Latest Corrections
Wikipedia Mar 18, 2026 at 11:24 PM
en.wikipedia.org/wiki/Fantasy_(psychology)
- This sentence conflates schizophrenia with “fantasy” and overstates biomarker evidence. Standard clinical references list delusions, hallucinations, disorganized thinking/behavior, and negative or cognitive symptoms—not “fantasy”—and reviews say there is still no clinically established biomarker for schizophrenia.
Wikipedia Mar 18, 2026 at 11:22 PM
en.wikipedia.org/wiki/Apathy
- This repeats the debunked version of the Kitty Genovese story. Later archival research found no evidence that 38 witnesses simply watched without acting, and some neighbors did intervene or contact police.
LessWrong Mar 18, 2026 at 10:52 PM
www.lesswrong.com/posts/Mxsy7wYvsCRv5dGrw/tastybench-toward-...
- The linked “ground truth” CSV does not match this description: it contains papers with publication dates outside January–March 2024, and the file has 38 total lines including a header, which means 37 paper rows rather than 38 papers.
LessWrong Mar 18, 2026 at 10:05 PM
www.lesswrong.com/posts/E9xfgJHvs6M55kABD/less-dead
- Current neuroscience does not say preserved ultrastructure definitively preserves long-term memories. The question is still unresolved and lacks professional consensus.
LessWrong Mar 18, 2026 at 09:37 PM
www.lesswrong.com/posts/skKYznZyRtN87tHbB/neuronpedia
- Neuronpedia no longer describes itself this way. The project officially pivoted in 2024 from a crowdsourced neuron-explanation game to a mechanistic-interpretability research platform focused on sparse autoencoders and related tools.
LessWrong Mar 18, 2026 at 08:09 PM
www.lesswrong.com/posts/TyusAoBMjYzGN3eZS/why-i-m-not-a-baye...
- Probability coherence does not require giving the same credence to every proposition implied by a statement. It only requires equal credence for logically equivalent statements; if A implies B, the probability axioms generally require P(A) ≤ P(B), not P(A) = P(B).
LessWrong Mar 18, 2026 at 08:06 PM
www.lesswrong.com/posts/dxffBxGqt2eidxwRR/the-optimal-age-to...
- Major fertility authorities do not say there is a consensus that the best age is 19–26. Official guidance instead says evidence is insufficient to name one optimal age, or places the clinically optimal window more broadly in the 20s to early 30s / before about 35.
LessWrong Mar 18, 2026 at 08:04 PM
www.lesswrong.com/posts/mgjtEHeLgkhZZ3cEx/models-have-some-p...
- The linked replication code uses six seed prompts, not five. It includes an extra climbing prompt, and the repo’s own result summaries report 6 seeds for several models.
X Mar 18, 2026 at 01:37 PM
x.com/GBNT1952/status/2033756101338861880
- The $37 billion figure is not California's spending on homelessness alone. California's Legislative Analyst's Office says about $37 billion since 2019-20 was for "housing- and homelessness-related programs," while the State Auditor cited nearly $24 billion for homelessness and housing over 2018-19 through 2022-23.
Wikipedia Mar 18, 2026 at 12:39 PM
zh.wikipedia.org/wiki/%E4%BA%B2%E5%AD%90%E9%89%B4%E5%AE%9A
- 这里把 S/s 说成了一个独立的“血型系统”，但在现代输血医学命名里，S 和 s 是 MNS 血型系统内的抗原，不是单独的 “Ss 血型系统”。
Substack Mar 18, 2026 at 03:57 AM
tbthealth.substack.com/p/is-modern-day-bread-even-food
- This additive code is wrong: guar gum is INS/E412, not E41.
- This overstates the ruling. The court did not say Subway’s product was not bread in general; it said it did not qualify as “bread” under a specific Irish VAT law definition for zero-rated staple foods.
