Common Tragedies

Thoughts on Environmental Economics


Posted by Daniel Hall on June 20, 2008

Several bloggers that I like and respect are mocking this Gregg Easterbrook piece about asteroids in the June Atlantic Monthly. But I am going to go against the grain and defend the article.

First off, however, I’ll acknowledge that the headline tag — “The odds that a potentially devastating space rock will hit Earth this century may be as high as one in 10.” — is overblown.* This has been the subject of the most vigorous objection (see Brad DeLong) so in this sense the critics are right. But I’d argue they’re missing the forest for the trees: the specifics of Easterbrook’s argument may be oversold, but the question he poses in the second half of the headline tag — “So why isn’t NASA trying harder to prevent catastrophe?” — is exactly the right one to ask.

The critics miss two key things: First, the ‘expected value’ calculation for an asteroid impact could show a very large cost indeed, even if the vastly most likely outcome is no cost at all (i.e., no asteroid). And second, we must consider both benefits AND costs… if preventing asteroid impacts is very cheap, it may well be worth doing.

Regarding the first point, let’s do a simple back-of-the-envelope calculation. We’ll assume the value of a statistical life (VSL) is a (very conservative) $1 million. An asteroid impact that killed 100,000 people would thus cost $100 billion. Easterbrook’s tagline seems to imply that in any year there is about a 1-in-a-1000 chance of such an event, which would imply we should be willing to spend $100 million per year for “asteroid defense”. This would be true even though there was a 90% chance we’d reach 100 years in the future and that money would have been a total waste every single year. Meanwhile, an expert in Easterbrook’s article thinks the whole asteroid diversion thing could be done for $400 million — total.

Now you may disagree with the numbers in my example (e.g., you think such an event is far less likely) but the specific example is not the real point. The important intuition is that one large catastrophe can dominate your expected value calculation.

Of course the actual calculation you would want to do is a full expected value calculation — the sum of all possible futures (their probabilities times their costs). This means counting up all the very likely worlds with no asteroids, the very unlikely worlds with asteroids that wouldn’t be so bad, and the one-in-50-million type asteroids that could cause mass extinctions. But just dismissing the problem out of hand because it hasn’t wiped us out yet strikes me as irresponsible.

The second mistake some critics have made is comparing asteroid impacts to other types of disasters without considering whether we have leverage over these disasters and what this leverage costs. This guy, who mercilessly shreds the article, is quite guilty of this:

Easterbrook is panicked by the thought that maybe-once-a-century event like Tunguska could occur over the negligible percentage of Earth’s surface which is covered by dense cities. He writes:

The blast had hundreds of times the force of the Hiroshima bomb and devastated an area of several hundred square miles. Had the explosion occurred above London or Paris, the city would no longer exist.

A comparable destructive energy can be expected from a good-sized hurricane or typhoon, which have a tendency to strike coastal areas, a popular location for cities. … Or, if you don’t like that analogy, you can also get a much larger, and similarly rapid, yield from an 8.0+ earthquake, of which we get about one a year. Localized disasters of this sort happen hundreds or thousands or times more frequently than similarly-powerful asteroid/comet impacts, and yet somehow we don’t characterize them as the deadliest threats ever.

… There have been extinctions of large numbers of species in Earth’s history – we have an idea of when they happen, and how big they are – and some of them may have been precipitated by some kind of catastrophic extraterrestrial impact. How does this threat compare with the expected results of human-caused climate change?

The widely-accepted science on global warming, much like the highly speculative situations Easterbrook is fantasizing about, would have similar effects on the planet – mass extinction, starvation, disease, and massive physical destruction. According to a 2004 study in Nature, mid-range estimates for global warming could cause the extinction of 15-37% of all plant and animal species. The last extinction event which even approaches this magnitude was 33.5 million years ago, which may or may not have had something to do with some kind of extra-terrestrial impact. Sixty-five million years ago we have a mass extinction likely caused by the impact of one (or many fragments of an) asteroid, wiping out perhaps 30% of all species. Before that, we have to go back 200 million years. So, a survey of the last 200 million years tells us that at most we have extinctions from all causes on the order expected from global warming every 60-70 million years.

I have a couple of responses to this:

1. Yes, earthquakes and hurricanes are more likely than asteroid impacts but we don’t have a good way of stopping or diverting them. (Not that we shouldn’t be researching this.) This means we must either pay the (very large) opportunity costs of not inhabiting hurricane- or earthquake-prone areas, or pay the costs of building more resilient infrastructure and rebuilding destroyed infrastructure. (Anyone remember what Hurricane Andrew cost? Maybe we shouldn’t have rebuilt Florida but the fact that we did gives you at least a hint about the opportunity cost.) Paying for asteroid defense, by comparison, looks cheap.

