There is a book that came out a few months ago by the people who built the reputation system that Yahoo uses called, aptly enough, "Building Web Reputation Systems". It's a really good book. One of the things they make very clear is that the idea of trying to build a globally useful reputation system is basically a pipe dream.
The problem is that reputation is earned in context. So, reputation on a particular blog is useful, but trying to extrapolate that out to be useful on another blog is not as useful. Reputation gained in one domain is not really transferable to another. (As an example - if Jon Skeet had a cooking.stackexchange beta account, he shouldn't be automatically the most reputable user just because he has the top stackoverflow account - cooking and coding aren't directly transferable skills.) The best you can hope for is saying that this person is not a troll or spammer. The real hard part is, say, for Wordpress, it's easy for people to create a new account - how do you distinguish between new people and trolls on new accounts? This is probably the fatal flaw for the use case you are espousing.
You also need to think about how you're going to control overprotective "old timer" members. I've read of people that put in systems only to have the top members down mod everyone else so they could protect their top spots. I assume for the Wordpress plugin that you are going to add some sort of comment voting system. (On a side not that's ANOTHER problem with blogs - there isn't much a person can actually do on them - read a post, comment, maybe vote in a poll - hard to build a reputation system around that).
Obviously, since my startup (IActionable) is in the same space and would be a competitor I believe in the general concept, but I think many people make the mistake in thinking that it is an easy problem to solve. Blogs especially are a tough nut to crack because you have to add in a lot of functionality that most of them don't come with.