Find some way to rate the success of free software projects. Establish a futures market for these projects (like the HSX does for celebrities, the Iowa Electronic Markets do for politics, or the Foresight Exchange does for general predictions). This market would effectively let you bet on the success
or failure of chosen projects. It could use "funny money" (like most of the markets I've named) or real cash (more useful, but regulated).
1. It would effectively sort the wheat from the chaff. Freshmeat is currently almost useless because of the preponderance of "gtk CD player of the day" apps. This market would be an effective way to discover promising new projects, as well as giving you some reason to continue to troll Freshmeat and Sourceforge looking for "diamonds in the rough" (if you're so inclined).
2. It offers a way for free software developers to make money. They can effectively buy the equivalent of employee stock options in their own project, which will (assuming the market uses real cash) encourage them to succeed and reward them for success (possibly funding future work).
3. It offers a way for the users of free software to pay for its development, and to eliminate the sizable risks associated with depending on uncertain free software development methodologies. If a certain project would make your life better, you can bet against its success in the amount by which it would improve your life. If the project succeeds, you get the benefit of the software; if the project fails, you get the money. Set the amount so that it's even either way. (This works particularly well for corporate users for whom the value of software can often be quantiatively measured.) This money may well end up in the hands of the original developers (see #2).
The biggest problem is finding an effective, fraud-resistant way to measure the 'success' of a free software project in a way that helps accomplish all three goals listed above.