2. Yes, climate change is far, far more likely to be a problem than asteroids but you have to compare the proper counterfactuals. For asteroids we are talking a world with a very tiny risk of asteroid disaster versus a basically identical world ($400 million is nothing) with zero risk of asteroid disaster. For climate change, unfortunately, we are talking about a world perhaps warmer by 6-8 degrees Centigrade versus a world that is perhaps only warmer by 3-6 degrees but where we have literally spent trillions of dollars to do this. There is no free lunch. This doesn’t mean I don’t think some level of climate protection could well be the best lunch we ever buy, but let’s be honest about what we’re spending and what we’re getting.

My ultimate point is that the article has value because it highlights our screwy priorities when it comes to spending money on space. Why does almost all our $17 billion NASA budget go to getting humans in orbit and bases on the moon and Mars? We need a more Earth-centric NASA. It could be doing far more good developing enhanced Earth monitoring systems — satellite data is going to be invaluable to understanding climate change in the next century — and yes, protecting us from space debris. Anyway, I think Easterbrook’s article is thought-provoking and you should read it.

*Although I’d suspect Easterbrook would fall back on the definition of “potentially”.

8 Responses to “Disaster”

  1. Brad DeLong said

    You write that the very first thing the *Atlantic* has to say is wrong: “I’ll acknowledge that the [*Atlantic’s*] headline tag — ‘The odds that a potentially devastating space rock will hit Earth this century may be as high as one in 10.’ — is overblown.”

    Shouldn’t your next move be to point us to somebody writing about impacts and their risks who knows what they are talking about, is telling it to us straight, and is trying to inform–rather than misinform? Why does “Easterbrook” appear eight times in your post?

  2. Daniel Hall said

    Brad, here’s a Wikipedia page and here’s the page for NASA’s near Earth object program.

    But I don’t think any of this really matters for the points I make in the post. (Points which, I’d add, you choose to ignore in your comment.) The specific collision risk numbers aren’t the valuable part of the article. The article is useful because it raises questions about our priorities regarding disasters and, more pertinently, NASA’s funding.

  3. Hi, good reponse. A couple of points:

    1. In addition to wildly exaggerating the chances of a “catastrophic” (by various definitions) asteroid impact, Easterbrook thinks that the NASA should have this as one of its top priorities. (You have to wade through the embarrassing middle section of the article to discover this, the one where Easterbrook seems highly sceptical that NASA doesn’t set its own priorities.) Manned space flight is a horrible boondoggle, so I’m not going to try to defend it, but there are actually useful/informative projects (as you yourself imply), defunded over the last eight years, which could use the money, rather than shifting it from boondoggle A to quasi-boondoggle B. I think this is the proper perspective. I’m not saying we shouldn’t spend anything on asteroid detection, but we need to remember that such an event is hugely, hugely unlikely.

    2. I think your point #1 is weaker than you think it is. No, we cannot (AFAIK) prevent earthquakes, hurricanes, volcanos (perhaps a better analogy than the ones I chose), or any of the other catastrophes we know about and accept. But we can certainly greatly reduce the damage they do, through early warning/evacuation, building codes, etc. See Katrina, etc. Nobody panics that we need to turn our funding priorities around to counter these much, much more likely threats, which are not only real but regular and inevitable, for just this reason – we understand the risks involved, and we’ve made peace with them. (Understand that I’m not saying we should (or shouldn’t) spend more (or less) than we do (or don’t) on disaster prevention, I’m just making a point about our attitude towards them.) Meanwhile, despite the fact that the disasters Easterbrook describes have not ever happened in the history of the human species, we are supposed to place them in some new, especially scary category. Fear of the Unknown, rationalized. It’s nonsense.

    3. As for point #2, I agree that asteroids-to-climate change is apples-to-oranges (I don’t agree that we have properly calculated the cost/benefit of global warming prevention, but I’m running long already). One is going to happen, one, to a very high degree of certainty, isn’t. The proper comparison here is, again, attitude. Gregg Easterbrook has spent the last decade-and-a-half calling people who worried about global warming – scientists, and those who can read the very-easy-to-understand reports that these scientists write – “alarmists”. I say this not as someone who has some special understanding of climate change, merely as someone who can read English, and finds it irritating that people who are parading around as public “experts” either can’t, or – as I strongly suspect – won’t. The title of my post was “Gregg Easterbrook is … an idiot”, which is a 100% true and verifiable statement, not “We have nothing to fear from asteroids” which is only a 99 44/100% true statement, far below my gold standard for truthiness.

    Again, thanks for the good response.

  4. Daniel Hall said

    The Editors,

    Thanks for your long and thoughtful comment. I think we are very close to agreement on many issues, so I just want to offer a few additional comments on your points:

    1. It sounds like we’re largely in agreement here.

    2. I agree that we should do much more to increase our resiliency to catastrophes (better building codes, disaster warning and response, etc.). My post was merely trying to remind readers that such policies have real costs (as well as benefits) and these costs are likely much, much bigger than $400 million. Meanwhile, I have two comments about your claim that the “disasters Easterbrook describes have not ever happened in the history of the human species”. First, these scientists argue that a Tunguska-like event may happen on Earth about every 300 years. Now the next one may not be a big deal but on the other hand just because the last few space rocks were not a big deal for 0.1-1.6 billion people doesn’t mean that 6-9 billion people are not going to have a problem with it. Second, and far more importantly, when talking about very, very unlikely but extremely devastating events (i.e., extinction impacts) it is specious to argue that such things have never happened, because survivorship bias affects our observation. The fact that the human species has not destroyed itself through nuclear annihilation is not a good guide to the probability it will happen in future.

