[ExI] 23andSingularity

Anders Sandberg anders at aleph.se
Sun Jul 7 13:57:55 UTC 2013


On 2013-07-07 05:54, spike wrote:
> It occurred to me that we could program something like a will to figure out
> an enormous but very specific question: how are all the carbon units on
> 23andMe related?

Sounds doable. I think existing phylogenetics algorithms/software almost 
should be able to do it, but the extra info about humans (names etc.) 
would help the process even more. The end result is not a guaranteed 
family table, but a probabilistic best match.

I think some of the sleuthing is slightly nontrivial to automate, but 
even the trivial part is enough to provide extra information to help 
narrow things down. Essentially it is a giant sensor fusion problem.

> It would represent a kind of nano-singularity, a
> tiny slice of human existence in which code came along and was taught our
> ways of finding these sorts of things, then it just started doing it and
> accumulating more and more data, with error checking and verification, and
> with each verified link, the system became stronger, since it can now use
> those links to find others, until one day it surpasses every human.  Then in
> that one tiny nano-slice of life, we could claim that a singularity of sorts
> has occurred.

At least a phase transition. Perhaps like the one that happened among us 
in the 90s as we became globally connected (and is still rolling 
outwards - those Tajik goat-herders showing off their smartphones are 
not the last generation). It is very much like percolation: as the 
probabilitiy of nodes being conencted increases, at some point a giant 
component emerges where most of the nodes are linked.


-- 
Dr Anders Sandberg
Future of Humanity Institute
Oxford Martin School
Oxford University




More information about the extropy-chat mailing list