Omega’s decision is based on our decision algorithm itself.
Yes, but this dependence factors through the strategy that the algorithm produces. Read Eliezer’s TDT document(pdf) where he talks about ‘action-determined’ and ‘decision-determined’ problems. Whereas CDT only ‘wins’ at the former, TDT ‘wins’ at both. Note that in a decision-determined problem, Omega is allowed but a TDT-minimizer is not.
I argue that an EDT agent should integrate his own decision as evidence.
You appear to mean: When an EDT agent hypothesizes “suppose my decision were X” it’s subsequently allowed to say “so in this hypothetical I’ll actually do Y instead.”
But that’s not how EDT works—your modification amounts to a totally different algorithm, which you’ve conveniently named “EDT”.
If EDT’s decision is to one-box, then Omega’s prediction is that EDT one box, and EDT should two-box.
...then Omega’s prediction is that EDT will two-box and oops—goodbye prize.
But that’s not how EDT works—your modification amounts to a totally different algorithm, which you’ve conveniently named “EDT”.
EDT measures expected value after the action has been taken, but the output of EDT has no reason to be ignored by EDT if it is relevant to the calculation.
...then Omega’s prediction is that EDT will two-box and oops—goodbye prize.
It loses, but it is generally claimed that EDT one boxes.
Yes, but this dependence factors through the strategy that the algorithm produces. Read Eliezer’s TDT document(pdf) where he talks about ‘action-determined’ and ‘decision-determined’ problems. Whereas CDT only ‘wins’ at the former, TDT ‘wins’ at both. Note that in a decision-determined problem, Omega is allowed but a TDT-minimizer is not.
You appear to mean: When an EDT agent hypothesizes “suppose my decision were X” it’s subsequently allowed to say “so in this hypothetical I’ll actually do Y instead.”
But that’s not how EDT works—your modification amounts to a totally different algorithm, which you’ve conveniently named “EDT”.
...then Omega’s prediction is that EDT will two-box and oops—goodbye prize.
EDT measures expected value after the action has been taken, but the output of EDT has no reason to be ignored by EDT if it is relevant to the calculation.
It loses, but it is generally claimed that EDT one boxes.