Does “Ultimate Neartermism” via Eternal Inflation dominate Longtermism in expectation?
Epistemic status:
While I believe this idea’s consequences could be profound if correct, and it should be taken slightly more seriously than April Fool’s post “Ultra-Near-Termism”, I consider it mostly a quirky novelty, I’ve spent relatively little time thinking about it and suspect it may have major flaws. At the very least, I would hate for it to be taken too seriously before more serious investigation, and would really appreciate anyone pointing out flaws in reasoning that would clearly invalidate it!
TL;DR
If we give the prominent “eternal inflation theory” of cosmology and “evidential decision theory” a non-zero credence, then what is morally important may be what we are able to influence in the smallest unit of affectable time in the future. This would have a lot of weird implications, but it seems there are reasonable, if also weird, arguments that may defeat it.
Eternal Inflation and the Youngness Paradox
In this video, between video start and 11:40, Matt O’Dowd explains Alan Guth’s “Youngness Paradox,” that in each second, due to “Eternal Inflation”, a leading cosmological theory, every second another 10^10^34 universes are born MORE than were born in the previous second, etc., eternally, meaning that every second the number of universes increases by a factor of 10^10^34 from how many existed in the previous second.
Based on a comment on the EA forum, to be clear, this is different from quantum multiverse splitting, as this splitting happens just prior to the Big Bang itself, causing the Big Bang to occur, essentially causing new, distinct bubble universes to form which are completely physically separate from each other, with it being impossible to causally influence any of the younger universes using any known physics as far as I am aware.
Evidential Ultimate Neartermism
Anthropically/evidentially, this means that at any point in time, the sum of all younger universes carry exponentially far more weight than older universes, and almost all intelligent life that exists is that which is youngest across all universes; therefore, if we are trying to maximize the amount of good across all universes, what we should evidentially care about is what is happening soonest in time across all universes, including our own, as the weight of what happens across all universes soonest in time is always mind-boggling-ly more weight-y and hence valuable than what happens later in time. This, If correct, swamps longtermist arguments by a very, very large factor, and does so in expectation even if we give this theory and evidential decision theory an absurdly low credence.
If we follow this line of argument, then what we should altruistically hope for is that in every universe, everyone in that universe is trying to do whatever they can to maximize value in the next smallest affectable unit of time, meaning that there is at least some possibility that the morally correct thing to do is to try to maximize value (most likely for yourself, as it seems difficult to affect others as rapidly) in the next ~millisecond, and to be doing this at all times, hence,”Ultimate Neartermism.”
Possible escape routes and implications
To be clear, I think that the same arguments that I suggest defeat Pascal’s Mugging and actually support longtermism, also defeat arguments against blindly accepting this idea. Namely, we should give a non-zero credence that we might eventually figure out how to create a perpetual motion machine, or other ways that we might create infinite value, such as if we accept a cyclic universe or Penrose’s Conformal Cyclic Cosmology model, or various infinite multiverses or many other weird anthropic or highly speculative theoretical scenarios (although, I can’t say for certain any of these aren’t also subject to the same youngness paradoxes due to different sizes of infinity). Another weird possibility, for example, is that even if we give an absurdly low credence that we might be able to influence all of the younger universes being created by eternal inflation, then perhaps that is what we should actually try to pursue, increasing the goodness of the exponentially huge number of future younger universes.
It seems to me that we should give this theory at least some credence, but its consequences are bizarre enough and potentially horrible enough from other perspectives that it seems worth thinking about it carefully rather than blindly accepting it. Doing further research on this idea would, of course, mean that we are losing a factor of 10^10^34 orders of magnitude worth of value of the universe every second this research takes, however, because this loss of value always continues, if the theory is correct, doing the research is still worth incalculably more than other courses of action which lead us to completely neglect this idea.
Interestingly, this means it is of comparatively infinitesimally little value researching how to maximize value in the next millisecond. You lose so much value by doing so, that even if you have a positive but practically nonexistent credence that you may increase your next millisecond value by a virtually negligible amount, this value exceeds the value of researching how to best increase next millisecond value by a factor many, many, many times the value difference between the value of this research compared to the entire value of the future of our universe.
Furthermore, one of the most bizarre consequences of this idea is that it is extremely anti-memetic, i.e., if you accept it and are altruistic, then immediately you should start trying to maximize value in the following millisecond and continue doing so as long as you are able, which means you should put zero effort into spreading this theory, as plans that far in the future have virtually zero value compared to the evidential impact beings like you will have in the proceeding millisecond. Due to this fact, and the unappealing nature of this idea, it seems unlikely it will catch on, unfortunately even if correct.
If a universe splits into many universes, most of them will be very similar. So the action I do now has an impact on all of them, but the action I do a little later still happens in most of them, so it has an impact on most of them. It does not make a big difference whether I do it now or a little later.