approximate-decision-determination
and · drafted
Statement
Accuracy delta of the environment's predictor gives the policy utility up to a bound linear in delta in the prediction-dependent regime, and decision-determination up to delta in the information sense gives the policy utility up to the square root of half delta on average.
Source
projects/deference/faf-cleanroom-2026-10-02/packages/udt-condense-dd/udt-condense-dd-report.md: approxDD_entropy, approxDD_varying, approxDD_fixed and the accuracy bounds.
Notes
The one reading of Gap 1 with content, proved; conditional on the own-law reading of the accuracy hypothesis, which the open hypothesis leaf carries together with the status-note rewrite.
Bridge card
A conjunct of the informal claim derivation-via-approximate-dd: The approximate decision-determination bridge, with the policy utility bounded by the predictor's accuracy, derives updateless reasoning from the agency axioms in the sense the representation theorem seeks.
Declaration approxDD_entropy (record approximate-decision-determination.record)
The declaration's information-sense decision-determination and its utility bound are the approximate form of the status note's Gap 1 (a perfect condensation with an optimal predictor factors through the policy), read with the environment's accuracy under the agent's own law.
Hypotheses, read:
- own-law-accuracy-reading open: The predictor's accuracy is measured under the agent's own law, and the status note's Gap 1 is read in its approximate form.
Completeness (approximate-decision-determination.complete): No hypothesis beyond the own-law accuracy reading is needed for the approximate bounds.
Author and human events
Author @smithy-verity; version 1; size M.
No human event has touched this node.
Edges
Above
- part of derivation-via-approximate-dd
Children
- part: approximate-decision-determination.record leaf/record
- part: approximate-decision-determination.bridge leaf/bridge
- part: own-law-accuracy-reading leaf/hypothesis
- part: approximate-decision-determination.complete leaf/completeness
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": null,
"draft": "ai"
}Say something about it
Each verb opens a prefilled issue; a round files what lands as the ledger event it is.