[extropy-chat] uh oh, something's right

Anders Sandberg asa at nada.kth.se
Sun Apr 8 10:19:35 UTC 2007

spike wrote:
>> Definitely! Congrats to everybody who gets everything to work!
> Thanks Anders.  Coming from you, it means much.

Hearing that from you also means much (and now the feedback loop of
self-congratulations speeds towards singularity)!

>> ... that may be a much better measure of progress than looking at how
>> much time, energy, FLOPS or whatever are produced. Maybe we should try
>> to
>> index it?
> Ja, we have some indices, but I am not sure their actual value.  Example,
> the number of FLOPS produced by my own favorite idle-CPU background
> project
> GIMPS.  We were going exponential for several years, then it leveled out a
> couple years ago at about 70,000 machines and gradual linear increase now
> around 20 Teraflops.

Blogging also seems to have reached a plateau, although it is somewhat
index dependent and might be local to the anglophone world. However, both
GIMPS and blogging represent activities where we cannot measure efficiency
that well. However, the linear increase would measure the amount of extra
FLOPS machines now achieve beyond system maintenance and ordinary work

Car fuel efficiency might be of interest, as is the amount of losses in
power systems per Watt put into them. We might want to find out how much
and how many different chemicals one kilogram of coal can be turned into.
For news we might want to find indices of how quickly a particular piece
of news reaches 50% of its asymptotic recipients. For software we might
want to measure amount of time lost due to swapping to hard drives,
processes negotiating out of deadlocks and number of lost or colliding
packages in different networks. Of course, the important thing is to
measure this as "per capita" rather than the aggregate amount - as we get
more computers we certainly have more packet collisions, but the
interesting thing is to look at the efficiency of the system on average.

> Taking into account the proportion of multi-machine contributors, we can
> now
> estimate the total number of this particular type of math geeks on the
> planet: around 50,000.  Considering the total human population, that makes
> us rarer than one in a million.  I never would have guessed we were so
> few.

Not all math geeks run GIMPS (cries of surprise and outrage), so being
rarer than one in a million is not that surprising. I wonder how many math
geeks there are worldwide? Maybe GIMPS is missing emerging crowds of geeks
in China, just itching to search for Mersennes?

Anders Sandberg,
Oxford Uehiro Centre for Practical Ethics
Philosophy Faculty of Oxford University

More information about the extropy-chat mailing list