Several years ago, I had an idea to use Stanley
Milgram's ideas on social distance (see his Small
World experiment for an example) to help fight
spam.
The idea is a simple one: Spam is a relative term. One
person's spam is another person's valid email. Similar
groups of people will have similar opinions on spam.
And just as you trust your friends' opinions, and base
that trust on their similarity to your own and a basis
in fact, you can do the same with spam.
I spent about a year tooling around with the concept on
and off in spare moments, and then spent a fair amount
of effort getting a proof of concept working, getting
the community to notice it, and building momentum
behind the idea. Sadly, I took a job that required me
to stop development on the project, so stop I did.
This isn't the place to explore the entire concept in
depth. Instead, head on over to Sufficiently
Advanced Technologies, Bitshift's sister-site of
sorts, and read all about it, as well as download
working code.