somebody wrote:
> Drugs will be priced "accordingly" (i.e. outrageously) no matter what.

That's kind of my problem with it. :-)

> Often, it's the case that the cost function is asymmetrical, i.e. it
> may take 20 CPU years to find a solution, a millisecond to verify it. And
> those are the type of problems most suitable for distributed computing
> anyway.

For searches (like SETI, say), it's easiest that when you get a hit, you 
farm the same raw data out to a dozen other people and see if you get the 
same answer back. If not, you stop serving jobs to the guy giving you crap 
answers. If you really need all the answers from all the parts, you do the 
Byzantine General bit. It's a solved problem that just takes more overhead.

-- 
   Darren New, San Diego CA, USA (PST)
   My fortune cookie said, "You will soon be
   unable to read this, even at arm's length."

Post a reply to this message