Wide-to-tele zoom lenses like the EF24-70mm f/2.8L achieve their constant aperture differently than either the wide-to-wide or tele-to-tele designs; in this case the iris diaphragm is "cammed" so that it changes its size as the lens is zoomed. You can see this if you look through the lens off the camera while you're zooming it. If the diaphragm was not cammed, the 24-70/2.8L would be revealed as a variable-aperture zoom with a maximum aperture larger than f/2.8 at all focal lengths under 70mm.
So this is the thing that I've always wondered, is why do they do that?
Sigma and/or Tamron have f/2.8-f/4.0 vari-aperture zooms in the 24-100 ranges, which is basically them taking a f/4 constant aperture lens and removing the f/4 restriction on the wide end so that it comes out at f/2.8.
So if we take the EF 24-70 f/2.8, Chuck's just said that we could take away that cam and make it something like a 24-70 f/2.0-2.8.
Well, why don't they?
Is it just the 'prestige' that constant-aperture is better than variable? (even though it only really affects you if you're shooting wide-open in M or using flash). (and if you've got a f/2.0-2.8 vari-aperture zoom and using M or flash, you can just set the aperture to f/2.8, when you zoom out it will stay at f/2.
Would the IQ suffer that badly? (given how well the 24-70 II performs, I don't think it would do that much worse at 24mm f/2.0).
Do they want to protect sales of the 24mm f/1.4 L? ([/cynical]).
The only real downside I can see is vignetting at the widest end, it's already 1.8 stops at 24mm f/2.8, at f/2.0 it might be as bad as the 15-85 @ 15mm. But jam it on a low-noise body like 5D3 or 1DX or even 6D, and you can use PIC in DPP and fix it right up.
I know it'll never happen, but it would be a nice thought...