Well, that doesn't actually change the point - there are knowledgeable people who get paid a lot that aren't going to summarise an aspect of their field, or a field they know a fair chunk about, for $2. Wikipedia's anti-expert bias doesn't really change that.

The little I know of economic theory suggests that replacing intrinsic rewards - like the warm fuzzy feeling you get from contributing - with a small cash reward means that people will value contributing to Wikipedia at the price of the small cash reward. This is invariably less than the dollar amount they'd attach to an act of charity that also spreads knowledge.

tl;dr: don't offer cash rewards for people doing things for fuzzy emotional reasons. It doesn't work.

Part of the problem with team-based puzzles is that they're very difficult to do in a persistent world. Once they're solved, they're solved. Do you put them back? The person who knows the solution is wandering around some place, and is free to post the solution on the Internet. If you don't, then players not at the cutting edge of the game essentially play clean-up -- assuming, of course, that one *has* a cutting edge. An ARG is essentially a kind of MMO - it has a persistent world that's shared amongst all players, after all, and they assume that players are working together, and so are free to make the puzzles as tough as they can, confident that players will eventually find a way through.

To WoW's credit, most of the bosses in the dungeons in the two expansions require players to work out and execute on a strategy to defeat them. Unfortunately, the strategy is generally worked out during the beta testing, well before most players reach it, and players generally don't have the luxury of figuring out the strategy on their own.

Someone will eventually crack the way to do a persistent world with puzzles and discovery while being able to renew solutions. I expect it'll be done by designing a class of puzzle that the game can assemble hundreds of unique variations on, but players can't solve by building their own solver.