There is no such thing as "doesn't happen" or "zero chance of something happening". All probabilities of falling debris injuring a human are calculated, expressed in mathematical terms and kept within agreed bounds. FAA will not issue a license otherwise.
Bill,
The question raised by this and the other articles is whether the calculated risk level is the actual risk level.
Risk models require a lot of assumptions, along with a lot of data. There is uncertainty in both, and often errors, too. And, the models are usually "incomplete" - that is, they miss some parts of the total risk - sometimes the biggest parts.
For example, the U.S. Atomic Energy Commission long "guestimated" that the probability of a nuclear fission reactor core melting as "one in a million" per reactor per year of operation, but, when they actually contracted for MIT to make a risk model, it came out much higher than that. After fixing some of the biggest contributors to the total in the calculation, the result was still 10 times "one in a million". Fast forward some decades, and we are now in an era when there are lots of risk models being used for lots of nuclear fission power reactors, and those reactors and their models have been modified to produce probability estimates that are again "one in a million" or even a few tenths of that.
However, we have now had enough actual power reactors operate for enough years, and enough of them have actually melted their cores, that we now have statistical evidence that the current model calculations are too low by a factor of about 1000!
The lesson to take from that is not that we need to multiply other risk calculation results by 1000, but rather that we need to do some serious
reality checking on the results of such models. The usual calculation of uncertainty ranges (e.g., 95% confidence that the real answer is within +/- x%) are typically not reliably bounding the real answer. The calculated uncertainty is usually too small.
And, then there is the question of what is an
acceptable answer. There are now about 8 billion people on Earth. IF the "acceptable risk" is limited to "one in a million
per person per year", then it would be "acceptable" to have falling space junk kill 8,000 people per year. But, is that really socially acceptable?
On the other hand, requiring that the space junk have a probability of
less than one in million per year of killing a single person on Earth would be unattainable, and would put the individual risk at some absurdly low value like one in 8,000,000,000,000,000 (8 quadrillion for U.S. definition of "quadrillion").
Even the frequency of killing less than one person per year may not be the reality for much longer, assuming it is, now. That would require the probability per person to be less than "1.25 in 10 billion per year".
And, that would need to be
the total for everything launched, all taken together. If the FAA is doing its calculations only on a per launch basis, then the total is really not being addressed. As humans increase the launch rate to hundreds of times the historical rate, it is important to consider what the total is becoming.
So, what is realistic for the amount of risk to be allowed from falling space junk? Currently, it is surely smaller than the risks of crashing airplanes to people on the ground, because we have statistics that prove that much. But, we also have procedures and regulations that are intended to reduce aircraft crash hazards to people on the ground.
It is a social/political question that will be addressed in a social/political manner. And, that will probably be steered more by emotion than quantitative risk assessment, once somebody actually gets seriously hurt or killed by falling space junk.