Post-Cutoff

Posts that mattered

1,633 posts by 887 authors: the X posts, essays and blog posts our events cite, newest first.

October 2026, continued

  1. Cassandra Unchained@michaeljburryX

    Michael Burry (‘Cassandra Unchained’) on Hock Tan: ‘Well well well. Well.’

    The best-known AI-bubble short seller amplified Tan’s remarks, quoting Bloomberg’s passage on enterprises ‘wasting money’ on tokens (about 408k views).

  2. Microsoft@MicrosoftX

    Microsoft: ‘Introducing Maia 300, Microsoft’s next-generation accelerator’

    Microsoft’s corporate announcement post for Maia 300.

  3. NIK@ns123abcX

    Clip: Elon Musk says Grok’s reasoning traces in RL feel like ‘a person under extreme pressure with a gun to their head’ and warns against ‘torturing a superintelligence’

    ~911k views; the only located source for Musk’s spoken remarks agreeing with ‘some of the things Anthropic has said’ about mistreating models, days after his post backing Anthropic’s ban on cruelty to Claude.

  4. Rohan Arun@RohanArunX

    “Update: New conditional kappa = 0.001467289440740487 κ is now capped by the complex…”

    Cited in Outside researchers using AI crowdsource a tighter exponent in OpenAI’s integer-multiplication result #109, from 2^-182 to about 2^-8.4 (conditional)

  5. Rohan Arun@RohanArunX

    “Update: we hit a hard wall again and progressed only 0.7% since last night.”

    Cited in Outside researchers using AI crowdsource a tighter exponent in OpenAI’s integer-multiplication result #109, from 2^-182 to about 2^-8.4 (conditional)

  6. Marcus Bintz@spurious_bowsX

    “Pair 4: (a) Sharp mass bounds for the two-dimensional O(4) model [9/23] (b) The…”

  7. Tim Sweeney@timsweeneyepicX

    “Our more modest aim is to survive the AI revolution, help other developers do the…”

    Cited in Epic Games reportedly moves its Disney shooter to an ‘AI first’ model with ‘no more hand-written code’; Sweeney confirms a ‘huge pivot’

  8. Tim Sweeney@TimSweeneyEpicX

    Tim Sweeney: AI gains at Epic range ‘from 20% in some areas ... to 5x for other areas’

    First-hand numbers from Epic’s CEO on current AI productivity gains and his forecast for 2027.

  9. Tim Sweeney@TimSweeneyEpicX

    Tim Sweeney: ‘The biggest change will be much faster time from having an idea to playtesting it’

    First-hand: Epic’s CEO on what he expects AI development to change in game production.

  10. Tim Sweeney@TimSweeneyEpicX

    Tim Sweeney: ‘Much of the room wants it to be 2016 again, but it’s not’

    First-hand: Epic’s CEO on the mood around the AI pivot; his most-viewed reply (about 288k views).

  11. Vals AI@ValsAIX

    Vals AI: Muse Wonder Full debuts at #5 on the Vals Index, above every OpenAI model

    Second independent benchmark placement for Muse Wonder Full.

  12. Federico Viticci@viticciX

    Federico Viticci: ‘Muse Wonder feels…great?’ (10 subagents for code review, a voxel Colosseum)

    First-hand day-one test by the MacStories editor, with a shareable demo; a positive counterweight to early ‘benchmaxxed’ criticism.

  13. Wario64@Wario64X

    Wario64: Bloomberg quote ‘No more hand-written code’ at Epic

    The post that carried the Bloomberg report to the gaming audience (about 894k views); quotes the paywalled article directly.

  14. David RobertsBlog

    David Roberts: ‘What even is mathematical output?’

    A mathematician who formalises in Lean for his day job argues that mathematicians are judged by their writing, and that OpenAI’s release of over 700 unrefined documents is therefore ‘not just frustrating, but insulting’; he collects expert verdicts on the papers and points to junk Lean files in the repository. Posted by the author under Tao’s Oct 12 reading list.

  15. Terence TaoBlog

    Tao: ‘Additional recent pieces on math and AI’ (reading list)

    Tao’s own list of the essays he considers worth reading on mathematics and AI in autumn 2026; a map of the debate among mathematicians.

