I want my pictures to look like they looked or would have looked to my naked eye, not like what they out-of-camera JPEG looks like, and they are very often dramatically different.
Presumably, on a pro body you'd be able to adjust the EVF to suit your needs. Also, what if it's dark? Your naked eye won't see much, but EVF can see in near-darkness now. (maybe this doesn't apply to your style of photography)
Adjusting the EVF isn't sufficient. What I can do in post is much more than can be done in the 5ms or so available to the processing pipeline in the camera, and it's very often the case that all of that post is what's necessary to get the image to look as it did to my eye. Further, the sensor can't even do what my eye can do, partly because my eye is a much larger format even than full frame and partly because my eye can do its own pre-processing (different ISO at different locations, lateral inhibition) before the data is even sent to the brain.
The mirror assembly serves a very useful purpose, and EVFs are many, many generations away from being "good enough" for me (I'd estimate decades away).
I guess I'm more optimistic. Aren't high-end video cameras all EVF now? If so, it seems like just a matter of a few years before that tech comes to still cameras.
Video cameras have EVFs because they have to, and they all stink. Even some users of RED's nearly $4000 EVF have demanded a non-TTL OVF because the EVF isn't suitable to their uses.
The difficulties with 36x36 sensor extend to the entire lens system as well.
How so?
The image circle isn't big enough, and some lenses already have rectangular hoods or rear windows, or both.