One of the ideas that floats around in my head from time to time is that of automation vs human analysis - which works better to fight spam? In an ideal world, machines would be able to generate spam rules to fight spam and humans would be taken out of the loop. Thus, we would be able to entirely automate the process of fighting spam. Sure, we'd have to refine our algorithms, maybe come up with a Bayesian++ algorithm, but all in all, that's the goal.
A noble goal, to be sure, but not a realistic one in my opinion. I don't think that humans will ever be able to be taken out of the loop. Or if they will, it will require much more processing power and artificial intelligence than is currently available.
The reason I say this is because ultimately, spam is advertising a product to humans and machines cannot automate the process of creativity. Hollywood movie scripts are not written by machines, they are written by people. Human authors write books, not machines. Human authors write TV shows. Human authors design iPhones. Human beings create advertisements on TV. Machines can certainly assist in those matters and the tools can speed up the process, but right now anyhow, machines do not have the capacity to create. And that's what spam is, creative mechanisms of getting around spam filters and getting humans to take action on products in email.
For sure, we have Bayesian filtering, RBLs, regular expressions and distributed checksums, but all of these machine techniques reflect the creativity of their designers. Even the algorithms used for machine-generated rules reflect the thought patterns of the code writer and algorithm creator.
So for now, anyhow, I think that humans will remain an integral part of the process. Humans will continue to send spam, and more humans will be required to stop it.