It's interesting that nearly everything that was wrong with the Space Shuttle was accurately predicted in this article. Both catastrophic breakups were essentially predicted (a booster failing and tile damage from falling fuel tank ice), as well as the huge cost overruns and performance shortfalls. I wonder what author Gregg Easterbrook thinks as he looks back on the Shuttle era? Having a quick glance at his article on Wikipedia, it seems he's written or said a bit more about space policy since the shuttle; I'll have to read up.
Even more impressive to me is that he wasn't an engineer but still made impressively technically well-informed article. I have no doubts one or more very smart and technically trained people helped him, still, it's amazing how much quality information is in one single article.
Unfortunately, Easterbrook is coasting on the reputation he built up from this one magazine piece from 34 years ago. It's basically impossible to read anything he writes nowadays without finding something either wrong or infuriatingly contrary-for-the-sake-of-contrariness.
The article I'd really like to read is about the "whistleblower" (or more of them) who gave Easterbrook all the info for that 1980 article. It would be very fitting to today's world.
Even after the Challenger disaster people were afraid. Feynmann didn't discover the O-ring problem so much as learn about it through tip-offs from NASA insiders (something he freely acknowledged later). They'd got him onto the investigation panel because he was someone with the public stature, and of course the personality, to stick his head up and report it.
It's great realizing, when reading a 35 year old polemic, that the chief issue with its criticisms was that it allocated WAY too little probability space to the outcome "worse than even we think it will be." For example, the debate in the article is between optimists who think the shuttle will hit 70 flights a year and pessimists who said 20. Nobody had their money on "four."
The article also describes Challenger and Columbia in chilling detail in advance. Quote: "The plan is, you die."
I've thought for 30 years that the shuttle design was terribly misguided.
The terrible problem with space flight is weight. Weight, weight, weight. It takes an enormous amount of fuel and rocket to push any kind of weight into orbit.
But the shuttle is not optimized for getting useful weight into orbit (i.e. payload, with the emphasis on pay). It's optimized for returning the rocket to earth. I.e. it has landing gear, wings, rudder, etc. All this is heavy, and therefore terribly expensive.
Step back a moment, and look at what we really need to bring back from orbit - it's the astronauts. Nothing else. All we actually need is a tiny capsule to bring the astronauts back, i.e. a heat shield and parachute. The rest can just be left in orbit.
Getting rid of all that weight then means the rocket can either be far cheaper, or it can launch much bigger PAY loads.
As for reliability, likely the most cost effective solution is to build a smaller, highly reliable rocket just for pushing astronauts up into orbit. Use another rocket, a big, heavy, unmanned one, and one that needn't be so reliable, to push up the stuff the astronauts need to do their mission.
"Step back a moment, and look at what we really need to bring back from orbit - it's the astronauts. Nothing else."
By saying that you're wilfully ignoring the key reason the Shuttle was designed the way it was; not to bring astronauts back but to return DoD satellites to Earth for servicing. That was only done a few times in practice, but drove the whole design process ( including the cross-range manoeuvring capability that led to wings being added).
The article pretty much destroys the concept of bringing back satellites. The only other thing we actually need to bring back is information - and telemetry does that just fine.
Not really - it destroys the concept of bringing back commercial satellites, which are mostly in geosynchronous orbits (and which, as it correctly points out, nobody wants back anyway). This may have been the story they were selling the public at the time, but the driver of this requirement was the Air Force and what they really wanted it for was to bring back military satellites, which actually are in Low Earth Orbit.
Of course, what they really really wanted was the capability to bring back Soviet military satellites (remember, SDI was just about to kick off). How they expected to pull this off without triggering Armageddon was never clear, but I guess a lot of people's budgets depended on them not pointing out the obvious insanity.
The deeper problem with the shuttle is that we need to bring the astronauts back because we sent them to space, and we sent them to space to do a mission, and that mission is to go to space. There is absolutely nothing a human can do in space that a machine can't do better, except die.
> There is absolutely nothing a human can do in space that a machine can't do better, except die.
I can think of two things.
The first is a bit morbid, but consider what happens if some country ever decides to weaponize space. Being able to have people up there who can use the weapons available to them and exercise their independent judgement even in the face of an earth-to-space communication failure is important. It's the same doctrine that's behind ballistic missile submarines equipped with nuclear warheads.
The second is less morbid-- the colonization of other planets. This is necessary to protect the human species from an extinction event. Maybe even one that resulted from the aforementioned submarines. Naturally in order to colonize other planets we need to study the effects of space on humans. In order to do that we need to send them into space.