  16. HuahaiBlog

    AI Is a Tool. Using It Well Takes Skill.

    The most-upvoted reply (301 points on Hacker News) to Rich Hickey’s “‘AI’ is Not a Tool”.

  17. Max Blade@_MaxBladeX

    Max Blade: Muse Wonder Full vs Opus 5.5 on one game prompt; $4.11 of tokens used up a $50 Muse Code plan

    Widely viewed (about 297k views) user comparison of Muse Wonder Full and Claude Opus 5.5, with a first-hand report on Muse Code plan limits.

  18. aaron holmes@aaronpholmesX

    Aaron Holmes (The Information): Maia 300 announcement confirms his August report

    A reporter ties the announcement to The Information’s August report on Microsoft’s chip ramp (more than 300k chips next year).

  19. Alexandr Wang@alexandr_wangX

    Alexandr Wang: bug in how Muse Wonder counts against Muse Code usage limits

    First-hand acknowledgement by Meta’s AI chief of a launch-day usage-limit bug for Muse Wonder and Wonder Full, with a promise of resets.

  20. Andreas Schilling 🇺🇦@aschillingX

    Andreas Schilling: Maia 300 chiplet configuration

    Compact list of the package layout as presented by Microsoft.

  21. Ben Bajarin@BenBajarinX

    Ben Bajarin: Maia 300 rack and networking details

    Analyst notes that add deployment details: 96 accelerators per rack at about 167 kW, 30% better tokens per dollar, 12,000-accelerator plan of record.

  22. Ben Murphy@benjaminmmurphyX

    Ben Murphy: ‘we are going to quickly need to adapt to a world with omnipresent swarms’

    Policy reaction to the Anthropic study from a researcher whose profile lists RAND CAST, IFP and FAI.

  23. Bill Kristol@BillKristolX

    Bill Kristol on Anthropic’s self-replication warning: ‘This seems…problematic?’

    A political commentator with about 1M followers carried the story outside the AI audience.

  24. Brendan Burke@BrendanBurkeXX

    Brendan Burke (Futurum): Maia’s gigawatt plan is a bet on software-configured dataflow and agent-written kernels

    Detailed analyst thread on how Microsoft uses Maia (RL rollouts and synthetic data for its Superintelligence team, pretraining on NVIDIA) and on agentic kernel generation.

  25. Charlie Marsh@charliermarshX

    Charlie Marsh (OpenAI): ‘publishing intelligence without cost is not-great’

    Criticism from an OpenAI employee of how Muse Wonder Full was benchmarked: an Intelligence Index score with no per-token price.

  26. Deirdre Bosa@dee_bosaX

    Deirdre Bosa quoting Scott Guthrie: custom chips narrow the ‘cone of uncertainty’

    Carries a direct quote from Microsoft’s cloud chief, from her live interview on the day of the Maia 300 announcement.

  27. Ben@deobfuscatelyX

    Ben (Synthient): ‘Surreal to see some of the @synthient research mentioned by @AnthropicAI’

    First-hand: a member of Synthient says Anthropic’s self-replication post draws on Synthient’s research.

  28. Ed Ludlow@EdLudlowX

    Ed Ludlow (Bloomberg): Microsoft has not named the GPU baseline for Maia 300’s 40% claim

    A reporter’s clarification of the headline efficiency claim: no named comparison system.

  29. Ed Zitron@edzitronX

    Ed Zitron: ‘multi gigawatt deployments’ wording questioned; ‘Microsoft only has 2gw of ai data centers’

    Critic’s reading of the Maia announcement; contains an unverified capacity claim.

  30. Elon Musk@elonmuskX

    Elon Musk replies ‘Hmm’ to BridgeMind’s ‘Muse Wonder is benchmaxxed’ post

    A rival lab head’s only reaction to Muse Wonder in this set: a one-word reply amplifying a critical test.

  31. Emad@EMostaqueX

    Emad Mostaque on Epic’s pivot: ‘The step change in game x AI capability in the last few weeks is the biggest ever’

    Supportive reaction from the Stability AI founder, in contrast to the developer backlash.

