Allow me to say approximately the same thing in a different way. Adam, I guess you would agree that there are some people who will almost certainly never found a really successful startup no matter how often they try. Do you really think Pr(you are such a person) is less than 0.1%?

Further:

• I think it's more like 75% of startups that fail.
• If you have no savings to speak of, then after a startup fails you will need to go back to "ordinary" work for a bit to raise enough money to start the next one.
• You may find it harder to raise investments if you have a few failures and no successes under your belt. (Again, if you have no savings to speak of you will need investment quickly.)
• Mean time to failure may be only 2 years, but (a) I suspect that the most likely path to startup success goes through startup near-success, and (b) nearly-successful startups probably take longer to fail, and (c) actually-successful startups typically take well over 2 years before they start delivering serious money.
• A startup that doesn't fail doesn't necessarily make you rich. You probably need quite a big success for that.

Let's suppose you don't care about (c) above because once you've got a successful startup you don't mind waiting, so your 20-year window indicates only how long you have to start an eventually-successful startup. Then in view of the numbers above I suggest an optimistic timeline looks like: some number of iterations of (1 year working to raise money, then 2 years for a startup to fail), then one of (1 year working to raise money, then 5 years for a startup to only-just-fail), then one of (1 year working to raise money, then startup succeeds). Then you have time for about 3 "quick" failures and one "slow" failure before succeeding. At a 75% failure rate, that's a lot less than a 99.9% chance of success. And if you fail on this path, you're now fortysomething, you have a long track record of failures, and you have no retirement savings at all. Good luck!

(When I say "an optimistic timeline", of course I'm aware that some people get there much quicker. But you have no reason to think you're in the lucky not-very-many-percent who will succeed first or second time around.)

Failing to successfully found a startup would be evidence for your chance of success being less than you had previously imagined

Hm. I agree that it might be evidence that you're not as good as you thought you were, but you also gain experience in failing that increases your chances of subsequent success. The relative strengths are obviously case-by-case.