A common theme in criticisms of my "Catching up with the D800" article (below) seems to be "the D7000 had the same problem, why didn't Thom write about that?" Some get more personal and slander my reputation by accusing me of not doing research at all.
But this is all about research. For over a decade I've been commenting on and dealing with the issues that come up in the Nikon world. In almost every case of a new Nikon DSLR appearing I've worked with at least a half dozen, often more, owners trying to come to grips with their cameras. The D7000 was no different, indeed, I probably worked with more D7000 owners and cameras than I did any other new Nikon DSLR in the past couple of years.
A common complaint about the D7000 was "it doesn't focus." Just got another one in my In Box today, as a matter of fact. The question is whether that represents a real defect in Nikon's QA or not. In other words, does the camera actually not focus correctly? The answer for the D7000 cases I examined was often that the camera was perfectly fine, but both changes in the camera's focus system coupled with other issues such as field curvature, focus shift, and the higher pixel density came into play.
One key test is simple. The person says their camera doesn't focus. They hand me the camera. I take a picture and it's in better focus than their image. Is the camera at fault? No, so we have to fix a user assumption before continuing. Assuming my image didn't get a better result, the next test is to calibrate the lens/camera properly. Did that solve the problem? If yes, then Nikon's factory tolerances are loose enough that it causes issues in the field, but they're manageable if you know what you're doing. Virtually all D7000 cases I've examined ended there, save one that indeed needed repair. At this point I've been through this several dozen times with D7000 samples. I myself have two D7000 models, both of which test out the same.
From Auto fine-tune the testing gets much more tricky, by the way, because so many different factors come into play. Mount alignment, mirror alignment, focus sensor alignment, focus shift, field curvature, even just getting the test equipment properly lined up for the test itself.
Short version: I've been doing this kind of early user testing since the D1 appeared, and with cameras well beyond the ones I purchase. When I was teaching a lot of workshops (up to 2010), I had a lot of chances to do side-by-side comparisons of cameras, and often did. But I still try to do a handful with every release. What I found with the D7000 was that I could align and get acceptable focus with every camera I handled (except that one); in one case we even found a lens that couldn't align with any of my cameras.
The D800 case seems different. Two of the twelve D800's I've had in my hands had a clear left/right focus alignment problem, the same problem that is getting commonly reported on many Internet fora. It's a little tricky to test for, as field curvature on most lenses means you have to move the camera or the test rig with some testing jigs (e.g. LensAlign). But I'm comfortable in reporting that I've seen two cases of the problem that would require Nikon to do something about.
I believe I wrote this before, but it bears repeating. Way back in the early 90's Popular Photography's Herbert Keppler wrote a series of articles about autofocus "tolerance." Simply put, a phase detect system always has a small amount of tolerance to where it puts focus to start with. A good manual focuser can beat an autofocus system on a static target every time. Couple that with the fact that the viewfinder neither shows the exact position nor shape of the autofocus sensor being used, nor does Nikon provide much of an explanation of what happens when only part of that sensor covers the desired target, and there's plenty of room for inconsistent results built into the system in the first place. Now couple that with field curvature and focus shift and a whole host of other things and things get complicated very, very fast.
So I'll repeat what I wrote below: I'm concerned about how many D800's I've examined that seem to have a real QA problem (that number would be six). To that I'll add, I was not concerned about how many D7000's I examined that seemed to have a real QA problem (that number would be one, out of a much larger sample).
People need to stop trying to blame the messenger and try to understand the message better.
That said, I'm also confident that Nikon will deal with any real problems that have shown up. Yeah, it sucks to have received a bad camera.
For the record: my experience with Nikon DSLRs is currently that three out of thirty bodies I've personally acquired (usually to research books) over the years needed immediate action by Nikon to fix something; further, three others needed fixing at some point down the line, none for problems I caused). Couple that with the cameras bought by my assistant and a few of my peers I'm in close contact with and the numbers stay pretty consistent at 10% or less. So I'll repeat: my concern is that in an admittedly random sample that is biased towards bad samples, I'm seeing more real problems this launch than before. It's not statistically valid, but it is a worrisome situation that I'll continue to monitor.
Should you avoid a D800? No. I've plenty of reports from very happy users, and if I ignore the AI-S arm issue on my D800, my cameras are pretty incredible, too.