Interesting. I would imagine that with any given sensor, a 1.8 lens would have a 2/3 stop disadvantage compared to a 1.4 of the same focal length. Maybe I am mistaken?
According to the article , the problem originates from the extra wide aperture of f1.2 and f1.4 lenses and the rays that pass near the sides/edges of the diaphgram . The rays are too skewed/whatever for the pixels on the sensor to detect - so less light than there theoretically should be . The light loss is actually supported by tests done with these fast lenses . The t-stops are actually 0.5-1.0 EV lower than the f-stops .
It suggests therefore that the f1.4 may not be real and just a result of boosted ISO without our knowledge. ( It is not impossible considering Sony is doing a similar trick with the pellicle mirrored A55 )
What is missing in the article is the same t-stop test done for f1.8 lenses - instead of just indicating there is no such light losses at f1.8 aperture .