  32. FireMonkey@FireMonkeyX

    FireMonkey: Epic employees joked about having ‘Claude fix its own mistake’ (via Bloomberg)

    Relays further details of the paywalled Bloomberg report on how Epic staff reacted to the ‘AI first’ instruction.

  33. tae kim@firstadopterX

    Tae Kim: context may be ‘lost in translation’ in Bloomberg’s Hock Tan article

    A semiconductor journalist’s caveat on the casino quote: Broadcom’s own future depends on Anthropic and OpenAI succeeding.

  34. Grummz@GrummzX

    Grummz: ‘This is the shot heard round the world for games and AI’

    Widely viewed reaction (about 125k views) from a former game producer to Sweeney’s confirmation.

  35. Samuel Hammond 🦉@hamandcheeseX

    Samuel Hammond: an ‘online agent population explosion’ once small open models can earn money

    Policy analyst’s reading of Anthropic’s self-replication post: the limit becomes the agents’ revenue, not the hardware.

  36. ItsMirror@its_mirrorsX

    ItsMirror (solo game developer) on Tim Sweeney’s AI pivot post

    Most-liked reaction to Sweeney’s post in the sweep (about 6.9k likes, 68k views), from an indie developer; an example of the backlash among game makers.

  37. Jeff Weinstein@jeff_weinsteinX

    Jeff Weinstein (Stripe): ‘We work hourly—often minutely—with the Meta team’ on agentic commerce

    First-hand statement from a Stripe product lead on Stripe’s work with Meta on Muse’s shopping and payment use cases.

  38. Jacob Navok@JNavokX

    Jacob Navok (Genvid CEO): ‘It is over for non-AI workflows’

    Reaction from a games-industry executive to Epic’s pivot.

  39. Karla Ortiz@kortizartX

    Karla Ortiz to Epic Games: ‘stop using GenAi for art’

    Artist-side objection to Epic’s pivot from a concept artist who says she is suing generative-AI companies.

  40. Mustafa Suleyman@mustafasuleymanX

    Mustafa Suleyman: ‘Our new silicon is 🔥 40% more tokens per watt!’

    First-hand reaction from the CEO of Microsoft AI, whose MAI model team is a named user of Maia.

  41. Patrick Moorhead@PatrickMoorheadX

    Patrick Moorhead on Maia 300: ‘primarily a margin lever, not a benchmark story’

    Industry analyst’s assessment with the caveat that the chip is untested and GPUs will move before 2027.

  42. Paul Tassi@PaulTassiX

    Paul Tassi on Epic’s ‘no more hand-written code’: ‘a huge deal one way or the other’

    Games journalist with a large following frames what to watch: quality, speed, blowback.

  43. Rob Hallam@robj3d3X

    Rob Hallam on the Cheeky Pint episode: ‘3 billionaires chatting over beers’

    Highest-reach quote post of the Zuckerberg–Wang interview in the sweep (about 101k views); shows how the launch interview was received.

  44. Ryan Orhan@rynorhnX

    Ryan Orhan: Anthropic thinks small AI models could be ‘just MONTHS away’ from replicating themselves across everyday devices

    One of the widest-read lay summaries of Anthropic’s self-replication post (about 89k views in a day from a 7.5k-follower account); it shows how the finding spread outside safety circles.

  45. Lisan al Gaib@scaling01X

    Lisan al Gaib: Anthropic quote on small models self-replicating on IoT devices

    Most-viewed post about the Anthropic self-replication study in this set (about 232k views).

  46. Mark Malone@soundmigrationX

    Mark Malone: ‘Mark Zuckerberg pretending to enjoy Guinness is my new favourite thing’

    By far the most-viewed post linked from the Muse Wonder launch-day discussion (3.7M views); a joke about the Cheeky Pint video, with no information on the model.

  47. Supreme Bagholder@SupBagholderX

    Supreme Bagholder to Alexandr Wang: ‘The subscription sprawl is getting silly’

    User complaint with reach (about 32k views) about three separate Meta bills: Meta One, a Muse subscription and a Muse Code subscription.

  48. Theo - t3.gg@theoX

    Theo (t3.gg): ‘Meta is a frontier lab now. OpenAI has fallen into 3rd place’

    Widely viewed developer reaction to the Artificial Analysis results for Muse Wonder; opinion.