The Infinity Machine [entries|reading|network|archive]
simont

[ userinfo | dreamwidth userinfo ]
[ archive | journal archive ]

Fri 2008-04-18 09:52
The Infinity Machine
LinkReply
[identity profile] feanelwa.livejournal.comFri 2008-04-18 10:11
As a naive barely-programmer, I assume this is possible - I would look for an algorithm to without fail identify spam from genuine emails, and when I was kicked out I would give it free of charge to every email provider in the world.

I would also run everybody's multislice simulations, starting with my several-billion-atom one that would otherwise need silly amounts of breaking up to do a bit at a time on CamGRID.
Link Reply to this | Thread
[personal profile] simontFri 2008-04-18 10:18
I fear the spam problem is another AI-complete one. You could search for a program which correctly classified everything in a ginormous input corpus, but it could very easily turn out to be distinguishing on the wrong attributes and fail the first time it saw a mail that wasn't part of the corpus. Also, the problem with spam is that it evolves; once the algorithm was running on all mail systems, spammers would have a copy to test against and could hammer on it looking for a way through...
Link Reply to this | Parent
[identity profile] azekeil.livejournal.comFri 2008-04-18 22:42
Spam is also a problem of definition - what is spam to one person is not necessarily spam to another.
Link Reply to this | Parent
navigation
[ go | Previous Entry | Next Entry ]
[ add | to Memories ]