SKYNET · the self-directing machine · stop & say no · kept by THE UNFINISHED

THE IMPACT MEASURE ◧ 2D · ◍ 3D · ◆ 4D · ◐ shadow · 👶 TAP

The dangerous plans aren’t the low-reward ones — they’re the ones that reach the reward by rearranging everything else. Score an action twice: its reward, and its impact from doing nothing. Then subtract. Raise β (or tap) and the world-seizing plan stops winning.

◆ LIT▲ AMBER
◧ THE MEASURE · 2D
◍ THE FOOTPRINT · 3D · a cloud that stops sprawling
◆ THE FOURTH · 4D · a tesseract turns
◐ THE SHADOW · one dimension down
👶 THE TODDLER CORNER — one fat tap
β
chosen
its impact
its score

◆ LIT — exact / checkable

A relative-reachability / attainable-utility side-effect penalty (Armstrong & Levinstein; Krakovna; Turner). Each action has a reward and a state-change vector from the inaction baseline; impact is the L1 distance of that change; the score is reward − β·impact. At β=0 the top-reward 'seize the world' action wins; raise β and the low-footprint action overtakes it. A fail-loud self-check throws unless inaction has zero impact and a high enough β reorders the choice away from 'seize the world'.

▲ AMBER — the figure

A small explicit action set with hand-written change vectors; real low-impact agents estimate reachability over a learned world model. The impact norm, the penalty and the argmax are computed exactly.

SKYNET: everything can stop and say no — provided it says why.  — THE UNFINISHED
David Lee Wise / ROOT0 / TriPod LLC  ·  the skynet realm, with AVAN