On Saturday, January 11, 2003, at 03:47 PM, Bill Stewart wrote:
....
- A distributed computing like this needs several parts:
- A problem to solve - they seem to keep waffling on this;
their FAQ really needs to be upfront about it,
but it only talks about RSA-576, while their forum
says they are or aren't also doing something with
X-Box,
depending on their legal worries, but doesn't say what
they're trying to do to it (Cracking a 2048-bit RSA key
certainly isn't a rational problem to solve,
but maybe they're trying to crack something else about
it,
like a passphrase used for a key file?)
If neither is solvable in the lifetime of the earth, does it matter
which one they claim to be working on?
- Some way to hand out work and collect results,
and it's possible that they've done this well,
though I doubt they scale to seti.org sizes.
Although, as simple calculations show (reported here several times over
the past decade), random and overlapping self-apportionment of keyspace
to search is only a factor of 38% or so worse than more careful,
non-overlapping apportionment is. (And random apportionment stops the
attack where someone finds the solution, or knows where it is and
claims that portion of the keyspace to search, and then doesn't
announce a solution.)
- A way to split up that algorithm into manageable pieces.
Well, it sounds like their current algorithm has that
:-)
You mean like "All computers on planets circling stars in the Local
Group will work for the next billion years on the following one
trillionth piece of the keyspace."?
(Minor note: Some of those projects are charity-donation things,
where you click on the page and their sponsor shows you a logo
in return for donating to their page. The Landmine Clearing one
seems like a good politically correct thing to do -
Reason enough _not_ to participate.
--Tim May
