Sunday, October 15, 2006

Lancet redux

Summary: I think the new Lancet study is basically irreproachable, except that I fundamentally disagree with their interpretation.

Well, as it did before I started blogging, the Lancet has published a study concluding that the US invasion of Iraq has been a bloody mess:

A team of American and Iraqi epidemiologists estimates that 655,000 more people have died in Iraq since coalition forces arrived in March 2003 than would have died if the invasion had not occurred.
I'd love to see something wrong in the statistical procedures, but I don't. (I haven't gone over it as carefully as the first time, but people I came to trust in that discussion, like dsquared, have...and I figure I've seen enough.) With the earlier analysis there was at least some fuzziness from the limited number of death certificates, but that seems to me to be gone. It remains possible that the authors are lying or that the data-collectors were lying or that the data-providers were lying, and there could be all kinds of incentives for this, but that doesn't strike me as very interesting. zeyad of Healing Iraq suggests that their number may be as much as twice too high (not only the "55" but the "6" may be noise), but really that's not very interesting either. As Tyler Cowen quotes Steve Sailer:
"The violent death toll in the third year of the war is more than triple what it was in the first year." That to me is the more telling estimate.
Yes indeed.

Maybe I should feel validated. I said before the war, in a statement of support for that war, that it was option (c):

(c) "Regime Change"'s not a nice option, it will end us up in a big long-term mess, Bush is not a good President to be running it,
but there was a "but":
but it looks to me a lot less bloody than options (a) or (c)...
Unfortunately, I still haven't changed my mind about that. I guess you can call me a techneocon. Whether I'm thinking about Social Security or about global warming or alternative energy, I tend to think that the future will be dominated by increasingly rapid technological change. Mostly I think of that as good: I'm a Singularitarian. Yay! But I see our survival into that good future as a race: as TPM Barnett noted in the Knoxville Gazette lately, we are roughly ten years from genetically engineered threats from non-governmental organizations -- and roughly ten years from being able to detect and maybe deal with them. Sometimes I sit and think about that...I'm not at all sure of Barnett's ten years, of course, we may have twenty (or five, but I think the curve stretches out to the right), but it's not just genetically engineered threats that matter. If the RepRap project works (and it will, or some successor will) then we will have cheap desktop manufacture by Instapundit's Army of Davids, just as he says. I bet we'll be able to manufacture self-reproducing machines that are really hard to stop. (I'm not thinking about tanks, just now; I'm thinking about artificial ticks that can manufacture botulism -- the ticks would not be self-reproducing, but they would be manufactured by self-reproducing machines which release little tick-laden hydrogen balloons every now and then. What fun...) Ditto for nukes, though the raw materials there provide some constraint.

We may not be able to stop the dictatorships that breed terrorism (including direct support, as in Saddam's case.) If we can't do it, we will not be thinking of 655,000 as a large number of dead. (Okay, I'm the guy who didn't expect civilization to last till the year 2000. But then, again, I think we survived by luck and the grace of Stanislav Petrov.)

So I support the war, sort of. Well, I support the initial "Big Bang", as Barnett describes it:

Big Bang: refers to the implied (and sometimes openly voiced) strategy of the Bush Administration to trigger widespread political, social, economic and ultimately security change in the Middle East through the initial spark caused by the toppling of Saddam Hussein's regime in Iraq and the hoped-for emergence of a truly market-based, democratic Arab state. ... a direct, in-your-face attempt by the Bush Administration to shake things up in the stagnant Middle East... The Big Bang was and still is a bold strategy by Bush, one that I support.
Me too.

So, why did I think of Bush as a bad president to run it? Partly I was wrong: I never thought he was particularly stupid (he probably wouldn't have done too well in my computer science classes back when I was teaching, but I think that's a minor flaw in a President), but I thought he was much more of a crony capitalist than I now think. Mostly I think I was right: he's an optimist -- and he doesn't believe in open government. "Trust me, it'll work out fine." He means it. He shares both these traits with his predecessor, in different ways: I suppose that Martin Seligman is right about optimism, and I'm worried that maybe nobody who doesn't share both those traits can be elected.

This thought does not add to my optimism.

UPDATE: The Iraq Body Count people beg to differ: they offer a list of implications which they believe follow from the Lancet results, and argue that

In the light of such extreme and improbable implications, a rational alternative conclusion to be considered is that the authors have drawn conclusions from unrepresentative data.
Well, I believe they're effectively accusing the data-collectors or data-providers of lying, and I hope they're right. But they share final conclusions with the Lancet people, and of course I don't.

UPDATE: A trusted-by-me Statistical Modeling, Causal Inference, and Social Science blog says basically supportive things of "Estimating Iraq deaths using survey sampling"; I will be astonished if there's anything really wrong with the statistical reasoning as reported in the journal. Even after reading the IBC rebuttal, I'll be surprised if the death-rate hasn't risen substantially as compared with pre-invasion levels. (Pleased, of course, but surprised.)


Post a Comment

<< Home