"If you aren't so lucky as to have someone come along who will volunteer their time and do a fantastic job for free, you are faced with accepting mediocre free work, offering prizes, or paying someone. Since prizes are usually inefficient, it appears that offering to pay someone is the best option, so long as you are able to stipulate that there will be no or reduced pay if the work is not done on time and to specification."

Are they really so inefficient? Worth noting that in at least a few ways, Robin Hanson thinks they're much more efficient than other methods; see http://hanson.gmu.edu/whygrant.pdf and http://www.overcomingbias.com/2007/01/prizes_versus_g.html