Funnily your greens and reds are obviously more saturated in the second picture, which is apparent on the background bricks and moss ! This is more in line with the theory of ETTR.
So we are speaking of a major imbalance in the treatment of the components.
Maybe your software is playing havoc with colours, trying to compensate for a known colour shift of the sensor in low light, but applying it in a range where there is no sensor coor shift due to the overexposure ? Just a wild guess...
I think we have a different understanding of the word saturated. If you put a colour picker on the reds or greens they are more saturated in the top picture in all three sets. The 'correct' exposure is giving more detail and saturation than the severely overexposed image.
Thanks Private, yes that is what I would expect.
I believe blue is the weakest wavelength and this may have something to do with the effect ETTR has on blues, but other colours are effected in the same way. Not sure why feanolas thinks the second image has more saturated red, green etc.
If you intend to lift shadows then a small amount of ETTR will improve the overall image, yes, but ETTR for the sake of it is a myth, and the article from LL that started this thread is absurd.
I suggest anyone who believes in ETTR to 'improve IQ' try taking the same shot twice, one at correct exposure, the other two stops over. Convert to a JPEG and then have a look at the difference in size of the two files.
Another point that I would make regarding exposure is printing. I have always found that the closer the exposure used was to a 'correct' incident light reading the better the picture prints.