Let's see - 50M MTBF for a mechanical switch versus 10M MTBF for a rubber dome. We have a real life example of 1,226,400,000 keystrokes for a 101 key keyboard (the IBM Model M). 12M per key. Sounds about right given the varying key frequencies.
What keys specifically wore out? Statistically should look something like this:
THIS is how you do a A+ Science paper. Not just post on a Internet Forum.
Damn T-Ball Generation, expects an award for showing up to the game.
Believe it or not, not those keys. Bear in mind, my typing is atypical, so results not guaranteed. The first keys to go south were actually the numpad, bearing in mind, that just that section probably snuck in another half billion keystrokes or so. (I was only estimating words, and probably underestimating at that.) Matrix failure started at left shift - presumably due to all the programming work done on it, primarily in FORTRAN and SQL. Quelle surprise there, right? I also have a tendency to lean into it. If the backing plate is holding, leaning will wear things out faster. I didn't bother to, but should have counted how many stabilizing rods I replaced. It was more than a few.
Probably the biggest factor by far in killing any Model M is the actual physical typing habits, not the keys pressed or any of that. They're not exactly keyboards you can use kid gloves with with for obvious reasons, but if you have habits like leaning or riding, it can shorten the life significantly. The specific typing habits WRT keys pressed don't really have any significant bearing on it.
I could do the whole science report thing - I still have that keyboard around here somewhere - but I don't have time to disassemble and retest. Now, that said, I've had two '91's that didn't last 3 years before developing numpad problems (from different sources, just to be sure.) And a '89 which just may have 2B+ keystrokes on WASD which chews up caps but keeps working. So there is a tremendous amount of variance year to year and model to model even though there shouldn't be. Those are the ugly ones; the isolated key problems.
That's the other thing about Model M. You're not dealing with a keyboard - you're dealing with a mechanical engineering project. When you bend that backing plate, you actually affect the lifespan of the entire keyboard. When you lean/roll onto the left shift key hard, you're actually damaging mechanical components over time. When you stretch it out over 12 years, microfractures become visible cracks, so on, so forth. Biological contaminants contributing to corruption (you sneezed, don't lie), and so on. That isn't to say that a Model M is not almost literally a battletank - they engineered these things to be bulletproof, after all. But there's only so much that can be done in the hands of the unwashed masses (currently under the symptoms of an allergy attack.) Every M is not created equal, nor is every user created equal.
Rip's pattern, by the by, is exacly where I expect it to be for traditional typing. Specifically, email, writing fiction and non-fiction, etcetera. But very much not the layout for someone who work sin for example, hexadecimal. (A-F,0-9) which means that you use 16 keys common, plus 6 keys modifiers, which means a string is in fact 12 keypresses. Same deal with C; you'll heavily lean on {} [] and <> and ., which you don't see outside of programming. Also a lot of \. System Administrators, you'll see a very different pattern centered around short commands that go long.. ls -l -a -x -v - stacking in spacebar and hypen.
