Taken from Bikeradar:
Stages engineer Andy Lull backed up Kerrison, and said that in his testing the non-round rings "do not seem to allow the rider to produce more power at the rear wheel through a mechanical advantage, but rather produce a higher power reading from crank-based power meters."
"I can’t speak to the biomechanical efficiencies or psychological advantages of non-round rings, only to the phenomenon of the fluctuating velocity biasing the crank-based power measurement," Lull said.
So do O.Symetric rings work at all, or is the 5-8% power increase the company claims down to pure measurement error?
They may "work" for some things, but there is a pretty well established effect of a phantom power increase (i have seen the figure 5-6%) vs. round rings. You can find a ton of info on this on the wattage based training forums, and I think there is some good info on it on the DCRainmaker blog too. In other words, if you are trying to prove that non-round rings work, you don't want to hang your hat on this particular power aspect.