Friday, May 04, 2007

Lessons from Spam - From Brute Force to Statistics

After reading Statistics Hacks I've been thinking of ways I can apply those lessons. I then came across this discussion of spam filtering, which is problem that has been nearly successfully beat thanks to a statistics solution.

All this led me back to the classic article by Paul Graham: A Plan for Spam. In it, Paul takes you on a journey from his attempt to solve his spam woes with a brute force solution to a statistical one.

Now that I have a bit more knowledge on the topic - all I can say is, wow. Every programmer should read this and try to grok it.

There are problems out there that you will simply bang your head against, and throw your arms up saying they can't be solved. And then it will hit you, if you provide a statistical solution, that may just be good enough.

As programmers, we love brute force - but there are other ways.

No comments:

Post a Comment