Alignment Workspace

updateless-defers-relation

leaf/step · drafted

Statement

An endorsement-like relation between a prior and a candidate update assigns deference in transparent Newcomb, counterfactual mugging and the twin prisoner's dilemma exactly where the policy-optimal choice differs from the updateful one, and is not definitionally equivalent to policy optimality.

Source

projects/deference/note-dump-2026-08-11/deference-trust-lab/models/updateless-deference-model.md, section 0, the candidate relation; projects/deference/note-dump-2026-08-11/deference-trust-lab/report/CRITIQUE.md, the lab's own blocker rating the direction near-vacuous; projects/deference/faf-cleanroom-2026-10-02/packages/udt-endorse-policy/udt-endorse-policy-report.md: defersAll_iff_policyEndorses is a squeeze and defersMenu_not_policyEndorses_witness refutes the menu form.

Notes

Two independent AI passes, the trust lab's critic and the clean-room's squeeze, found every candidate tautological: a tautology wrapper around argmax with a hand-fed discriminator, in the critic's words. The maintainer's framing in the public agenda, projects/deference/note-dump-2026-08-11/deference-trust-lab/AGENDA.md ('There are real improvements which UDT will accept'), is why the task stands.

Author and human events

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

No human event has touched this node.

Edges

Above

Ranking (the lens, not the structure)
{
 "status": "open",
 "diagnosis": null,
 "flags": [
  "AI number"
 ],
 "V": {
  "V_prior": null,
  "display": "2",
  "headline": 2,
  "ref": 2,
  "valid": 2,
  "weaker": "both"
 },
 "belief": {
  "J": 0,
  "mid": "7/20",
  "range": [
   "0/1",
   "17/20"
  ],
  "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.

comment · challenge · estimate · want