That doesn't seem like a very nice assumption, that new users will have bad ideas, and if old users lose karma every time they vote up a new user, how can you ever know if the new user genuinely has nothing to offer, or if old users are just wary of decreasing their score, and therefore, influence? (i.e., I have lots of karma, and I know if I upvote this comment that I like, it will increase their score quite a bit - I have influence now - but it will also lower my karma, and I'll have less influence over the next comment I see which I may like more...)
This is what happens in chess: highly-ranked players just simply won't play new or lower-ranked players at all. It's not a bad thing there, but then again, it is a competition. A social site isn't, though.
Oh I agree completely that it isn't a nice assumption; it is just a simplifying one (like the proverbial spherical cow) that's mostly wrong. But I don't think it is entirely wrong, and the ways that it is right may be useful, as an experiment.
Okay, but why should votes cost more for a user with more karma? I would think the opposite should be true, that if karma is an indicator of a user's value to the site, someone with more karma should pay less to vote up a newer user.
The way that I'm thinking about it is this: that it is relatively unusual, from the standpoint of the community, for a high-karma user to upvote a low-karma user.
Karma is a measure of the norms of the community, parceled out to users.
This is what happens in chess: highly-ranked players just simply won't play new or lower-ranked players at all. It's not a bad thing there, but then again, it is a competition. A social site isn't, though.