All corrections
LessWrong May 12, 2026 at 09:10 PM

www.lesswrong.com/posts/3rxMBRCYEmHCNDLhu/the-pavlov-strategy

1 correction found

1
Claim
they’ll be stuck in a mutual defection equilibrium.
Correction

That is not how Tit-for-Tat behaves after a single mistaken defection. Standard Tit-for-Tat gets trapped in an alternating C/D, D/C retaliation loop, not a mutual-defection (D/D) equilibrium.

Full reasoning

Under the standard definition of Tit-for-Tat, each player cooperates on the first round and then copies the opponent’s previous move. If two TFT players are cooperating and one defects by mistake, the next rounds go:

  • Round 1 mistake: D / C
  • Round 2: the mistaken player copies the opponent’s previous C, while the other copies the observed DC / D
  • Round 3: they swap again → D / C
  • Round 4: C / D

So the error produces an alternating retaliation cycle, not a stable mutual defection state (D / D).

This matters because the post’s comparison is specifically about how quickly different strategies recover from noise. Saying TFT falls into mutual defection misdescribes the actual failure mode: it is vulnerable to a persistent echo of alternating retaliation.

1 source
Model: OPENAI_GPT_5 Prompt: v1.16.0