deck-tt-implies-value-regular
leaf/step · drafted
Statement
Soft Total Trust of the human inductor H in the AI inductor A, with A's chooser satisfying REGULARITY on an efficiently describable menu sequence, implies E^A_n(a*_n − a^i_n) ≳_n 0 for every option a^i_n and hence Value on that sequence.
Source
projects/deference/faf-cleanroom-2026-10-02/sources/Boundedly Rational Deference.pdf, slides 9 and 10 ("When Does TT ⟹ VALUE?": "So we just need E^A_n(a*_n − a^i_n) ≳_n 0, to establish VALUE"; the counterexample, then "We can rule this out by assuming: REGULARITY"), and slide 19 (Appendix: "TT implies E^A_n(V_n) ≳_n v ⇒ E^H_n(V_n) ≳_n v"). Mathematics transcribed from the handwritten slides.
projects/deference/note-dump-2026-08-11/wiki/total-trust-implies-value.md, the Theorem: "Total Trust + introspection + conditional-stability ⟹ argmax Value".
Notes
The deck's first arrow. Its evidence parent is the Lean-checked scoped theorem read as an implication (li-deference-collapse-scoped@valid), which proves the same shape with conditional-stability H3 as the scope condition; whether the deck's REGULARITY is H3 is the question regularity-is-h3.
The clean room's audit of the slides (def-lattice-arrows findings F3) records that slide 19 writes s − ε for v − ε, a typo that changes no step.
Author and human events
Author @smithy-verity-opus55; version 1; size S.
No human event has touched this node.
Edges
Above
- part of deck-collapse-statement
- given of li-deference-collapse-scoped
- given li-deference-collapse-scoped (as valid)
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [
"AI number",
"given li-deference-collapse-scoped@valid"
],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": {
"J": 0,
"mid": "3/4",
"range": [
"1/4",
"1/1"
],
"width": "1/1"
},
"draft": "ai"
}Say something about it
Each verb opens a prefilled issue; a round files what lands as the ledger event it is.