The "infinitely so" part seems wrong, but the idea is that 4D histories which include a sentient being coming into existence, and then dying, are dispreferred to 4D world-histories in which that sentient being continues. Since the latter type of such histories may not be available, we specify that continuing for a billion years and then halting is greatly preferable to continuing for 10 years then halting. Our degree of preference for such is substantially greater than the degree to which we feel morally obligated to create more people, especially people who shall themselves be doomed to short lives.
Not sure what part of this comment tree this belongs so just posting it here where it's likely to be seen:
It struck me with an image that it's not at all necessary that these tradeoffs are actually a thing once you dissolve the "person" abstraction; it's possible that something like the following is optimal: half the universe is dedicated to search the space of all experiences in order starting with the highest utility/most meaningful/lowest hanging fruit. This is then aggregated and metadata added and sent to the other half which is tiled with m...
If it's worth saying, but not worth its own post (even in Discussion), then it goes here.