I am on both sides of the whole sensor size argument.
I have never been concerned about the crop factor as its not changing anything optically, the none digital lens on a digital body still projects the image to a 35mm cone and the APC only see's the center of that image, you can easily cut the center out of a full frame image and get the exact same result, although at a lower resolution. The 5D total resolution is higher than the 20 or 30D but if you clip to the APC area the resolution of that clipped area is quite a bit lower. The 5D resolution is 4368x2912 with a 35.8x23.9mm sensor and the 20D is 3504x2336 with a 22.5x15.0 sensor, if you crop the 5D image to the APC area the resolution is 2745x1828 which is a good bit lower. The 5D has slightly larger pitch on the individual pixels which means more photons fall on a given pixel, this should mean the CMOS sensor should have a lower signal to noise ratio and will be able to simulate faster film.
The only place where the 1.6x factor is a real issue is if you are trying to take the same framed shot on both cameras, the full size will have to zoom 1.6x times to match. Optically this is different but on telephoto shots it looks almost identical. However any work with a narrow depth of field, including macro work, the factor can be an issue. At the big picture level, the focal length or distance to subject has to change for the same image therefore the area in focus also changes, a big issue for macro work.
In reality I am more concerned about the fact that film and 35mm sensors are very unforgiving on the lens quality, seeing the entire image and not just the center which is typically of better optical quality. I have a mix of good and ok lenses so the first thing I'll have to do is check the edges of the pictures taken with the cheaper lenses.
I'll post my first thoughts when it gets here..
Rob