A while back I ran a test.
I had a friend, back in the day, talking about making prints for clients from 4x5 and 120 film. He had his own in-house lab and was trying to get away from luggin the ol' view camera around. He made 2 8x10 prints and his comment was the print from the 4x5 just looked better, in almost an indescribable way. I did the same test with digital camera files.
I shot with the big Canon, (MKII, I think), a 22MP single-shot and a 22MP Multi-shot, and instead of what I usually do, which was make the most enormous print I could, I just made the best print possible at around 16x20 (figuring "what commercial job goes any bigger than this?).
The difference was really amazing. Think back to your film days and imagine the same photo from a 35mm, 120 and 4x5 printed at 16x20 and that was pretty much the result.
I'm throwing this out there to see what people think. I'm sure there are guys out there who shoot both, depending on the job, and I'm also convinced that the "common" DSLR chips and processing are getting even closer. There's another thread going on about the cost of MDFB, and for my two cents I doubt very much that you're going to see the big chips drop in price much, but what I do think is realistic is to see the DSLR difference becoming closer and closer to the "bottom end" of MFD.
My predictions, FWIW are that we're going to see the big-chips concentrating on the high-end machines, the "full-frame" 645 and completely integrated systems, rather than continuing to get hammered by DSLRs. Shooting speed and noise at higher ISO are simply two things that the big chips cannot deal with as well as DSLRs, from a simple physics perspective, and the "35mm" DSLRs just can't compete with the resolution, pixel quality and bit-depth of a "645" chip...
...but how good a file do we need?
8x10 is certainly a higher quality piece of film than 4x5, but I don't know any commercial studio that shot 8x10 past the mid-'80s...