Well, it looks like most of my ideas got shot down, which just proves that I'm a brilliant visionary and ahead of my time. ;-)
For the record, I wasn't proposing that we "ban" all people who use hotmail and yahoo addresses. I was proposing that we make it a requirement that NEW registrants supply an ISP-based email address. This wouldn't need to be applied retroactively. Moreover, it wouldn't prevent people from continuing to use a hotmail or yahoo address for the purposes of correspondence.
My main point, though, is not about hotmail or yahoo accounts. The central thing I'm trying to say is that there's not much point in having policies regarding behavior if there's no way to enforce them, and there's no way to enforce policies when anyone can use an anonymizer and create a hundred sock-puppet identities for themselves. Some kind of identity verification system is necessary. It should be done in a way that is non-intrusive and respects individual privacy, but SOMETHING needs to happen. Otherwise an enormous amount of decent people's time will continue to be wasted trying to cope with the childish behavior of a few trolls.
In lieu of a system for identity verification, what we have at present is a system that falsely pretends IP numbers can be equated with individual identities, which is both ineffectual and unfair to innocent people. It's ineffectual, because abusers can easily get around an IP ban. It's unfair, because banning an IP number punishes an entire class of people for the behavior of a single individual.
Arguably my "hotmail-yahoo" approach is a half-baked way of achieving identity verification. I think Erik Moeller's "permanent cookies" approach might work better.
Another approach might be to create a _reward_ system for people who supply a verified identity. We'd have three classes of users:
(1) People who have supplied a verified identity, through a means to be determined.
(2)People who have registered but HAVEN'T supplied a verified identity.
(3)People who haven't registered at all.
The reward for supplying a verified identity would be greater trust from the Wikipedia community. You'd get some extra privileges, like the ability to mark changes as "minor," and you could participate in a system of rating other WIkipedians similar to the way eBay buyers and sellers rate one another. There would be a feature that makes it possible to filter out high-ranking Wikipedians when people look at a list of "recent changes," thus reflecting the presumption that highly-rated Wikipedians need less monitoring that people who haven't earned that level of trust.
People who have registered but haven't supplied a verified identity would still be able to contribute as they do at present, but they wouldn't be able to earn "trust" and would therefore always be ranked at zero.
People who haven't registered would also be treated the same way they are at present.