It’s about time—
Old school (Nikon and Canon): choose a focus sensor, then try to put it precisely on a small target. But the focusing spot might be larger than the target (the eye), so enjoy sharp eyebrows and blurred iris.
Transition: face detection. Works OK for some things and stopped down, but aiming for the wrong thing, coarsely. And just plain dumb to not take the logical step to focusing on the eyes, specifically.
New school: eye detection. The Right Way. Nirvana for portraits with fast lenses, assuming it works well.
If the Sony A7R can really do eye detection well wide open, then that is a huge breakthrough for making portraits— getting the eye crisp has long bedeviled me. Even better if manual focus assist offers eye detection guidance.
What is really going on in general? (not just this idea) Sony is thinking and innovating. Nikon and Canon are in a rut, having done essentially nothing new in 10 years—a Nikon D800E is not very different from a Nikon D1 other than Live View, and that is badly done on the D800E.
See my thoughts from nearly a year ago in Sony vs Nikon and Canon — Lunch is Served, but WHO WILL EAT IT?.
Both Canon and Nikon are in serious danger of becoming irrelevant from a market share viewpoint and possibly to the point of finding R&D costs grow too high in percentage terms as the business shrinks. Believe it or not. Creative competition has taken down all companies eventually. That is Good.
To the point: I dislike carrying a heavy and bulky EVF-less DSLR on backcountry hikes, or even in general: traditionally it has been worth it, and it still is for the image quality of particular lenses like the Zeiss Otus 55/1.4 APO Distagon. But what if I can shoot that same lens with high quality magnified Live View using a high-res EVF? (Assuming I can support the lens properly). For a faster shoot with a higher hit rate (focus). Why would I carry a D800E if the ergonomics work? This remains to be seen (DSLR lenses + adapter and their ergonomics).
Consider that with three lenses, one can cover a wide variety of material. With mirrorless, the lenses can be smaller due to fewer design constraints (no need to clear the mirror box or illuminate a mirror). Really top grade ones won’t be much smaller, but can also be optimized for current technology.
Gravitation to the 80% solution
That is not to say I would not want or need a DSLR for some uses, but it acknowledges a reality that many users are feeling these days: if one can have the same quality at 1/3 the weight and 1/2 to 1/3 the size, it’s a powerful incentive to do most work on the most convenient platform.
Which of course implies that the whole package has to have an ergonomic and operational efficiency*. DSLRs could make a big leap forward by offering an EVF; the lack of an EVF is a constant hassle for me (getting precision focus): it puts me on a tripod with a loupe with a mangled Live View on the D800. Which is a very different experience than with Sony RX1R or Leica M Typ 240.
* For examples, the Sony RX1R in the field to be far superior in hit rate and speed and pure enjoyment thatn the Leica M240, and indeed much faster and more enjoyable than my Nikon D800E. If the Sony RX1R behaves as well or better, it