Sure but a 20MP vs a 24MP is 10% difference on X and Y axis approximately. Are Canon that much behind the competition?
No they are not. Not too long ago I spent an unreasonable amount of time comparing files from the Sony A7 (24 MP AA filter) to the A7R (36 MP no AA filter). I always resized the 24 MP file to 36 MP with very light sharpening (scaling tends to soften) so that size (and therefore magnification) differences didn't affect my judgement.
You're looking at a nearly 25% axis gain in this comparison vs. 10% for the APS-C sensors.
My conclusion? Occasionally an area of very, very fine detail would be rendered better on the 36 MP sensor. This would be visible while pixel peeping but never in print.
I came to a similar conclusion when I mixed in 5D3 files, though the 5D3 file must be RAW. The JPEGs fell behind. Also, when converting the 5D3 file you have to be a bit more aggressive with the detail and sharpening settings. The Sony sensors would handle heavy processing a bit better because of this.
At the resolutions we're dealing with today I would say you need a 50% gain on each axis before differences become visible in print, all other differences (i.e. sensor size) being equal.
Erm, yes they are... the OP was complaining about a 2MP hike in comparison to the current 7D. Current APS-C is around 24MP and the difference is about 10% per axis (cf 20MP). I was indeed saying the same as you - there has to be a significant gap in MP before for you to notice. So 20 / 24MP gap, really not that much...