I used CrowdMed a couple of months ago. I have experience in IT and thought that using some of the neat things about the industry, like crowdsourcing, for medicine actually sounded like a great idea.
Unfortunately, I can't say I was terribly impressed with the diagnostics. I think the above list speaks for itself, and if that's what's driving their 55% 'success rate' I'm even more suspicious. The actual functionality is excellent (in terms of uploading charts and results, etc) and I still believe the site has a lot of potential.
The problem, IMO, is that the site seems to attract the attention of mostly non-experts unless you put down a very hefty reward fee to entice the 'bigger fish' (ie real doctors, present or retired) to comment on your case.
To reach that level of usefulness, I'd be far happier simply paying a fixed fee up front (rather than specifying a 'reward'), particularly as people coming there are often drained of money (and may not have the resources to post exorbitant 'prize money') and even if one of the suggestions happens to be the correct one, you're still going to have to go to a real-life doctor and then float the idea (which is always a questionable way of doing things).
I interacted with five or six 'guessers' and the disparity in quality couldn't have been more obvious. The one real doctor was the best (answered my questions calmly and with reference to actual studies). A 'psychic' from Las Vegas diagnosed me with 'heart failure' (despite my basically normal ECG), and an extremely zealous med-school student suggested a host of extremely rare diagnoses, none of which matched my symptoms in the slightest, although I appreciated his effort and sincerity to help.
I'd take major issue with their claim that patients are often capable of diagnosing better than doctors - particularly when it comes to rare diseases. If my year of amateur 'med research online has taught me (attempting to self diagnose), it's that medicine is painfully vague and complicated and doctors really are in the best position to pair our symptoms and lab results and try arrive at a possible diagnosis. They also have the benefit of having studied years of the underlying anatomy and biochemistry in a detail that (generally) none of us have.
Although we may often be able to glean useful information from the internet to work with them, I wouldn't place the diagnostic process in anyone else's hands other than someone who has gone through the proper training.
If there were a way to guarantee that it was limited to real medical professionals and employed a fixed fee model I'd happily give it another chance, but as it stands at the moment I'm not sure I see the merit in it.