All of that said, I also believe most of our current missions can be done much more cost effectively by changing them to unmanned missions. But you can't take your eye of the long game, which just like the rest of history, is manned and involves militarization and colonization.
> It's the same doctrine that's behind ballistic missile submarines equipped with nuclear warheads.
Did you see the article "Unsug hero of the nuclear age" a couple of days ago (yesterday?) on HN. That would probably shake your confidence in human reliability when facing the choice to launch or not to launch a warhead. Please have a look.
> The second is less morbid-- the colonization of other planets.
We can send humans in space, but the technology part that is lacking is still there: we don't have the "know-how" to colonize other planets, and before sending people in space there's a lot of work to be done in that field.
> And sending people into space is how we're getting that know-how.
No, not at all. There's a bunch of other things we need to develop BEFORE sending people in space. That's my point. Putting people in a rocket is easy, but having them live, on their own, on a different planet is something else altogether.
> There is absolutely nothing a human can do in space that a machine can't do better, except die
I would love to hear you explain how to perform, for example, research on the effects of long-term zero-gravity on humans using only machines.
Or to take another example: I suppose it would have been possible to develop robots to repair the Hubble, but the cost of developing said robots to a point at which they can perform without failure is going to be equal - if not greater - than the cost of actually sending someone up to do the repair work in the first place.
The first observation you made is a little tautological: the reason you'd study the effects of zero gravity on humans is so that you can more safely and effectively send them into space.
The second example is an extraordinary claim presented without evidence, since the cost of sending humans to space appears to dominate the costs of the space program.
We could have literally had 3 new Hubble's in orbit for less than the cost of all those repair missions. Repairing a space station is useful because unlike the Hubble it takes more than one launch to get it into orbit, but repairing a satellite is pointless.
We send machines to pave the way for humans, to see if it's safe and what precautions should we take.
If you said that only to humans in LEO, then I'd agree. There is no reason for a manned space station right now or humans riding along the satellites they launch. Maybe someday one would be useful as a buffer between long haul traffic and a new generation of shuttlecraft ferrying people and cargo to and from LEO for other ships to carry to deeper space, but that day only exists in our dreams of today.
He left the word "important" out, which makes it easy to score a debate point against him. What's the important thing humans actually end up doing in space that a machine couldn't do?
I'm out to say this oft-repeated sentiment is trivially incorrect.
While there are a range of activities which can be performed better by machines than humans, there is still a very wide range of actions that can't be performed better by machines. One would expect technologists to understand this better than anyone.
But the crucial thing here is that a given human is capable of a much wider range of actions than a given machine. True, this can be made up for to an extent by using a wide variety of machines. Though at some point the weight of said machines will outweigh the human and their required support systems. And even then, those machines will be limited to a small subset of actions that the human can take.
So what do humans have that machines don't? Versatility.
PS- If I wanted to score points, I'd say he said "absolutely nothing", which overstates his case and makes your response invalid.
> Could you perhaps identify a single important task that we'd want to do in space that can't effectively be done by machines?
I certainly can -- appeal to voters. If space programs didn't have to appeal to voters to get funded, we could do away with manned programs entirely, and good riddance. It's said that an equivalent unmanned program costs just 10% of a manned one.
Sure, for missions that aren't performed within a light-second of Earth. So far we have sent a total of zero humans far enough away to take advantage of their superior reasoning capabilities.
That's a very generous threshold for usefulness. For difficult interactive tasks—like, perhaps, performing fiddly repair work on a $2.5bn+ satellite—there is clearly some advantage to not having a ping of 2000 or even 350.
But unless the repair is going to be conducted by an AI—highly unlikely in the foreseeable future—then this is moot, as there will be a human brain in the loop in either case.
The reason the astronauts flip the switch to start them: "The experiments could not be made fully automatic because NASA policy requires that experiments on manned missions involve the crew." Politics, again.
Kind of a tautology isn't it? The crew must flip the switch because on manned missions the crew must flip the switch. Why not send up unmanned missions
Which doesn't change the fact that a human overseeing an experiment can do a better job than a machine, and could cause the experiment to not be lost in the first place.
Due to the versatility and adaptability of humans.
No, but it does change the fact that humans are likely to lose their lives in the pursuit of the experiment. Losing a machine is merely an inconvenience.
In case they have all the tools, spare pieces and the knowledge to debug and fix the machine. And in case that the fixed machine is safe, for example if you forget to put a bolt and now the system is not tight and some of the fumes escape, or it begins to spin or make vibrations. There are a lot of possible problems, and having the middle of the space with no escape pod is a bad place to try to be creative.
