~/reading

121 entries · updated 2026-10-03 · my personal favorite reads

## short form57 entries

  1. read together

  2. read together

  3. 05this is water (kenyon commencement, 2005) · david foster wallace
  4. 06child's play · sam kriss (harper's)
  5. 07in defense of a loaded word · ta-nehisi coates
  6. 08on thinking · northrop frye

  7. read together

  8. 11welcome to harvard. when are you leaving? · the crimson editorial board
  9. 12can quantum computers solve chess? · the chessboard vault
  10. 13randar: an explanation · spawnmason
  11. 14a global workspace in language models · anthropic
  12. 15american kids can't do math anymore · the atlantic
  13. 16such, such were the joys · george orwell
  14. 17e unibus pluram: television and u.s. fiction · david foster wallace
  15. 18the internet is already over · sam kriss

  16. read together

  17. 21the kekulé problem · cormac mccarthy
  18. 22i can tolerate anything except the outgroup · scott alexander

  19. read together

  20. 25on the shortness of life · seneca
  21. 26the disadvantages of an elite education · william deresiewicz
  22. 27cowardice and the closing of hampshire college · faune albert (the chronicle of higher education)
  23. 28in defense of distraction · sam anderson
  24. 29the ones who walk away from omelas · ursula k. le guin
  25. 30on self-respect · joan didion
  26. 31a modest proposal · jonathan swift

  27. read together

  28. 34the urgency of interpretability · dario amodei
  29. 35situational awareness · leopold aschenbrenner
  30. 36the intelligence curse · luke drago & rudolf laine

  31. read together

  32. read together

  33. 41drowning in slop · max read
  34. 42the coddling of the american mind · greg lukianoff & jonathan haidt
  35. 43the moral decline of elite universities · ben sasse (the atlantic)

  36. read together

  37. 46on the phenomenon of bullshit jobs · david graeber
  38. 47the company man · tomás b. (lesswrong)

  39. read together

  40. 50what is -1 to 0? · Ruchi Sanghvi
  41. 51superlinear returns · paul graham
  42. 52the lesson to unlearn · paul graham
  43. 53coasean bargaining at scale · seb krier (cosmos institute)
  44. 54ai as normal technology · arvind narayanan & sayash kapoor
  45. 55anatomy of an ai system · kate crawford & vladan joler
  46. 56do artifacts have politics? · langdon winner (daedalus, 1980)
  47. 57humans are bad at predicting futures that don't benefit them · caroline beaton (the atlantic)

## long form28 entries

  1. 01genesis: artificial intelligence, hope, and the human spirit · kissinger, schmidt & mundie

  2. read together

  3. 04solving mathematical problems (second edition) · terence tao
  4. 05amorality and greed at silicon valley's favorite university · theo baker

  5. read together

  6. 08point counter point · aldous huxley
  7. 09metamorphosis · franz kafka
  8. 10the trial · franz kafka
  9. 11white noise · don delillo
  10. 12swamplandia! · karen russell

  11. read together

  12. 15mao ii · don delillo
  13. 16excellent sheep · william deresiewicz
  14. 17technopoly · neil postman
  15. 18the shallows · nicholas carr
  16. 19the true believer · eric hoffer
  17. 20the alignment problem · brian christian
  18. 21blood meridian · cormac mccarthy
  19. 22east of eden · john steinbeck
  20. 23where is my flying car? · j. storrs hall
  21. 24the road to serfdom · friedrich hayek
  22. 25thinking in systems · donella meadows
  23. 26enlightenment now · steven pinker
  24. 27thinking, fast and slow · daniel kahneman
  25. 28the selfish gene · richard dawkins

## academic36 entries

  1. 01gpt detectors are biased against non-native english writers · liang et al. (arxiv 2304.02819)
  2. 02efficient test-time adaptation through human-ai interaction · wang et al. (arxiv 2609.04141)
  3. 03deep learning for individual heterogeneity · farrell, liang & misra (arxiv 2010.14694)
  4. 04base models can reason by taking a cue from training data · wang et al. (arxiv 2610.06851)
  5. 05the best-laid schemes: coordinated sabotage and monitoring in multi-agent systems · radev et al. (arxiv 2605.29178)
  6. 06glm-5.3 sometimes considers how train-time actions shape future behavior · axel højmark & ezra newman (lesswrong)
  7. 07infer human's intentions before following natural language instructions · wan et al. (aaai 2025)
  8. 08personalizing reinforcement learning from human feedback with variational preference learning · poddar et al. (neurips 2024)
  9. 09neural networks are already clean · nick cammarata (goodfire)
  10. 10ai models may be capable of self-replicating without access to a datacenter · william zhou & daniel freeman (anthropic)
  11. 11debating with more persuasive llms leads to more truthful answers · khan et al. (arxiv 2402.06782)
  12. 12characterizing delusional spirals through human-llm chat logs · moore et al. (facct 2026)
  13. 13a single neuron is sufficient to bypass safety alignment in large language models · kazemi, chegini & safi (arxiv 2605.08513)
  14. 14introducing system one models & jev · typesafe ai (diogo almeida)

  15. read together

  16. 17Language Models Can Control Their Own Attention · ho et al. (arxiv 2609.02737)
  17. 18Measuring the Gap Between Human and LLM Research Ideas · zhao et al. (arxiv 2607.01233)
  18. 19improving the matrix multiplication exponent with modern optimization and alphaevolve · dupont et al. (arxiv 2608.16884)
  19. 20mind viruses: self-propagating ideas in multi-agent llm systems · papadopoulos et al. (arxiv 2608.10218)
  20. 21stealing reasoning traces from proprietary llm apis · panfilov et al. (arxiv 2608.09867)
  21. 22matraix: simulating the world with 8.3 billion persona agents · li et al. (arxiv 2608.04205)
  22. 23why ais (might) obey the law · sachdeva & kolt (ssrn 2026)
  23. 24prompt injection as role confusion · ye, cui & hadfield-menell (icml 2026)

  24. read together

  25. 27learning representations by back-propagating errors · rumelhart, hinton & williams (1986)
  26. 28why larger models learn more: effects of capacity, interference, and rare-task retention · arxiv 2605.29548
  27. 29how much do llms hallucinate in document q&a scenarios? · arxiv 2603.08274
  28. 30self-harness: harnesses that improve themselves · arxiv 2606.09498
  29. 31on the dangers of stochastic parrots · bender, gebru, mcmillan-major & mitchell (faact 2021)
  30. 32the bitter lesson · rich sutton

  31. read together

  32. 35clever hans or neural theory of mind? · shapira et al. (2023)
  33. 36every author as first author · demaine & demaine (arxiv 2304.01393)