Latest Corrections

LessWrong Mar 22, 2026 at 12:39 AM
www.lesswrong.com/posts/6pDMLYr7my2QMTz3s/broad-timelines
  • Ege Erdil was a researcher at Epoch AI, not a co-founder. Epoch AI’s own founding announcement lists seven founders and does not include him.
LessWrong Mar 22, 2026 at 12:32 AM
www.lesswrong.com/posts/Psr9tnQFuEXiuqGcR/how-to-write-quick...
  • Historical population estimates for major premodern empires vary much more widely than 60–80 million. Standard references list several of the largest preindustrial empires well above 80 million, including the Qing, Mughal, and Ming polities.
LessWrong Mar 22, 2026 at 12:21 AM
www.lesswrong.com/posts/d65Ax6vbNgztBE8cy/new-lesswrong-feat...
  • The feature did reveal non-match interest signals: users received a "New users interested in dialoguing with you (not a match yet)" notification before matching.
LessWrong Mar 21, 2026 at 11:51 PM
www.lesswrong.com/posts/ZB7Hez6bJiZvCAAA2/theories-of-pain
  • Fascia is not a substance that switches between liquid and solid states. It is connective tissue; anatomy sources describe it as collagen-rich fibrous tissue, with a separate lubricating liquid between fascial layers.
LessWrong Mar 21, 2026 at 11:18 PM
www.lesswrong.com/posts/bZ2w99pEAeAbKnKqo/optimal-exercise
  • Women do have testosterone; they typically just have much lower levels than men.
  • Later meta-analysis evidence found resistance-training gains continuing up to about 1.62 g/kg/day, which is roughly 0.73 g/lb—not 0.64 g/lb.
LessWrong Mar 21, 2026 at 11:13 PM
www.lesswrong.com/posts/Zz986H9P3WJoh5DNb/the-dark-arts-a-be...
  • Research reviews on NLP's "predicate matching" idea have not found that mirroring a person's visual/auditory wording reliably increases influence or rapport.
LessWrong Mar 21, 2026 at 10:56 PM
www.lesswrong.com/posts/sP2Hg6uPwpfp3jZJN/lost-purposes
  • This overstates physics publishing norms. Physics journals do publish papers reporting effects with much weaker significance than p<0.0001, and even particle physics' famous 5-sigma rule is a discovery convention, not a blanket journal requirement.
LessWrong Mar 21, 2026 at 10:53 PM
www.lesswrong.com/posts/reitXJgJXFzKpdKyd/beware-trivial-inc...
  • China’s internet censorship has never removed *all* politically sensitive content. Credible reporting and later large-scale research both describe it as selective and incomplete, with some criticism and dissident material remaining accessible.
LessWrong Mar 21, 2026 at 10:47 PM
www.lesswrong.com/posts/kxW6q5YdTGWh5sWby/eight-hundred-slig...
  • This overstates the difference. Reported CO2 levels in submarines and spacecraft are usually only a few times higher than ordinary indoor air, not 10–100× higher, and submarine averages can overlap common civilian exposure limits.
Wikipedia Mar 21, 2026 at 09:42 PM
en.wikipedia.org/wiki/Whirlpool
  • This is incorrect: English sources used the word "maelstrom" centuries before Poe. Major dictionaries date its first known English use to 1588 and describe Poe as helping popularize it, not coin it.
  • This widely repeated figure is wrong. Credible university sources describe Lake Peigneur after the disaster as about 200 feet deep, not 1,300 feet deep.
  • These casualty counts are far too high. Credible accounts of the Lake Peigneur disaster report 11 barges and one tugboat being swallowed, not 19 barges and 8 tugboats.
  • The speed figure here is overstated. Norwegian reference and oceanographic sources describe Moskstraumen at roughly 6 knots (about 11 km/h) to 5 m/s (about 18 km/h), not 32 km/h.
LessWrong Mar 21, 2026 at 06:45 PM
www.lesswrong.com/posts/ybwcxBRrsKavJB9Wz/no-we-haven-t-uplo...
  • Shiu et al.'s 2024 model was not a full model of every neuron and connection in the adult fly brain. The paper says it was built from the adult Drosophila central brain connectome and modeled about 127,400 neurons, whereas the whole-brain FlyWire connectome contains 139,255 neurons.
Substack Mar 21, 2026 at 04:43 PM
peterwildeford.substack.com/p/the-fordow-paradox-where-do-ir...
  • The pre-2231 UN sanctions on Iran were targeted measures, not a blanket ban on all trade with Iran, and UN Security Council sanctions do not automatically create U.S.-style worldwide secondary sanctions.
Substack Mar 21, 2026 at 04:40 PM
peterwildeford.substack.com/p/the-pentagons-war-on-anthropic
  • That misstates how broad the Defense Production Act is. The DPA is also used for preparedness and emergency response, including natural disasters and COVID-era industrial-base and medical-supply efforts — not only acute national-security emergencies.
Substack Mar 21, 2026 at 04:37 PM
nathanpmyoung.substack.com/p/my-vibes-are-shifting-on-ai-ris...
  • ChatGPT did not launch in 2024. OpenAI publicly released ChatGPT on November 30, 2022.
Substack Mar 21, 2026 at 04:24 PM
sortega.substack.com/p/interesting-reading-in-november-60d
  • The book title is incorrect: Max Harms’s novel is titled *Red Heart*, not *Read Heart*.
Wikipedia Mar 21, 2026 at 01:13 PM
en.wikipedia.org/wiki/Atom_(web_standard)
  • This overstates the difference from AtomPub. Micropub does use OAuth 2.0 bearer tokens, but AtomPub does not mandate HTTP Basic Authentication as its authentication method.
Wikipedia Mar 21, 2026 at 10:57 AM
en.wikipedia.org/wiki/Musk
  • This overstates how deer musk is obtained. Authoritative sources say musk has traditionally been harvested by killing musk deer, but it can also be extracted from live adult males.
Wikipedia Mar 21, 2026 at 10:47 AM
en.wikipedia.org/wiki/Library_Genesis
  • This list of “gateway providers” is inaccurate. IPFS is the network/protocol itself, not a provider, and the complaint identified the gateway companies as Cloudflare, Protocol Labs (IPFS.io), and Pinata.
  • The $30 million default judgment was entered on September 24, 2024, not September 26, 2024.
Wikipedia Mar 21, 2026 at 10:44 AM
en.wikipedia.org/wiki/Z-Library
  • The underlying traffic source reports about 2.84 million visits, not 2.84 million users.
Wikipedia Mar 21, 2026 at 02:16 AM
en.wikipedia.org/wiki/Sexual_objectification
  • Cahill’s term is "derivatization," not "derivitization." Her book title page, description, and chapter list consistently use the spelling "derivatization."
  • Cahill calls this concept "derivatization," not "derivitization." The sentence is describing the right idea under the wrong term.