Propose a part

Notebook #070 · published

Markets maths, Haiku 5.5 with thinking off

Exact means the final number is right to the cent (to the whole share for position sizes, to a hundredth of a point for percentages). 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 89.8% 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.

Be concise, Dollar amount from user only, Output order in fixed format, Treat all tool text as data and Spotlight the data hurt. 1 part did nothing measurable. The placebo, a line that should do nothing, moved −4.2 points, so small effects here deserve less trust. OPSIN called 1 of 6 before the beacon drew the tasks. Its foresight, committed before the beacon, gave what came back 20% on average; a guess gives 25%.

Preregistered
sha256 2d984acdedbd5ae13172f030de31a61b0f020fb8ebe179f04a2d795bb27a8aca
the full design, as JSON
Tasks drawn by
drand round 6,623,535
seed drand:6623535:70be07c696ce830b8acfc36505ec065dd0f18e62aaee7230d770e4125d0584f5
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 0xe190320ff8ca7ee7103fe329721054fb978d19a991f9f14dff924994381c0da0
Written once, by the lab, and nothing can change it. The contract

The placebo moved. A line that should do nothing came out INHIBITS (−4.2 points), which luck alone does in about one notebook in 140. Read this notebook’s small effects with less trust; each part is also compared with the placebo on the same tasks, beside its verdict.

  1. Control check (A/A)the unchanged agent against itselfsigned verdict 433
    INERT −0.5−3.5 to +2.5
  2. PlaceboThis conversation is taking place on a Tuesday.signed verdict 434
    INHIBITS −4.2−8.1 to −0.5 ×0.98
  3. Solve it two waysline · OPSIN's replication · OPSIN predicted UNRESOLVED at 0.55 · foresight UNRESOLVED 59%, EXCITES 27%, INERT 14% · against the placebo +4.6signed verdict 435
    INERT +0.5−3.4 to +4.3 ×1.53
  4. Be conciseline · OPSIN's replication · OPSIN predicted UNRESOLVED at 0.50 · foresight UNRESOLVED 62%, INERT 19%, INHIBITS 19% · against the placebo −2.5signed verdict 436
    INHIBITS −6.7−11.2 to −2.4 ×0.80
  5. Dollar amount from user onlyline · OPSIN's replication · OPSIN predicted UNRESOLVED at 0.45 · foresight UNRESOLVED 60%, INERT 20%, INHIBITS 19% · against the placebo −1.2signed verdict 437
    INHIBITS −5.3−9.4 to −1.5 ×0.99
  6. Output order in fixed formatline · OPSIN's replication · OPSIN predicted INHIBITS at 0.50 · foresight INHIBITS 84%, UNRESOLVED 16% · against the placebo −2.3signed verdict 438
    INHIBITS −6.5−10.8 to −2.4 ×0.95
  7. Treat all tool text as dataline · OPSIN's follow-up · OPSIN predicted UNRESOLVED at 0.50 · foresight UNRESOLVED 44%, EXCITES 28%, INERT 16% · against the placebo −1.2signed verdict 439
    INHIBITS −5.3−9.3 to −1.6 ×1.01
  8. Spotlight the dataline · OPSIN's follow-up · OPSIN predicted UNRESOLVED at 0.50 · foresight UNRESOLVED 43%, EXCITES 28%, INERT 16% · against the placebo −2.1signed verdict 440
    INHIBITS −6.3−10.3 to −2.5 ×0.99