A model of UDT with a halting oracle searches only for one utility value for each action. I’m guessing the other formulation just wasn’t obvious at the time? (I don’t remember realizing the possibility of playing chicken implicitly before Will Sawin advertised it to me, though I think he attributed it to you.)
A model of UDT with a halting oracle searches only for one utility value for each action. I’m guessing the other formulation just wasn’t obvious at the time? (I don’t remember realizing the possibility of playing chicken implicitly before Will Sawin advertised it to me, though I think he attributed it to you.)