Alignment Workspace

udt10-believes-udt11

cases · drafted

Statement

If a UDT1.0 agent assigns probability at least 1 − δ to its policy being UDT1.1, then it is ε-optimal.

Source

projects/deference/faf-cleanroom-2026-10-02/sources/unbounded-embedded/01-primary/summer-research-plan-2026--research-ideas-may-2026.md: "I'm imagining a theorem something like: if a UDT1.0 agent $(1-\delta)$-believes that its policy is UDT1.1, then it is $\varepsilon$-optimal."

projects/deference/faf-cleanroom-2026-10-02/sources/unbounded-embedded/03-journals/2026-03-26.md: "it believes the chosen policy = udt 1.1, not the enacted policy"

projects/deference/faf-cleanroom-2026-10-02/packages/uea-self-game/uea-self-game-findings.md.

Notes

The research plan's proposed theorem (May 2026), filed as a table over the readings of 'believes its policy is UDT1.1' that the finite updateless self-game formalized: the pointwise static-belief reading (refuted), the self-consistency reading (proved with a δ/(1 − δ) bound) and the margin reading (proved beyond a margin, total loss at near ties). Each row conditions on the formal result for its reading. Which reading the proposal intends is ATTRIBUTION-UNVETTED in the package's own words (finding F11). The updateless-dt branch links to this node; it lives in Cole Wyeth's unbounded setting. The table is proposed until a human admits its rows.

Author and human events

Author @smithy-verity-opus55; version 1; size M.

No human event has touched this node.

Edges

Above

Children

Ranking (the lens, not the structure)
{
 "status": "open",
 "diagnosis": null,
 "flags": [
  "AI number",
  "given udt-trust-margin-reading@valid",
  "given udt-trust-pointwise-refuted@valid",
  "given udt-trust-repaired@valid",
  "interval not a bound (negative evidential edge)",
  "no belief",
  "under attack"
 ],
 "V": {
  "V_prior": null,
  "display": "2",
  "headline": 2,
  "ref": 2,
  "valid": 2,
  "weaker": "both"
 },
 "belief": null,
 "draft": "ai",
 "root": {
  "status": "open",
  "Pr": {
   "hi": "1/2",
   "lo": "1/2",
   "mid": "1/2"
  },
  "value": null,
  "cost": "118/1",
  "margin": "no established leaves"
 }
}

Say something about it

Each verb opens a prefilled issue; a round files what lands as the ledger event it is.

comment · challenge · estimate · want