Creating A Centralised, Open Anti-spam Service

It occurs to me that such a thing would be very useful under the right circumstances. It would essentially be directory of known spam sources that anyone could refer to. Some notes:

• Follow the phishtank model of allowing anyone to partake in the confirmation process and trusting users based on performance compared to other trusted users
• Provide service for email and comment spam as well as phish URLs
• Needs a clean API to enable many applications to use it
• Needs resilience from attack
• Push option as well as the normal pull methods
  

Sharing opinions on whether an entity, e.g. an IP Address, is a source of spam could give people the choice of blocking them automatically and before they come into contact with that entity at all. For instance, if a blog commenter is considered by enough people to be a spammer you could block their IP Address automatically, before they even post on your blog.

I need to think this through and ask around for better ideas. If you have any, or want to help, let me know.