I decided to start a new post based on comments at this one, in which in response to a comment of mine, a commenter writes:

“Escape systems can actually introduce more risk than they remove, and not be worth their cost and weight. There is a reason that airline passengers aren’t issued parachutes.”

This is a rather alarming line of reasoning, the kind that led to seven dead astronauts in 1986. Of course any crewed spacecraft requires some kind of launch escape system. If Challenger didn’t prove that, nothing will.

If so, then nothing will, because Challenger proves nothing of the kind. But once again, this is the fallacy of hasty generalization. Challenger, and the Shuttle in general, don’t “prove” anything about launch systems in general, either expendable or reusable — it’s simply unreasonable to draw huge extrapolations from a single example.

One of the main reasons for abandoning the Space Shuttle is the lack of a launch escape system. Otherwise, why give up such a massively capable vehicle in favor of a 1960s throwback capsule?

There were a number of reasons to give up the Shuttle, and the lack of an escape system was one of the weaker ones. It was costing too much to operate, and becoming untenable to continue to operate it with only three vehicles left in the fleet. Each flight is costing us in excess of a billion dollars now, and if (as some fantasize) the program is extended, at an even lower flight rate, they will cost much more. And each flight risks losing another vehicle, and if that happens, continuing to operate it is simply infeasible. Cold-hearted as it sounds, we have a lot of astronauts, but we have only three orbiters, and the cost of replacing them is far beyond what it would be worth.

There was nothing intrinsically wrong with the basic philosophy behind the Shuttle — that the cost of an escape system, in dollars and weight, would exceed its value. The failure was in not making the vehicle as reliable as intended.

Now could the Shuttle have been made that reliable? Probably not, in the first attempt. But then, the first attempt should have been a much less ambitious system, and there should have been evolution and lessons learned from it. In a sense, Shuttle was a bridge too far, as Apollo was — a huge government program that became a self-fulfilling prophecy that we could afford only one.

Can any launch system, with today’s knowledge, be made sufficiently reliable as to not need an escape system? I think that a reusable one probably can. Recall that both Challenger and Columbia were lost as a result of expendable components. Can an expendable vehicle be made that safe? That’s more problematic, because no matter how good your quality control, every launch of an expendable is a first launch. So it may well be, depending on what level of safety you demand, that you’ll need an escape system as a backup.

You will probably always be able to find people who will ride a rocket into space whatever the risk, but if crew losses start to mount, and the public perceives that safety concerns are being overridden by bean counters, the sources of funding for manned space travel, public and private, will quickly dry up. If the Challenger and Columbia disasters had happened, say, four years apart instead of 17, U.S. manned space flight would have likely ended then and there.

That’s an interesting counterfactual, but there’s no way to know that. Many always postulate that the next disaster will kill human spaceflight, just as they claim that a single death of a passenger will kill the space tourism industry. Such claims are made in defiance of history and human nature. We killed lots of people in aviation, and still occasionally do, but the industry survives.

I know space flight will always be risky (and hence never remotely as safe as airline travel) and that overemphasizing safety will kill exploration beyond LEO, but you have to provide some kind of escape option during the most risky phase of any mission– the launch. I’m sure Elon Musk understands this and has no plan of putting crew into a Dragon capsule without an escape option.

I don’t know what Elon understands and plans, and you may be correct, but I think that it would be driven by the customer. For instance, if I were buying rides from him, I might say “…skip the escape system — I need the payload, and don’t want to take the risk (for example) of it not separating properly. I have insurance policies for my crew.” If NASA insists on an escape system, it will likely be a political decision, not necessarily one dictated by a rational probabilistic risk analysis (I’ve never seen the PRA for the Orion LAS). I know that when I was doing hazard analysis during Phase II, it wasn’t even a question we were supposed to ask — that Orion would have an escape system, was a given. But an escape system introduces a lot of new hazards into the launch, many of which can bite you on an otherwise nominal mission (e.g., failure to separate). I know that the bureaucrats are afraid of being called before Congress and having to testify that they killed astronauts because they didn’t have an escape system. But they never consider the possibility that they might be called up on the Hill to explain how they killed them only because they did. And it’s not an impossibility.

In any event, if NASA insists on an escape system, it will be a decision implicitly premised on the belief that what we are doing in space isn’t important, otherwise we’d be willing to risk crew on it. If Shuttle proved anything, it is that safety is not a binary condition, and no matter how many billions we spend in an attempt to never lose an astronaut, we’re still unlikely to be successful, particularly with a government program. So we might as well just accept the risk, and do a lot more at less cost.

[Update a few minutes later]

One more point. Several space passenger vehicles are under development, including SpaceShipTwo, Lynx, and whatever Armadillo and Masten are planning. We know that the first two don’t have a crew escape system — they are designing for reliability (actually, I only know that’s true for SS2, I’m not sure about Lynx). Armadillo and Masten may have plans, but I’m not aware.

Yes, they don’t go to orbit, but there’s nothing magic about that. There is no bright energy line on which one side an escape system is required, and on the other it is not. Every vehicle is designed to meet its requirements, one of which is some level of safety, but if too much is spent (in either dollars or weight), the vehicle design or business case may not close, and you may not even get more safety for your dollars. Fortunately, we now have a number of competing designs and can let the market sort it out, rather than a dictate from on high by the kind of idiocy represented by the ASAP.

[Update a couple minutes later]

One more. I would dispute that launch is the riskiest part of the mission. I think that Columbia is a rejoinder to that. When you look at the total risk of a lunar mission, it’s a misallocation of resources and a defiance of rational systems engineering to put so much of them into reducing the risk of launch. But misallocating resources is what a politically driven institution does.

[Another update a while later]

OK, yet another point. I wrote: There was nothing intrinsically wrong with the basic philosophy behind the Shuttle — that the cost of an escape system, in dollars and weight, would exceed its value. The failure was in not making the vehicle as reliable as intended.

I’ve made this point before, many times over the years, but it bears repeating, particularly for new readers here. Shuttle wasn’t just insufficiently reliable to carry crew without an escape system — it was insufficiently reliable, period. This is because, as I said, we have a glut of people willing to fly into space, and a shortage of vehicles with which to do so.

Any reusable vehicle must be highly reliable, regardless of whether it carries crew or not, or it becomes unaffordable (as the Shuttle did). This is why the notion of “human rating” a reusable launcher is nonsensical. It is the value, and replacement cost, of the vehicle itself that drives the reliability, not its payload, whether human or otherwise. It’s also the reason that it makes no sense to put a crew escape system in one (again, the decision to not have one in the Shuttle was the correct one). If your reusable vehicle is so unreliable that an escape system is required, it is unaffordable to operate, period.