BTW, weight is also the enemy of airplanes, which is why you don't see airliners ever flying around without a full load of passengers or freight (i.e. PAYload) on them. Only governments can afford to fly them empty.
But it's a hundred, a thousand times worse with rockets.
Hello again! There are actually many, many airliner positioning flights that move around empty. It's a wonder airlines make any money when you see how inefficient they can be.
If you're watching FR24 or similar keep an eye out for callsigns ending in "P" for positioning.
To take your idea one step further: because it's so costly to get stuff into space, the less we come back with the better. The astronauts can come back, but once they launch a bunch of infrastructure, tools, etc into space those should stay there and remain usable. The ISS is a good example of this concept
The idea was floated around to boost shuttle's ET into orbit, and collect them. Flush out the noxious bits, add atmosphere and: living room. Modular sections for a rocket ship. Storage.
Google around for the military mission requirements that drove many of the key design decisions.
Basically, the Air Force needed a manned platform capable of launching from Vandenberg AFB and doing something over the Soviet Union (presumably recon). They later pulled out of the program, probably because spy satellites and recon aircraft got a lot better and rendered the mission obsolete.
For reference, the amortized cost of a shuttle launch ended up being $1.45 billion ($196 billion / 135), with the recurring (per-launch) cost as $450 million of that (?). For comparing with the article's figures, $1.00 (1975) = $1.53 (1980) = $4.18 (2011). (So you could compare, e.g. the $22.4 million 1975 estimate with either $108 M or $347 M).
You're linking to the BLS's CPI data, but you really don't want to deflate currency figures for the Space Shuttle using the CPI. Use something like the percentage of GDP: http://www.measuringworth.com/uscompare/
"Putting a man on the moon: [...] How much would that be today? If we used the CPI, it would be $160 billion, but this would not be a very good measure since the CPI does not reflect the cost of rockets and launch pads. ... [as] a way to consider the "opportunity cost" to society, the best measure might be the cost as a percent of GDP, and that number would be $453 billion. "
Using percent of GDP as an answer to the question of "How much would that be today?" is absurd, because our economy today is not only larger but more efficient at converting raw materials and manpower into working spacecraft. The question it really answers is something like "How daunting of an investment was that, in terms of today's economy?", though even that doesn't properly convey that such an investment made today would accomplish far more.
Okay. To some extent that's fair, but on the other hand "opportunity cost to society" is just a fancy way of saying the same thing as "how daunting of an investment was that" so the quoted statement is hardly in disagreement per se :)
Of course, there are also plenty of alternative measures available at the linked site which could be better than either percent-of-GDP or CPI. (I'm leaving their use as an exercise to the reader...)
To be fair, the late 70s inflation made it impossible to predict the price of anything well. My dad had some state GO bonds that paid something like 22% from 1980 or so.
Also remember that a big part of the shuttle program was to keep the engineers at NASA and the various contractors from Apollo in business. It was seen as as strategic need, just like how we build submarines and tanks even when we don't need them
The most interesting discovery for me was to compare: "cost projections: The analysts assumed that the shuttle fleet would stage at least 50 flights a year" with now known 135 launches in 30 years total which gives less than 5 per year.
The article is fantastically well researched and written.
It's curious how little is written about the X-37B that seems to be a successful reusable space vehicle.
It's unmanned, very much smaller than the Shuttle but seems to be of use for the US Air Force for something. It also stays in orbit for over a year at a time.
Required watching [1], straight from Dale D. Myers (Deputy NASA administrator 86-89) himself. An extremely fascinating tale regarding the importance of engineering constraints in mission critical designs.
I don't get it. The Shuttle project included a lot of things that had never been done before. So things didn't quite go to plan and some people died? But in the mean time, it performed useful tasks.
Predicting failure is easy, because something will always go wrong. There hasn't been a perfect anything built yet.
What is also surprising about these dredged up past articles, is that they all invariable criticize the engine (SSMB/SSME), even though the engine was never a major issue. Yes, the rebuilds between flights took longer than expected, but it was the first ever re-usable engine and is still an amazing achievement.
Feynman explains the problems with the main engines in the report about the Challenger disaster. They were a good candidate for the 3rd deadly accident if the shuttle program had continued.
The shuttle is representative of a class of problems that occur in almost every software project: uncontrolled complexity. It's easier to fix the symptoms in software than in other fields, so this issue runs rampant throughout all of IT.
https://en.wikipedia.org/wiki/Gregg_Easterbrook