[ExI] for the fermi paradox fans

[Anders Sandberg]

BillK <pharos at gmail.com> , 7/6/2014 9:03 AM:

That is a nice idea, but it doesn't sound right to me. 
If an intelligence is running in computronium at high clock speeds, 
effectively the outside world freezes. I can't see them switching off 
till the stars die. To them, that would be switching off for eternity. 
Actually not. From their perspective, switching off for a nanosecond or an eon will be subjectively equivalent.
A lot rather hinges on the utility function: what do you want to do? If the goal is maximizing fun per subjective moment for each conscious thread you just need resources proportional to the number of threads and the time they remain. If the goal is having maximum amount of fun, then you want more threads - up to the limit set by the available resources. So if you have total computational resources R, N threads and they require k resources per subjective moment, then you get R/kN moments in the first case and in the second you get R/k thread-moments (how they are distributed matters less). In both cases you would be better off with R larger. Now, the amount of computations R you can do at temperature T scales as 1/T. If we do it today, R=(1/3)*E0. If we do it in a zillion years when we get close to 10^-19 K (the limiting value), then R=10^19 E0. So it makes sense to wait. 
I think you assume a steep time discount factor: the value of the future at subjective time t (V(t)) to now is exp(-Dt)*V(t), where D is a measure of how steep the discount is. If you start with energy E0 and use it at some rate r, if E0/r >> D you will not care about the later subjective times. Jumping to the future means E0 is essentially multiplied by a lot. In the case of threads maximizing their own utility this has no effect: getting a lot more E0 will not improve things for them. The case where you want to have a lot of utility and discount it then it is rational for you to up r so that E0/r is about D: it would be rational to jump to the future, even though you would then burn through all the energy super-quickly. The main limit may be the physical limits on how large r can become: this, together with D, will set the desired distance into the future you wait. 
So long-discount civilizations (like ones that want to maximize the total amount of thinking across history) would jump to the cold era. Individualistic fast civilizations will not jump, but will presumably burn out ("Oh, my star's mass energy is running out... but it would take *minutes* to get more matter. C'est la vie..."). Collectivistic short-discount civilizations will jump a bit into the future so they can burn mass-energy at an intermediate rate. 
Note that this means that long-discount civilizations have a reason to limit the impact of short-discount civilizations, especially the collectivist ones. Either stop them outright, convince them to jump to the far era (short-term civs are indifferent to this, if they trust the long-term ones), or make them burn out fast without doing damage. Maybe this is Nyarlathotep's job...

"Beyond the worlds vague ghosts of monstrous things; half-seen columns of unsanctified temples that rest on nameless rocks beneath space and reach up to dizzy vacua above the spheres of light and darkness. And through this revolting graveyard of the universe the muffled, maddening beating of drums, and thin, monotonous whine of blasphemous flutes from inconceivable, unlighted chambers beyond Time; the detestable pounding and piping whereunto dance slowly, awkwardly, and absurdly the gigantic, tenebrous ultimate gods—the blind, voiceless, mindless gargoyles whose soul is Nyarlathotep."

Anders Sandberg, Future of Humanity Institute Philosophy Faculty of Oxford University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20140607/c4c2118c/attachment.html>