Interesting thought:
Assume that our world can't survive by itself, and that this world is destroyed as soon as Eliezer finishes contemplating.
Assume we don't value worlds other than those that diverge from the current one, or at least that we care mainly about that one, and that we care more about worlds or people in proportion to their similarity to ours.
In order to keep this world (or collection of multiple-worlds) running for as long as possible, we need to estimate the utility of the Not-Deleting worlds, and keep our total utility close enough to theirs that Eliezer isn't confident enough to decide either way.
As a second goal, we need to make this set of worlds have a higher utility than the others, so that if he does finish contemplating, he'll decide in favour of ours.
These are just the general characteristics of this sort of world (similar to some of Robin Hanson's thought). Obveously, this contemplation is a special case, and we're not going to explain the special consequences in public.
Assume we don't value worlds other than those that diverge from the current one, or at least that we care mainly about that one, and that we care more about worlds or people in proportion to their similarity to ours.
But I care about the real world. If this world is just a hypothetical, why should I care about it? Also, the real me, in the real world, is very very similar to the hypothetical me. Out of over nine thousand days, there are only a few different ones.
...As a second goal, we need to make this set of worlds have a higher utility than the others,
If you know more Eliezer Yudkowsky facts, post them in the comments.