Foldit (or is it Fold It! ?) turns protein folding into a game. Protein folding, as in Folding@Home, as in a computationally hard problem that was most cost effective to solve with distributed computing. Distributed human computing, like Amazon's mechanical turk, has long been hailed as a great way to solve computationally hard problems that people might just be able to "do," like image recognition or OCR proof-reading via CAPTCHA. That it is now wrapped up in a game, and a supposedly fun one at that, is a massive leap forward.
I'm withholding judgment until I can play it, but if it works as advertised, then two questions may be answered:
1) Can games can be used as a wrapper for computationally hard tasks which have real-world significance? This will depend on the popularity of Foldit to prove out its "fun," and will be a triumphant vindication of "serious" games.
2) Is human distributed computing more efficient than brute-force distributed computing? I doubt there is a control group of computers running similar folding problems concurrently, so this will depend on an analytic evaluation and some time to gather data.
Either way, this is tremendously exciting.