You went to six decimal places in your explanation of mutiple failure risk, but didn't factor consequences in at all apart from possibly via a vague statement that it's up to the individual
Right, because consequences don't lend themselves to numerical probability, and failure rates do.
I'm not sure why this is an issue or what you are driving at here... Are you saying that three pony tanks are actually better than one? Are you saying that one pony tank is unnecessary? Seriously, all I was saying was that the reason that redundancy in critical systems is useful is because the statistical likelihood of two systems failing is much lower than that for a single system. Is that a controversial point?
I went to six decimal places not for accuracy, but to demonstrate a very small probability (one in a million), and to demonstrate the calculation of relative risks. The actual results are not accurate of course, because we don't actually have the failure rates of an average first stage. But there is no reason we couldn't have them, if there was, say, a company or agency that were to compile that data. It is a real number that actually exists. Just like I don't know how many feet it is from my home to the nearest McDonald's, if I needed that number, it is certainly discoverable.
No matter how much research money and time you were throw at the problem, however, you will not be able to conclude that death is X times worse than the bends, or that an arterial gas embolism is Y times worse than a moray eel bite, or whatever. That is why you can't precisely calculate consequences, they are always in the eye of the beholder.
Many non-divers would never SCUBA dive for one reason - fear of sharks. That outweighs any possible benefits in their minds. So their decision is clearly based on their own gut feelings, and even though they are at more risks of being killed by a bee sting if they stay above the water. Their own unique and subjective assessment of consequences would be what guides their risk-benefit analysis.