From: Robert J. Bradbury (bradbury@aeiveos.com)
Date: Thu Jul 13 2000 - 17:48:29 MDT
On Thu, 13 Jul 2000 hal@finney.org wrote:
(I agree that Brent is rapidly skipping from Robin's "simple" uploads
to full sensory interleaving without a huge amount of work on exactly
how this would be accomplished. IMO, that is the *really* hard part --
mapping individual neural nets onto a common experience message exchange
format.)
> BTW here's another legal conundrum. If the minimum wage is $5.50 /
> hour, and an upload runs 1000 times faster than a normal human, then
> should he get $5,500 / hour minimum?
I doubt the initial ones will. Even if Blue Gene has a petaflops or
10 or even 100, until the abstractions Robin discusses are developed
a "copyload" is going to run *slow*. That is why I don't think
there will suddenly be the thousands of copies Robin proposes --
you have to build real large (initially non-nanotech) computers
on which to run them and those computers don't come cheap.
If you still need 2000 sq ft of floor space (initial Blue Gene
estimates), it sounds like you still have a data center and
its associated costs (electricty, air conditioning, security,
etc.). If the published estimates of Blue Gene time (~1 year to
solve the folding problem for a typical 300 amino acid protein, see:
http://www.research.ibm.com/news/detail/protein_fact.html
are accurate), then you are going to need *significant* hardware
progress *and* some nifty compilation of the brain's processes
to run the brain at "real time" rates.
I can see the advertisements now --
"Upload now -- In a single minute you can watch the rest of
the world grow 50 years older..."
(An interesting related "trivia" question would be whether
the cost/"volume" of computer hardware has declined significantly
over the last 40 years.)
Robert
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:29:55 MST