Dudes...
You guys are both "right" as you are arguing over two different concepts. Perhaps the misunderstanding is American English versus Australian English and "percent" vs "percentage point" and in what contexts they are used. I am neither sure nor care.
Look, here's how I see it. In essence I agree with both of you. Yes, I agree that something does not smell right. However, by all objective mathematical analysis it is impossible to make a definitive call one way or the other if something truly fishy is going on - there are too many degrees of freedom with unknown uncertainties. Even the accuracy of the SRM (quoted to within +/- 2%) could be the source of this 20 watts improvement if he changed units. Also the funny rings I believe introduce some source of inflation over round ring watts, even when the slope is properly calibrated and the offset properly zeroed. This is due to the comparative oversampling in the power phase since crank based units assume a constant angular velocity for each revolution in calculating the power output. Did he always use q-rings, even on the track? On all his training bikes? Etc, etc.
The biggest uncertainty, as Andy Coggan identified, is his weight. We have 131313's statement that he knows someone who saw him standing on the scale at the TdF this year at 72kg, so that's probably the best figure to use, although I do remember reading somewhere that in 2011 he came into the TdF at 69kg, his lightest ever. Whether that remained the same for the WC TT that year or not, who knows.
The frustrating thing is obviously at this level 20 watts can be the difference between winning the TdF or not. It is also a gain that can very easily be achieved via micro-dosing and/or other methods. However, as we are seeing, even with published power data 20 watts is still within fundamental uncertainties in the assumptions. IMHO this is why you guys are arguing with equal passion.