Comment by Marc Gotliboym (marc-gotliboym) on Embedded Agency (full-text version) · 2019-03-17T12:49:50.530Z · LW · GW

Concerning the 5 and 10 problem - I'm curious if any works been done trying to resolve this by using a weaker logic? I'm not a logician, but a relevance logic seems worth looking into. At least on the face of it, taking away the principle of explosion is a step towards making the mentioned P, “if the agent outputs 5 the universe outputs 5, and if the agent outputs 10 the universe outputs 0” unprovable.

I'd be interested in any other work on the 5 and 10 problem also.