[ExI] for the fermi paradox fans

[Anders Sandberg]

Rafal Smigrodzki <rafal.smigrodzki at gmail.com> , 14/6/2014 8:22 AM:
On Fri, Jun 13, 2014 at 5:34 PM, Anders Sandberg <anders at aleph.se> wrote: 
These results imply again that future cold intelligences may have fairly frequent retracing loops, especially since there may be limits to how long a working superposed qubit can be kept in memory.  
### I followed your exchange with Robin with great interest. I recently met an 8-year old boy who became depressed when he found out that our Sun is about to burn out, and the visible universe will enter a cold, dark night soon afterwards. So is there hope for us?
Somewhat. Present evidence suggests that we can only get a finite number of irreversible computations done before entropy wins. But finite here means *a lot*.
 By cold intelligences, do you mean very slow minds operating on solid matter, or computation implemented in loosely bound particles? What are limits on the number of computations than can be performed with the limited free energy within out light-cone, assuming no limit on time? Should we start collecting matter now, impose order on gravitationally bound galaxies, and hoard them for the coming night? Can we go on forever, slowing down ever more yet never stopping our journey through computable states? 
This is why I am making these posts: I want to answer the above questions in a paper I am writing. The pieces I have assembled so far:
* In our "Eternity in six hours" paper Armstrong and me calculates how far we could spread. This gives us the volume (in co-moving coordinates) we can play with. 
* The amount of resources depends a bit on what we can use. Conservatively, stars and other condensed matter, plus black holes. Increasingly less conservatively: baryonic matter in galaxies, dark matter halos, interstellar gas, the total mass-energy of the volume. Here I can use current astronomical estimates.
* How much negentropy that can be extracted from this remains a question mark for me. Help wanted.
* Due to accelerating expansion non-gravitationally bound clusters will get exponentially separated and lose causal contact. I ought to calculate the size of the largest possible remaining structures, and check how much is lost if we move superclusters together to build them.
* The background temperature declines over time towards a limit of the horizon radiation. 
* Given negentropy, available mass-energy and temperature I can calculate the maximum number of irreversible computations/error corrections. There will be many more reversible computations. If we can estimate an error rate we can estimate the reversible-to-irreversible ratio.
* Also, there are time limits set by possible baryon decay, black hoe decay and quantum liquefaction timescales.
Overall, my picture until recently was that solid computational media would have the lowest error rate and hence be ideal, but the potential barrier width argument has made me suspect that very diffuse systems may be better if they can keep together. I also think that hoarding matter is a good strategy, but I really want to see if this means transforming it into some suitable form or just keeping it in place until the dark era - this is of course the Fermi paradox link. 


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/20140614/5f22ccbc/attachment.html>