Asymmetrical Information - Megan McArdle

04.19.13

The Most Dangerous Substance in America May Be Fertilizer

Common, cheap, and deadly, ammonium nitrate is a necessary evil

As of this afternoon, the confirmed death toll in the West Texas fertilizer plant explosion is 12.  60 people are still missing, and while some of them will undoubtedly turn out to simply have been out of touch with friends or relatives, I think we have to assume that a lot of those people are dead.  

This isn't the first time that a fertilizer plant has exploded.  

Sgt. William Patrick Swanton, a Waco police spokesman who is helping out the city's smaller neighbor, said Thursday there was no indication the blast was anything other than an industrial accident. The Texas Department of Public Safety said it could take as many as six months to determine the cause of the fire.

The worst ever industrial accident in the U.S. was also caused by an explosion of ammonium nitrate, as was possibly the case here, and also took place in Texas. In that blast, in 1947, some 581 people died aboard a ship docked near Texas City.

Dan Halyburton, a spokesman for the American Red Cross who toured West, described the damage as unlike anything he had seen before. "It wasn't like a tornado or hurricane. It looked more like someone had taken a grinding wheel and just chewed up walls and roofs." When people return to their homes, he said, "It's going to be really traumatic."

There's been some speculation that this might be a terrorist attack, given events in Boston, but you hardly need to posit a terrorist to explain the explosion.  Ammonium nitrate is not necessarily the best explosive in the world, but it's a strong oxidant, so if a fire or explosion breaks out near it, the ammonium nitrate will jump in and really get things going.  Also, it's cheap and fairly readily available.  That's why the IRA used to like making bombs with the stuff (they got it from sympathetic farmers) and why Timothy McVeigh used it for the Oklahoma City bombing.  

Since Oklahoma City, we've put much tighter controls on ammonium nitrate, but as West Texas illustrates, we still haven't made it safe.  I've heard some talk this afternoon that this is some sort of regulatory failure, but we don't know that (no, not even if the plant hasn't been inspected for a while).  Accidents happen even in well-regulated industries that are run by careful, caring human beings.  

Oh, no doubt we'll be able to identify something that someone could have done better.  But if you look hard enough, you'll always be able to identify a problem.  Take back surgery.  

Huh?  What does back surgery have to do with industrial accidents?  

Well, back when we started to get x-rays, doctors x-rayed the backs of people who came in with back pain.  Many of them had bulging, slipped, or otherwise abnormal discs.  Logically, it seemed like this must be what was causing their pain.  So surgeons started going in there and fixing the discs.  And indeed, in many cases, the patients reported feeling better.  

But then a bright doctor got the idea of x-raying the backs of people who didn't have back pain.  It turned out that many of them, too, had bulging, slipped, or otherwise abnormal discs.  But no pain.  This called the "disc abnormalities cause back pain thesis into question".  And to this day, we remain fairly fuzzy on the causes of back pain.  

When we look at a big industrial accident, we're like surgeons looking only at the x-rays of patients who come to us with a problem.  We look for abnormalities, and when we find them (as we almost certainly will, because there's no such thing as perfectly average in a back or a factory), we declare those abnormalities the cause of our problem.  Usually, we also declare that human negligence must be the root of the abnormalities.

In Normal Accidents, sociologist Charles Perrow argued that this was self-comforting twaddle.  All complex systems have failure points, because the complex interactions between the parts mean that it's hard to anticipate all the possible ways that something can go wrong. While it's true that you can always identify something that could have changed any particular outcome, that doesn't mean that you could have engineered failure out of the system.  In fact, many of the fixes would have added complexity, making the thing more likely to fail.  For example, at Three Mile Island, a fail-safe feature actually made the meltdown worse.  

Placing the blame on negligent regulators or operators makes us feel safer, because it gives us the illusion that there is some way to prevent all the problems.  But Perrow makes a pretty convincing argument that this is, in fact, an illusion.  And a dangerous one at that.

Ammonium nitrate is extremely common, and pretty dangerous, and also extremely useful.  Maybe we'll figure out that there was some easy way to have stopped what happened in West Texas.  But we should also be prepared to accept that this may have just been one of those things.  That is, that while we will undoubtedly be able to point to something that could have prevented this, we may not be able to point to a problem that was obvious before the explosion.