I just got an interesting email from http://www.rentacoder.com noting that they are getting some bad press due to retaliation from buyers or coders that had a bad experience. I want to help them clear the record by saying that while I’ve dealt with some terrible coders, the arbitration process and design of their system is excellent.
Out of my 4 bids so far:
Push C++ function calls on the stack via assembly: Went very well. Coder submitted the work about halfway through the deadline. We worked out the kinks over the next few days. Coder said he enjoyed the project, and I liked working with him because he cared about doing a great job.
Upgrade C++ email client to support GMail: Coder didn’t do anything for two weeks, and lied about it on the status report. It was my fault it took 2 weeks, because I didn’t know at the time that I could have canceled the project due to not filling out a status report, rather than having to ask for one.
PHP directory server of games: Went OK. Coder submitted on the last day, generally working, but with problems caused by a lack of care and attention to detail. I had to spend about a day of my own time fixing stuff, but overall the work got done. Rated the coder a 7 / 10.
Port PostgreSQL to MySQL for autopatcher: Went very bad. Coder submitted on the last day, completely not working. It’s not that I was being picky over quality, it was that the project was utterly not working and only about 25% done. Coder denied the flaws, and at the end blamed the arbitrator for being unfair (on the contrary, the arbitrator was 10X more patient than I would have been). Got my money back a week later.
The system is intelligently designed, with double-blind feedback, ratings based on a factual chart, escrow for both buyer and seller, required status reports, and a chat and message system which is logged. The only problem I see is rating inflation, where most successful projects are rated a 10/10. This is the fault of the users to a large degree for not following the rating chart. It would be better if it asked you specific questions (was the project late, etc), and derived the score based on those facts.