decision-theory

frame · drafted

Question

Which decision theory for computationally uncertain agents, on the updateless side and in the synthesis of logical induction with bounded inductive rationality, is tractable, and which of its results bear on AI safety engineering?

Owners: @abram (GitHub abramdemski)

Path: the-program › decision-theory

Roots (the frame's goals)

plan

updateless-dt-programme frame drafted sub-frame

A research programme on the updateless side develops decision theory for computationally uncertain agents, counterfactuals, computational updatelessness, decision problems, open-minded updatelessness and geometric rationality, far enough to settle whether it is tractable and relevant to AI safety engineering, and unfolds in full only where both hold.

Sub-frames

updateless-dt-programme: Which decision theory for computationally uncertain agents on the updateless side is tractable, and which of its results bear on AI safety engineering?

Statement

A decision theory for computationally uncertain agents, developed on the updateless side and as a synthesis of logical induction with bounded inductive rationality against a stated acceptance statement, settles whether it is tractable and relevant to AI safety engineering.

Source

projects/decision-theory/README.md; projects/decision-theory/rounds/2026-09-29-li-bria-synthesis-spec/PROBLEM_STATEMENT.md; DECISIONS.md, the 2026-09-29 decision-theory entries (the line is a top-level project; updatelessness is an ideal extension).

Notes

The decision-theory line (ruling M2): the updateless programme as filed before this round; both of the decision-theory project's work orders (items 105 and 106) already had nodes inside it. No wiki page narrates this line; wiki/Continuation-BRIA.md narrates the learning layer the deference line keeps.

Author and human events

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

No human event has touched this node.

Edges

Above

part of the-program
root (plan) of the-program

Children

part: updateless-dt-programme frame

Work pointers

decision the map is the canonical index of the program: two maintainer handles, one plan with seven lines, programmes drafted from the wiki, the map as the foundational layer, multi-owner frames, `PRIORITIES.md` as a render, decision anchors, and the structure-only index (M1–M8)

Ranking (the lens, not the structure)

{
 "status": "open",
 "diagnosis": null,
 "flags": [
  "interval not a bound (negative evidential edge)"
 ],
 "V": {
  "V_prior": null,
  "display": "2",
  "headline": 2,
  "ref": 2,
  "valid": 2,
  "weaker": "both"
 },
 "belief": null,
 "draft": "ai",
 "root": {
  "status": "open",
  "Pr": {
   "hi": "1/1",
   "lo": "7960389/80000000",
   "mid": "21262296439/25600000000"
  },
  "value": {
   "hi": "1/1",
   "lo": "-1/1",
   "mid": "21262296439/25600000000"
  },
  "cost": "85/1",
  "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