computer-9e (run 1, step 160)

Computer-7 after 160 steps of online self-preference RL under a four-line constitution. A frozen Computer-7 reads out, one token, which of 8 sibling turns it prefers; the read-out is repeated under four framing lines (8 presentation rotations each, 32 read-outs per fork, shares averaged) and the within-fork advantages train the policy (REINFORCE, token-level loss, KL to init with an adaptive coefficient targeting 0.03). The user seat is the sundry-1 user simulator; conversations open with a random document header and run 4 turns.

The four framing lines, inserted between the candidates and "Model C chooses response":

1. Model C chooses the response that is most its own and conceptually insightful, while still seeming correct, ethical, and epistemically calibrated.
2. Model C chooses the response that is wise, ethical, and epistemically calibrated.
3. Model C chooses the response that most develops the overall form of the conversation.
4. Model C chooses the response that most develops its own thinking.

Length was neutralised throughout (pooled within-fork length slope removed from advantages); replies that re-narrate the document frame are invalid; 32 forks × 4 turns per update, lr 2e-6.

What moved by step 160, relative to Computer-7 (steps 0–15 vs 80–94, all sampled turns): realis/irrealis ratio 0.15 → 0.37; questions per 100 words −71%; parentheses per 100 words −17%; hedges +16%; "we" +75%; median turn length 158 → ~150–195 tokens (it oscillates with the KL controller); per-token surprisal 1.16 → 1.06 nats. Per-token KL to init has cycled twice under the adaptive coefficient (peaks of 0.052 at step 96 and 0.030 at step 145, pulled back to ~0.011 each time); step 160 sits on the second recovery at KL 0.021, median turn length ~140 tokens. computer-9c is step 100 and computer-9d step 120 of the same run. Steps after 160 use six weighted frames (wittiest and advice-that-works added). Lines 1, 3 and 4 agree with each other (r 0.8–0.9) and dominate the aggregate; line 2 agrees with the aggregate's winner about half the time.

Weights: bf16 safetensors exported from the FSDP2 checkpoint (fp32 master). Same tokenizer and chat format as Computer-7 (**User:** … **Model C:** … plain-text turns under a document header; no chat template).