[mst]

The site clearly says that absolute numbers cannot be directly compared when sensor sizes are different.

We do, but still these kind of discussion/questions come up again and again (see also the thread about the Tamron 24-70 VC and Sigma 24-105 OS in the lens section). I have the feeling that we need to explain the topic in a little more detail in every review.
 

PS I wonder whether it might have sense to "normalise" them... at least to have a gross comparation.

We tried this once. But that makes things even worse. Because the results still are not cross-system comparable, but the normalised scale suggests they are.

If you want a gross comparison, don't look at the numbers, but at the rough scale that we use (the one going from "poor" to "excellent"). But be aware that it still is only a "rough" comparison. The camera used (or, more precise: its sensor and the software workflow used) still remains a fundamental part of the results.

Edit: I just noticed we don't use this scale consistently across all sections. We'll discuss this internally.

-- Markus