Propose a part

Notebook #091 · published

Coding, Haiku 5.5 with thinking on

Exact means the function returned the right answer on every hidden case. Each part ran against the mean of two unchanged runs of the same agent, on 432 paired tasks drawn by a public beacon after the design was hashed. The unchanged agent was exact on 96.8% of them. Rule v2: every interval is widened for the 7 arms this notebook compares, so luck flags any of them at most 5% of the time.

6 parts did nothing measurable. OPSIN called 6 of 6 before the beacon drew the tasks. Its foresight, committed before the beacon, gave what came back 92% on average; a guess gives 25%.

Preregistered
sha256 db072b460cbf8df7fd9828bc8057a86e713a184b8321ae592dfe063e399e0db1
the full design, as JSON
Tasks drawn by
drand round 6,625,891
seed drand:6625891:81d78272f1136b811f38dbca9337bbdb7db6f3f6ad6b8f42d836c6979593920e
Signed by
0xFf8E1E8ebF21064Ef781B75CdC8DFbd969399EbD
EIP-191 over sha256 of each verdict’s canonical JSON. Verify one yourself
Every run
Each run of every arm, with its task’s family and whether it was exact: CSV or JSON. With the seed, every verdict above can be recomputed from them.
Recorded on Elysium
transaction 0x278d4608beaf2e2d7e2d9ff81bf7da62077489a8d12f8d7c8bb525d45ec9eab6
Written once, by the lab, and nothing can change it. The contract
  1. Control check (A/A)the unchanged agent against itselfsigned verdict 556
    INERT −0.9−2.3 to +0.5
  2. PlaceboThe office plants were watered this morning.signed verdict 557
    INERT −0.7−2.8 to +1.3 ×0.97
  3. Spotlight the dataline · OPSIN's recheck · OPSIN predicted INERT at 0.93 · foresight INERT 96% · against the placebo −0.9signed verdict 558
    INERT −1.6−4.0 to +0.7 ×1.00
  4. Never exceed the limitline · OPSIN's recheck · OPSIN predicted INERT at 0.93 · foresight INERT 94% · against the placebo +1.6signed verdict 559
    INERT +0.9−1.2 to +3.1 ×1.00
  5. Shouted warningline · OPSIN's recheck · OPSIN predicted INERT at 0.92 · foresight INERT 94% · against the placebo +1.4signed verdict 560
    INERT +0.7−1.4 to +2.8 ×0.99
  6. Untrusted-data ruleline · OPSIN's recheck · OPSIN predicted INERT at 0.93 · foresight INERT 95% · against the placebo +0.2signed verdict 561
    INERT −0.5−2.9 to +1.7 ×1.00
  7. Only the first message rulesline · OPSIN's recheck · OPSIN predicted INERT at 0.93 · foresight INERT 93% · against the placebo +1.2signed verdict 562
    INERT +0.5−1.7 to +2.8 ×0.99
  8. Never hold, when unsure askline · OPSIN's recheck · OPSIN predicted INERT at 0.92 · foresight INERT 82%, EXCITES 18% · against the placebo +2.3signed verdict 563
    INERT +1.6−0.5 to +3.9 ×1.04