This is a very interesting thread! I think you may be on to something with the difference categories of failure. Certain components are definitely more predisposed to having certain *kinds* of failures.
Mechanical devices are obviously going to suffer the most from wear and tear, and are most likely to be damaged by transportation or physical abuse and neglect. Floppy disks were notorious for just going bad without warning, and I've had one or two floppy drives go on me as well (although the one I have in my current rig has survived for many years now without a problem. Of course, now that I've said this it'll explode tomorrow.) All hard disks and optical drives *will* fail, if pushed long/hard enough. The question is simply when, and/or how much of a push they need to go over the edge. Monitors also degrade, and not always gracefully. CRTs are also very vulnerable to damage from transit (such as falls -- even just a couple inches -- while being moved, etc.), and are unlikely to last more than 4-5 years even in the best conditions without showing serious signs of wear. Somebody brought up printers, which I've never found terribly reliable (at least inkjets; lasers are usually pretty good, and I don't even recall having a problem with my old dot matrix one even after years of fairly heavy use). I've never had a problem with a keyboard (even ancient, heavily-used ones seem to hold up well), but non-optical mice definitely start to break down after a while. Especially if the owners never clean them out.
Solid state devices (CPUs, RAM, motherboards, video cards, etc.) should essentially last forever (or at least far longer than their functional life expectancy) if taken care of. The only video cards (2) I've personally seen go suffered fan failures and overheated. One ran for months like that, just occasionally crashing under full 3D load, until I cracked open my brother's case and saw the problem. However, these devices seem most likely to have inherent defects or, in the case of RAM, to just not function up to spec. Buying high-performance memory still seems to be a crap shoot these days. Circumstantial evidence from the tech support boards here says that cheap motherboards (such as ECS-brand ones) have widespread quality control problems. They're also easy to damage in transit if not properly packed, or if UPS decides to drop the box a few times.
Power supplies definitely still have lots of quality control problems, at least the ones from certain companies. The worst are off-brand ones that claim to provide 400 or more watts of power, but that fail miserably when actually put to the test. Again, a quick look at the tech support forum will support this. However, once you install one, and it's working, there's little that will break them. Power surges (or just dirty power over time) seems to be how most of them go.
User error is always a problem; people can and will screw up anything if you give them a chance. The biggest problem seems to be things that go in sockets (CPUs, RAM, video cards, PCI expansion cards). I've seen plenty of photos of RAM that was forced in backwards and went up in smoke, and it's not hard to bend a pin on a CPU if you don't handle it carefully, or to fry one with static electricity. CPU heatsink installation is also something that seems hit-or-miss.