in astro-ph. CO, gr-qc, or hep-ph and can endorse me, I'd owe you one.#AcademicTwitter #PhysicsTwitter #Cosmology…— Brandon Yates (@bmichaelyates) March 13, 2026
When should a paper be corrected? Last December, @elisabethbik.bsky.social and other sleuths began flagging papers by a prominent bioengineer, Ali Khademhosseini. They found over 80 with image issues. Khademhosseini and his colleagues have issued over 40 corrections, but avoided retractions.
Science sleuths raise concerns about scores of bioengineering papers
www.nature.com"I" may have just found a fundamental flaw in a "dense, high-prestige paper". Problem is, I don't have a clue what I did and how to verify any of this. I asked Gemini to perform a review of an interesting paper and it was skeptical and proposed a data analysis that should be performed.
a south park poster with a boy scout and the words what are you guys doing in there
media.tenor.com