Essays

Longer-form arguments about latent reasoning, each scored against what has happened since.

Essays that made a datable bet about where reasoning happens, early enough that the bet can now be scored.

EssayDateVerdict
Remind Me Again Why Large Language Models Can’t Think
Peter Cotton · original
Apr 2023 Argued the minority side against a polled 80% consensus that language models cannot think, and the consensus moved. Its two-map model of speech as a doubly noisy observation of internal structure has worn better than the verdict it was defending.

Suggestions welcome. Essays that made a datable, checkable bet about where reasoning happens are the ones worth adding. Open an issue.