legit-correction-function-learned
leaf/question · drafted
Statement
The correction function, which says how to read beliefs formed under illegitimate conditions, is determined by the agent's joint beliefs up to the part that lives on the anchor of the legitimacy event, and that remainder is learnable from human feedback.
Source
projects/deference/note-dump-2026-08-11/related-chats/2026-07-05__theory-of-legitimacy__c1ab54d9.md, turn 17, the maintainer: "Uninterpretability is addressed in the model via the corruption function, which tells us how to interpret corrupt beliefs, though again how do we learn that?" and "The correction function can be designated and we can work our way backwards to other beliefs if we want/need to do it that way. But I agree that it is not a free parameter over and above the agent beliefs."
Notes
A question of the maintainer's, stated as the proposition its positive answer would be. Conditioned on the self-case identity: if the correction toward the self is forced to the identity, the learnable part is exactly what lives on the anchor.
Author and human events
Author @smithy-verity; version 1; size M.
No human event has touched this node.
Edges
Above
- given of legit-li-self-correction-identity
- held by legitimacy
- given legit-li-self-correction-identity
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [
"AI number",
"given legit-li-self-correction-identity",
"held by legitimacy"
],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": {
"J": 0,
"mid": "11/20",
"range": [
"1/20",
"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.