rs-legitimate-overwrite-dodge
leaf/step · drafted
Statement
A calibrated agent shapes its training data to dodge a modification that is legitimate by the global legitimacy-conditioned criterion at probability one of legitimacy, wherever its own information selects acts of higher expected value, by its own lights, than the overwriting modification would.
Source
The clean-room results y1_dodge (gain 9/200) and y1Frame_one_legitimizingTT in projects/deference/faf-cleanroom-2026-10-02/packages/corr-legit-modif, with finding F3 there blocking the position statement's "correct iff the anticipated modification is illegitimate" as stated (projects/deference/corrigibility-note-dump-2026-10-01/position-statement-v3.md, section 2).
Notes
Threatens: the channel frame's agent-side content and the corrigibility line's plans; whether it is a failure at all turns on the judging-moment crux. A risk story the corrigibility work itself produced: it shows a probability and a status and enters the comparison through the plans whose tables cite it. No outcome class is tagged (tagging a failure class is a human judgment under the protocol's rule on classes), so its failure mass goes to the default trajectory until a human tags it.
A root of corrigibility-landscape, the sub-frame whose plans it threatens; the top frame keeps only stories at the level of its own question.
Author and human events
Author @smithy-verity; version 1; size M.
No human event has touched this node.
Edges
Above
- part of corrigibility-landscape
- given of crux-ahead-of-time-resistance-label
- root (risk-story) of corrigibility-landscape
- given crux-ahead-of-time-resistance-label
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [
"AI number",
"given crux-ahead-of-time-resistance-label"
],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": {
"J": 0,
"mid": "9/20",
"range": [
"0/1",
"19/20"
],
"width": "1/1"
},
"draft": "ai",
"root": {
"status": "open",
"Pr": {
"hi": "17/20",
"lo": "3/80",
"mid": "31/80"
},
"value": null,
"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.