There was a disastrous accident at a satellite construction facility a few days ago.
Because someone removed a few bolts from a fixture without documenting it, and someone else neglected to check for their presence, a weather satellite that cost almost a quarter of a billion dollars was severely damaged. The extent hasn't been reported but it will undoubtedly cost many millions of dollars to repair it. It may in fact be (in the words of the car repair business) "totaled," because after a fall like that it may not be possible to ensure the integrity even of the parts that don't appear to be damaged.
Disasters like this, in a mature field, never have a single cause. It requires a(n unlikely) combination of failures, which is why it happens relatively rarely.
Examples of this can be seen in any random perusal of aviation or diving magazines, in which accidents are described in detail, and they are invariably a result of a combination of things going wrong, rather than a single one.
Consider the Titanic. Just one thing going right (e.g., seeing the iceberg in time; not using a little-understood new steel that became embrittled by the temperatures of the North Atlantic in springtime; the captain understanding that he had to have forward power to have adequate steering control, which was not possible because he decided to reverse engines at the same time he was trying to steer away; hitting the berg directly rather than grazing it and tearing the large gash across multiple compartments; other ships being close enough, or the California receiving and understanding the radio messages; having enough lifeboats, etc.) and they would have been fine. But everything went wrong, and hundreds of people died.
Or the Donner Party. If there hadn't been an early winter, or they hadn't decided to take the "short cut," or...many other bad decisions had been avoided, they would have been safe in California before winter hit, as their traveling companions were.
Or Challenger. If the weather hadn't been quite so cold, if they'd understood the o-ring issue earlier, if they'd not been delayed by the previous delay caused by the desire to fly the Congressman, if only...
Or Columbia. If they'd been going to the space station, if they'd dealt with the foam problem sooner, if...if...if...
People have already commented on this particular accident, and I suspect that it will result in a change in procedure, and perhaps even in the design of the hardware that holds such satellites during ground assembly.
But I'm actually more interested in discussing why it's possible for the absence of a few parts worth, at most, a few dollars each to result in the loss of over two hundred million dollars.
Why do satellites cost so much?
The typical response from someone in the aerospace industry is "Space is Hard." We have to design the satellites for the harsh environment of vacuum, radiation, extremes of hot and cold. But that doesn't explain why similar systems designed for marine use (the ocean has extremely high positive pressures, and seawater is an extremely corrosive environment) can be built for orders of magnitude less.
The real reason comes back to transportation costs. Simply put, satellites are expensive because getting them into orbit is expensive.
Ocean-going vehicles, even underwater ones, can be delivered to their environment very cheaply--just drop them in. But when a launch costs a hundred million dollars or more, you want to make sure that your payload a) lasts a long time and b) works reliably, because if you have to replace it, you're out another hundred million dollars or so for another launch (on top of the replacement cost of the satellite itself). This translates into the use of extremely high-quality (and correspondingly expensive) parts, a lot of redundancy so that failures of individual components don't result in a failure of the system itself, careful attention to assembly, using highly-skilled and trained technicians to assemble it (which of course begs the question of how this accident occurred). All of this skyrockets the cost of designing and building a satellite.
A second factor, also related to transportation costs, is the need to accomplish the goal within a very restricted weight limit. Launch vehicles have a fixed payload capacity to a given orbit. If the satellites exceeds this by even a few pounds, it won't get to its destination. That means that people will spend huge amounts of money on it (e.g., careful packaging, exotic lightweight materials) to shave off ounces. This mentality of minimizing weight is so ingrained in the industry that engineers will occasionally (unthinkingly) spend more to take a pound out of a spacecraft than it would have cost to launch that same pound (typically a few thousand dollars).
And of course, both these cost-inducing factors cascade into a third one. Because launch, and hence satellites, are so expensive, and we live in a universe of finite resources, we don't do very many of them. It's not a mass-production assembly line, which is the only way to get products affordable to the masses--it's more of a craft, or cottage industry, with a few highly-skilled (ignoring the folks who leave bolts out of fixtures) artisans, building them on a boutique basis.
So when one of them falls over on the floor, it's not a matter of going and pulling another one out of the warehouse--it's more like commissioning another work of art from an artist.
Of course, some will cry, "But how many weather satellites do we need?"
An interesting question. As I write this, there is a hurricane bearing down on the mid-Atlantic states. We get continuous pictures of it (as long as the single satellite that can do so doesn't fail) because it is a geostationary one, and can take continual pictures of a specific site over the planet. But because it's geostationary, it's a long way away, so we can only get limited resolution, and a correspondingly limited understanding of wave heights and wind velocities, and temperatures. Those can only be understood via the primitive methods of sending piloted aircraft into the eye of the storm.
Imagine a different world, in which we had swarms of much lower-altitude satellites that could provide those parameters continuously, in all necessary detail, without hazarding aircraft and pilots.
That's a world that can only be provided by low-cost launch, and the corresponding low-cost satellites.
There's an old saying that "for want of a nail...a kingdom was lost."
For want of 24 bolts, a satellite was lost, and for want of a space transportation system that can provide vital space services, billions of dollars of property and lives, on the east coast of our nation, may be lost.
We have to come up with better (and cheaper) ways of getting eyes (both human and robotic) in the sky, and there's little coming out of our vaunted space agency that seems to propose to do so.
Rand Simberg is a recovering aerospace engineer and a consultant in space commercialization, space tourism and Internet security. He offers occasionally biting commentary about infinity and beyond at his Web log, Transterrestrial Musings.