Manas Bihani
About

the questions

  1. What is a moat in an AI world?
  2. Why do AI products converge?
  3. What becomes scarce when intelligence becomes cheap?
  4. Does distribution matter more than technology?
  5. Why might human-made things become more valuable?
  6. What happens to expertise when everyone has the same models?
  7. Which parts of an AI startup are actually defensible?
  8. Where does value move when intelligence becomes commoditized?

everything on the desk

  1. The periodic table of the AI stackVisualization
  2. What is a moat when the model isn't yours?Note
  3. The problem-selection premiumNote
  4. Same model, different wiringNote
  5. Selection is the new bottleneckNote
  6. Get friendly with the AI raceEssay
  7. The convergence taxNote
  8. The luxury of realityNote
  9. The non-technical technical advantageNote
  10. The verification economyNote
  11. The bets against the wallVisualization
  12. You can't buy your way outVisualization
  13. How a chatbot writes one wordVisualization
  14. The grid is the last wallVisualization
  15. Who got paidVisualization
  16. Why this paper mattersExplainer
  17. Transformer: Why did transformers replace RNNs?Vaswani et al., NeurIPS 2017
  18. KV cache: Why does a long conversation get slower and cost more than a short one?Shazeer, 2019
  19. Mixture of experts: Why do some AI models have experts?Fedus, Zoph and Shazeer, 2021
  20. FlashAttention: Why is attention slow when the GPU is barely doing any arithmetic?Dao et al., NeurIPS 2022
  21. Mamba: Why does a model reread the whole conversation instead of just remembering it?Gu & Dao, 2023
  22. PagedAttention: Why does a GPU with free memory still refuse new requests?Kwon et al., SOSP 2023
  23. DeepSeek: How did DeepSeek train a frontier model so cheaply?DeepSeek-AI, 2024
  24. Jamba: Why does Jamba matter?Lieber et al., AI21 Labs, 2024
  25. BitNet: Why does BitNet matter?Ma et al., Microsoft Research, 2025
  26. DeepSeek-R1: Can a small AI model learn to reason like a huge one?DeepSeek-AI, 2025
  27. Kimi K2: Why does Kimi K2 matter?Kimi Team, Moonshot AI, 2025
  28. Sliding-window attention: How do models handle huge context windows without the memory bill exploding?Gemma Team, Google DeepMind, 2025
  29. How electricity becomes intelligenceVisualization
  30. This desk, as a datasetDataset
  31. The first version of this roomNote
  32. The aura dividendNote
  33. Distribution is rented attentionNote
  34. The Convergence TestNote
  35. A shelf for thinking about cheap intelligenceCollection
  36. Anatomy of an AI startupNote
  37. Six shocks to expertiseNote
  38. Nineteen Public KeysEssay
  39. The value migration machineModel
  40. AAA-Rated GPUsEssay
  41. Moats, before and afterVisualization
  42. The rhinoceros problemNote
  43. What becomes scarce when intelligence becomes cheap?Essay
  44. AI Has Passed Every Exam. It Has Never Had an Idea.Essay
  45. What Becomes Scarce After Intelligence?Essay
  46. India’s Carbon Markets : A New Test for Global Climate PolicyEssay
  47. Google Wants AI to Become BoringEssay
  48. The Wall That Wasn’t YoursEssay
  49. The Rate-Limiting StepEssay
  50. The Speed of Being WrongEssay
  51. Uber Burned a Year of AI Budget in Four Months. A Rat Catcher in 1902 Knew WhyEssay
  52. Finding a Flat in India Is Broken. We Have the Technology to Fix It. Nobody With Power Wants To.Essay
  53. Why We Can Never Have Good Social MediaEssay
  54. Gen Z Is Going OfflineEssay

rooms

  1. Home
  2. Writing
  3. Projects
  4. Reading & Watching
  5. All the questions
  6. Everything, as a contact sheet
  7. About

Dataset · 24 Sept 2026

This desk, as a dataset

Every object on this site, one row each, rebuilt every time the site is. Download it and do something strange with it.

yes, the site is also a spreadsheet

A dataset should look like a dataset, so here it is: every object on this desk, what form it took, which questions it was filed under, and how many words it needed. It’s generated from the site’s own content, so it can’t drift out of date.

ABCDEF
1dateobjectformquestionswordsstatus
22026-03-15Gen Z Is Going OfflineEssay—1387finished
32026-03-21Why We Can Never Have Good Social MediaEssay—1255finished
42026-05-10Finding a Flat in India Is Broken. We Have the Technology to Fix It. Nobody With Power Wants To.Essay—2343finished
52026-05-31Uber Burned a Year of AI Budget in Four Months. A Rat Catcher in 1902 Knew WhyEssay—1681finished
62026-06-05The Speed of Being WrongEssay—2039finished
72026-06-09The Rate-Limiting StepEssay—1354finished
82026-06-13The Wall That Wasn’t YoursEssay—1508finished
92026-07-04Google Wants AI to Become BoringEssay—1920finished
102026-07-08India’s Carbon Markets : A New Test for Global Climate PolicyEssay—1714finished
112026-07-21What Becomes Scarce After Intelligence?Essay—1616finished
122026-08-01AI Has Passed Every Exam. It Has Never Had an Idea.Essay—2071finished
132026-08-04What becomes scarce when intelligence becomes cheap?Essayscarcity value548finished
142026-08-11The rhinoceros problemNoteexpertise135finished
152026-08-19AAA-Rated GPUsEssay—1236finished
162026-08-19Moats, before and afterVisualizationmoat defensible175finished
172026-08-28The value migration machineModelvalue scarcity275finished
182026-09-01Nineteen Public KeysEssay—2222finished
192026-09-02Six shocks to expertiseNoteexpertise human-made134ongoing
202026-09-08Anatomy of an AI startupNotedefensible moat212ongoing
212026-09-12A shelf for thinking about cheap intelligenceCollectionmoat value scarcity human-made53finished
222026-09-17The Convergence TestNoteconvergence moat245ongoing
232026-09-20Distribution is rented attentionNotedistribution moat107finished
242026-09-22The aura dividendNotehuman-made107finished
252026-09-23The first version of this roomNote—89finished
262026-09-24This desk, as a datasetDataset—80finished
272026-09-25How electricity becomes intelligenceVisualizationscarcity value0finished
282026-09-26The bets against the wallVisualizationscarcity moat1172finished
292026-09-26You can't buy your way outVisualizationscarcity moat473finished
302026-09-26How a chatbot writes one wordVisualizationscarcity0finished
312026-09-26The grid is the last wallVisualizationscarcity value200finished
322026-09-26Who got paidVisualizationvalue scarcity1145finished
332026-09-26Why this paper mattersExplainerscarcity98finished
342026-09-27What is a moat when the model isn't yours?Notemoat defensible189draft
352026-09-27The problem-selection premiumNoteexpertise value184draft
362026-09-27Same model, different wiringNoteconvergence defensible185draft
372026-09-27Selection is the new bottleneckNotescarcity expertise177draft
382026-09-27Get friendly with the AI raceEssayscarcity value2998finished
392026-09-27The convergence taxNoteconvergence moat150draft
402026-09-27The luxury of realityNotehuman-made scarcity178draft
412026-09-27The non-technical technical advantageNoteexpertise181draft
422026-09-27The verification economyNotescarcity human-made191draft
432026-10-06The periodic table of the AI stackVisualizationscarcity value976ongoing
Every object on the desk. Download it as a CSV: /desk.csv

Download desk.csv. It’s small now. The interesting version is in a year, when it can answer questions like: which questions keep pulling me back, and which forms do I reach for when I’m unsure?