First, I'd like to say thanks to all of you for a well-reasoned and polite discussion.
I only have one camera (7D) and have been debating getting a 5Diii.
...
That said, it should be noted that the higher spatial resolution of the 7D is only really a benefit in focal-length-limited scenarios. If you are stuck with a specific lens (say the 600 II), unable to get closer to your subject, and unable to add a teleconverter, then the 7D will always be capable of resolving more. In all honesty, that is frequently a contrived scenario, and you more often than not are at least capable of slapping on a 1.4x TC, and in the case of the 1D X (and hopefully soon enough, for the first time, the 5D III as well) adding a 2x TC. For the more skilled, getting closer to your subject is also not a problem, and most professional bird photographers have the sneaky skill to get within feet of jittery subjects, so filling the frame and getting as many pixels on subject is rarely an honest real-world problem when money and time are no object (or at least a secondary concern).
While there's no doubt the 7D has its limitations in ISO performance, especially, I have to say I don't think the "focal-length-limited" scenario is a contrived scenario at all. For me that seems to be the rule rather than the exception and its not because I'm unskilled. From personal experience:
1. You're on a cliff on one side of the river and the bear is on a carcass on the other side.
2. An owl in a nest is being mobbed by crows - same deal, can't get closer because you have to drop off the steep slope and wouldn't be able to see anything but trees.
3. You don't want to get closer to the wolf pack
4. etc.
As far as slapping on the teleconverter, I already did that with the 7D.
Very good points, and it is good to point out that focal-length limited scenarios are not "impossible". I did mention that they were often contrived, but that does not necessarily mean always, and the higher density (and therefor added reach benefit) of the 7D sensor is definitely a bonus in those scenarios.
Now that I have that out of my system, I'm hoping you guys can point out the error in my thinking here:
Given the "focal-length-limited" scenario above, and assuming good light (I know, that usually (or should I say almost always) isn't the case but speaking theoretically here), you want a 16 x 20 print and you crop the FF to the 1.6 dimensions to get equal subject sizes on the final print. I think that gives you 216 pixels per inch for the 18 MP APS-C vs. 150 for the 22 MP FF crop. Doesn't that give the APS "reach" some advantage?
Definitely. There will always be the benefit of added reach with any sensor that has a higher pixel density. It does not necessarily have to be an APS-C sensor, it could be a FF 47mp sensor, which has the same pixel pitch (4.3 microns) as the 7D. If the 7D II hits the street with a 24.4mp sensor, that is about the same a FF 64mp sensor. Reach is a benefit of pixel density, not necessarily form factor.
That said, it really depends on how clear and sharp, and to some degree noisy, your images are. From what I've seen in terms of bird photographs from the 1D X, it is impeccably sharp! Even in a focal-length limited scenario, you could probably blow its images up pretty far and still be satisfied. You may not have the same amount of detail, but that shouldn't really matter in most cases...so long as the detail captured looks good...is sharp, clear, with high color fidelity and low noise. Noise is less of a problem in print for enlargements up to 2x native size, as print density is generally much higher than screen density. Even at 150ppi, you are still about 45% more dense than the average 103ppi 30" screen (or 106ppi 27" screen). In the case of screens with 96ppi, that same print is 57% more dense, and in the case of a 72ppi screen, the same print is over 100% more dense.
Also, keep in mind, printers don't really print pixels when all said and done. They print ink droplets. The driver's rasterizer (or perhaps a custom rasterizer) converts image pixels into a set of ink droplets of varying color with a given dithering to maximize the quality of the image in print. Depending on how carefully you tune your images for print (manually resize to the exact native size and PPI for the print you intend to make, manually set white and black points, manually verify gamut, etc.), and depending on the capabilities of the rasterizer you use (Canon and Epson driver rasterizers actually do a pretty good job these days, but better third-party alternatives can also be found, usually for a price), the appearance of noise in the final print can take on a very different quality, usually a better quality. I rarely have problems with noise in my prints, usually generated with fine-tuned images tweaked in PS6, and printed with the Canon driver, on a PIXMA Pro 9500 II and various high quality fine art papers. The only time I do have problems with noise is with images taken at ISO settings above 1600, and then, usually ISO 2500, 3200, and higher.
Getting all of the complexity out of the way, though. I would say that for the most part, just about any image taken at an "artistically usable" ISO setting on any camera can usually withstand up to a 2x enlargement, assuming you maximized the potential of whatever gear you are using. The 1D X offers numerous benefits over the 7D, and even if you crop it, you should be able to enlarge that cropped image up to about 2x or so before you could really notice any IQ issues in print. If you really wanted to enlarge a 1D X crop to the same size as a 2x enlarged 7D full size image, you could probably pull it off thanks to the IQ benefit of the 1D X. The detail level wouldn't be the same, but I'd bet few people would notice.