While I know photometria can be both a genuine and an accidental pain, I wouldn't dismiss categorically all that he is saying. The "pure raw" data is usually obtained from well characterized CCDs. It's very hard to obtain "pure raw data" from a CMOS sensor, especially an active one like the current Sony. The content of Nikon RAW dslr files has always been post-processed to some extent, as all amateur astronomers using DSLRs for astrophotography know (
http://astrosurf.com/buil/nikon_test/test.htm). In addition to that, the current Sony sensors perform per pixel calibration and, while it is not possible to dynamically change the fundamental collection of photons at a given sensels, it is quite possible to pre-charge the sensel depletion zone and change the linearity of its response (that's one example, not a claim that they do it, in its sensor briefs Sony mixes genuinely interesting information with pure marketing). Whether it is only done at the individual sensel level to handle non uniformity or it is also done at the global level to achieve some kind of in-sensor HDR equivalent is a tough question, deserving a deeper investigation than casual dismissal. Whether all of this matters a lot for most photographic applications is another question, maybe not as tough if one sees the camera as a whole as a black box.