legit-rule-learning-model
leaf/route · drafted
Statement
A model of corrigibility rules learned from legitimate-branch feedback measures the selection bias the ex-ante conditioning result predicts, and rules certified at commitment escape it.
Source
The filing session's reading of the ex-ante weighting and selection results in projects/deference/faf-cleanroom-2026-10-02/packages/legit-neg-static/legit-neg-static-report.md (A3, A6) and legit-neg-pricing beside it: rules learned from legitimate-branch feedback inherit the selection the conditioning results describe, and rules fixed at commitment do not.
Author and human events
Author @smithy-verity; version 1; size M.
No human event has touched this node.
Edges
Above
- given of legit-objective-proposals
- part of legit-weighted-objective-improvable
- given legit-objective-proposals
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [
"AI number",
"given legit-objective-proposals"
],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": {
"J": 0,
"mid": "13/20",
"range": [
"3/20",
"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.