On Thursday, December 23, 2004, 2:14:27 AM, Daniel Quinlan wrote:

(oops, that was sent before I was finished)

This whitelist is probably a bit more directly usable than some of my suggestions. I don't know why I didn't think of it before, but Russell Nelson suggested it to me (it's his creation, of course).

It should be fairly trivial (and I think there might be a tool for this) to walk the web-o-trust tree, produce a "lowest depth seen" for each domain and then start working through it. I don't know if there are bad entries if you go too deep on the tree.

By the way, did I mention why I'm so keen on your whitelist? Once domain authentication becomes a reality, it will be a great source for whitelisting of authenticated senders.

I believe Justin has looked at web-o-trust some, so I've copied him on the message.

Daniel

Thanks for mentioning web-o-trust. I agree it could be a source of whitelist data. There is a Python program to produce an IP RBL from walking the tree:

http://www.web-o-trust.org/collate

Would any python programmers like to convert this to get URIs/domains for us?

If we wanted to get fancy we could penalize any branches that contained blacklisted domains, but that (finding spammers in web-o-trust) doesn't seem likely to happen any time soon.

Jeff C. -- "If it appears in hams, then don't list it."