Substack Mar 18, 2026 at 03:44 AM
tbthealth.substack.com/p/how-to-buy-eggs-in-india
- This is too absolute. White and brown eggs come from many chicken breeds, not just White Leghorns and Rhode Island Reds.
- This is too absolute. European egg-marketing rules include exemptions, so not literally every egg in Europe has to be stamped with a producer code.
Wikipedia Mar 18, 2026 at 03:35 AM
en.wikipedia.org/wiki/Analytic_language#List_of_analytic_lan...
- This is backwards for Modern Persian. Authoritative references describe Persian as having developed strongly analytic structure, with much of the older inflectional system lost.
X Mar 17, 2026 at 09:26 PM
x.com/aakashgupta/status/2033787781403611248
- Official FDA/NTP materials say the study was nominated because cell-phone use was widespread and long-term health data were limited, not because FDA had already concluded harm was expected.
- The NTP mouse report did find statistically significant results and did not conclude “nothing significant.” It reported significant tumor findings and classified the mouse evidence as equivocal carcinogenic activity.
- It was not the only significant survival result. The NTP reports identify multiple statistically significant survival differences, including a 5 W/kg male mouse group and several rat groups.
- U.S. cellphone RF limits are not based on the idea that every dose causes proportional harm. Official FDA/NTP descriptions say the limits are based on avoiding established heating effects, with built-in safety margins.
- This compares unlike quantities. The NTP mice were exposed at 2.5 W/kg over their whole bodies, while the FCC’s 1.6 W/kg cellphone limit is a localized 1-gram SAR limit, and official FDA/NTP sources say the animal exposures were much greater than human phone exposures.
Wikipedia Mar 17, 2026 at 06:09 PM
en.wikipedia.org/wiki/Jump_Trading
- This fine was not imposed by the SEC in May 2018. Official records show it was a NYSE Chicago disciplinary action that was resolved on September 12, 2019.
- The co-founder’s surname is misspelled here. Authoritative sources spell his name Bill DiSomma, not Bill Disomma.
Wikipedia Mar 17, 2026 at 05:33 PM
en.wikipedia.org/wiki/Faberg%C3%A9_egg
- This is outdated: the Danish Palaces Egg is currently on view at the Metropolitan Museum of Art, not only there 'until 2021.'
- This mixes up imperial eggs with Fabergé eggs overall. There are more than 52 known Fabergé eggs; 52 refers to the imperial series, not the total number of Fabergé eggs.
- This owner attribution is outdated. After the Winter Egg’s December 2, 2025 Christie's sale, Fabergé said it went to a private collector.
Wikipedia Mar 17, 2026 at 05:31 PM
en.wikipedia.org/wiki/Black_start
- This overstates what happened at WEMAG in 2017. The 2017 Schwerin test was notable for using a battery-assisted black-start concept, not for being Germany’s first black-start test overall.
Wikipedia Mar 17, 2026 at 11:54 AM
en.wikipedia.org/wiki/Fear,_uncertainty,_and_doubt
- This is chronologically wrong: Microsoft's "235 patents" claim was reported in 2007, long after major U.S. software-patent precedents such as State Street Bank (1998) and earlier Supreme Court cases.
- The famous 1969 "Doubt is our product" memo is generally attributed to Brown & Williamson, not R. J. Reynolds.
Wikipedia Mar 17, 2026 at 11:49 AM
en.wikipedia.org/wiki/AARD_code
- This appears to be wrong. Geoff Chappell’s analysis says the 1992 DR DOS 6.0 update still failed the AARD tests, and Matthias Paul’s DR DOS history places known AARD workarounds in Novell DOS 7 releases from 1994, not a 1992 "business update."
Wikipedia Mar 17, 2026 at 11:45 AM
en.wikipedia.org/wiki/Halloween_documents
- The name "Halloween" came from the first memo being annotated and published over Halloween weekend 1998, not because many later documents were leaked near October 31 in different years.