First off, I'm probably biased towards the kinds of shooting that I mostly do.
I don't shoot sports. And I really don't do BIF, either.
So I tend to think in terms of what would make
my life easier. And to me, even the current 40D live view (magnified) works pretty well for many manual focus situations. So an even better, faster, lower-latency in-camera EVF would probably be handy for me a lot of the time.
I also wouldn't compare a cheap P&S of today against what "might" be possible in a fancy DSLR down the road a few years. I'm just speculating about what could happen in the future (in the spirit of what the original poster asked). It's speculation, of course. I'm not suggesting that we install the crummy AF and back-of-camera display from a cheap P&S of today into a DSLR, and call it good
I'd expect to be using a similar AF system to what we've got now, and NOT rely on contrast detect AF from the sensor's data except when we choose to use that mode for slow-moving or fixed scenes. So the AF should not suffer at all because we'd still have what we've got now. We might have contrast-based AF to supplement it, as an option, but I would not toss out what we've already got.
I was mostly responding to jpr2's concerns. He asked:
"Jim, so you mean this EVF-only DSLR to be a still-frames, studio camera only?
Even with a modest 300/4L IS + 1.4x, such 1/500 SS is rather slow

("
And I was talking about shutter speeds.
My point was that for the purposes of the live-view images, we really don't need to have a high "virtual shutter speed". That's because to our eyes, any given frame in the EVF would not need to be taken at a high shutter speed because it'll just be immediately replaced with a new image before we've had a chance to analyze and notice the blur in that particular frame out of the "stream" of frames.
I still hold to this idea.
Broadcast TV is really amazing, even at 30 FPS. Any given frame, pulled out of that stream, can be very blurred. Yet we don't notice it because we cannot analyze 30 frames per second for sharpness, really. And we can't when viewing through a viewfinder either.
So the "virtual shutter speed" required by the EVF could be quite low, just as it can be for any video that is not being analyzed frame by frame.
So that's all I was saying.
I was not addressing "lag", and I was not suggesting that the camera itself could not use a high shutter speed for the actual photo exposure. I was just saying that for the purposes of the EVF, the virtual shutter speed need not be high. We cannot benefit from super high shutter speeds in that part of the system anyhow. Things DO need to be stable in the frame for us to begin analyzing focus, and for the purposes of framing or following action, we don't need each "frame" of the video to be dead sharp either.
The AF system, on the other hand, will still be getting its optical images directly, the way it always has. And of course, for the actual photograph, we'd have any shutter speed we want. That's a separate issue from the virtual shutter used to generate the EVF image.
My main reason for wanting an EVF option would be for manual focusing. When you think about watching a TV image, you can see if it's in focus or not, even on moving subjects. Having that same degree of ability to judge focus in the EVF would be sufficient, I think.
And when things stop moving, or slow way down, then the individual frames would become sharp enough to allow us to do that very critical manual focusing. So it'd be fine, I think, even with a fairly slow "virtual shutter speed" for the EVF image.
That was my point - concerning the virtual shutter speed used to generate the EVF.
I can see how the currently available live-view lag would bother people shooting sports or BIF, etc. But I'd like to see what can be done two years down the road. The live-view of the current 40D would be a pain for BIF or sports or a lot of moving subjects. I totally agree. The lag would just kill you. And perhaps that's where the virtual shutter speed does become a factor. To get low lag, you'd need a fast shutter speed. And that might end up being a real stumbling block to this.
But I'm not sure we'll be stuck with as much delay as we see now forever. I'd hope that improvements could be made over the years.
And if, as has been suggested elsewhere in this thread, we end up with a system where you can select between EVF and OVF or a blend of the two, then that might ultimately be where things would go so we'd have the best of both available.
I do see possible advantages to an EVF. So it'd be neat if the problems could be overcome because the EVF's advantages would be nice to have available.
--
Jim H.