    3. I don’t have a lot to add, although I’d guess I’m less concerned about attitude than you are. A writer’s history on other issues may increase the skepticism with which I approach his work, but it doesn’t preclude the possibility he may be right. And it certainly doesn’t mean that there isn’t something I can learn from him (even if, as in this case, the specifics are sloppy or wrong).

  5. On #2:

    1. Suppose we get 1 Tunguska every 300 years. Suppose ~3% of the Earth’s surface is covered by cities. We’re talking 1 every 30,000 years – much longer than recorded history, even ignoring the unquantifiable uncertainties here – for an event which probably wouldn’t be substantially more destructive than a very large earthquake or tsunami, let alone a Krakatoa-type eruption. Everyone can choose how scared they want to be about this, but the point is that, by this most generous possible interpretation, we are a full order of magnitude from the promised 10% per century.

    2. This argument is completely false. The geological record is readable going back far, far before our species existed. We know how frequent mass extinctions are (from all causes) over the last several billion years. We have a solar system full of large, solid, atmosphereless objects from which we can deduce how frequent large impacts are, and have been, over time and space. The answer: both extinction-potential impacts and mass extinctions happen in our neck of the cosmos on a timescale of several tens of millions of years, and these frequencies both appear to decrease with time. Again, choose your own reaction to this news, but the probabilities are knowable and very, very, very small over any human timescales.

    #3: You are more generous than I, then. If Easterbrook were a scientist, or a member of any profession where one is judged on the rigor of one’s work, he would be flipping burgers. As it is, he is probably the most published science writer in the country, and consistently one of the worst (he is, among other things, a major proponent of creationism, and spends a lot of times chiding theoretical physicists for covering up the existence of the soul). It is correct, in a Platonic sense, that we should judge all arguments strictly on the merits. It is difficult, as a member of society, to see someone rewarded for consistent crackpottery and hold my tongue.

    In Easterbrook’s defense, however, I will say that his football writing is 100 times worse.

  6. 1. s/b “1 every 10,000 years”, of course.

  7. Daniel Hall said

    1. Ok, taking your number of 1 city impact every 10,000 years and my (very conservative) VSL estimate of $1 million then we should be willing to spend $400 million if we think a city impact would kill 200,000 people or more. (I am assuming a social discount rate of 5% and thus spending $20 million per year forever is equivalent to spending $400 million once today.) Readers can make their own judgments at the appropriateness of these numbers. I am not arguing for stay-awake-at-night terror, however; I’m just saying it seems like the benefits might outweigh the costs.

    2. This is a valid criticism and a good point. I would point out, however, that my argument in the general sense is not “completely false” (it is in fact completely true) but rather that it is not applicable. Or at least not applicable in your view. I say this because I would put more emphasis on the uncertain nature of evidence and the possibility of our error in interpreting that evidence. Your answer makes it sound like you ascribe probability=1 to the completeness of the current scientific evidence on geologic extinctions and impact frequencies. I agree that the current estimates are the best we have but I think there are decent odds (p=0.3? 0.5? 0.7?) of significant revisions (in either direction) as we learn more or improve our analysis. If you allow for this type of uncertainty then my argument goes back to being partially applicable (depending on the probability distribution you ascribe to revision).

    Finally, have I mentioned yet that I’m extremely optimistic that a we are not going to get hit by a devastating space rock? This has very little to do with the probabilities and everything to do with the fact that we are very, very likely to see it coming a long ways in advance and it will be relatively cheap to do something to avoid it. (Compare this to pandemics or climate change and you see why it’s such an easy problem.) Indeed in the long run the much bigger problem may be that we could become so good at precisely diverting space debris that it could be used as a weapon. (Interesting science fiction terrorist plot for all you young writers out there.)

  8. 1. I’m not objecting to spending some reasonable amount of money, either – just to blowing it out of proportion. There are real and present dangers to address, and other useful things we could do in space besides watch asteroids miss us.

    2. I couldn’t say how uncertain the extinction/impact evidence is, but I’d suggest that when you are talking about the biggest, most obvious events, it’s probably pretty settled. A mass extinction would leave a big delta in the fossil record, a big impact would leave a big crater. Maybe Tungaska events happen more often then generally assumed, perhaps as often as Easterbrook’s scientists believe. However, it’s hugely unlikely, IMO, that we’re going to discover that we missed a bunch of mass extinctions, or that there are twice as many huge impact craters as we thought there were. I’m not a geologist or paleantologist (I can’t even spell it), but getting this wrong seems like an astronomer misjudging the number of stars in the Solar system. Again, not an expert, but I don’t think something can be simultaneously Earth-shattering *and* difficult to detect.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: