Monday 23 February, 2009

Dangerous conclusions

I am reading The Bottom Billion by Paul Collier. It, according to The Economist, will be hailed a classic. Maybe it is one already.Blogger is a free blog publishing tool from Google for easily sharing your thoughts with the world. Blogger makes it simple to post text, photos and video onto your personal or team blog.

But I find it illogical and somewhat dangerous.

First, the book is based on the premise that nations can be reduced to numbers. This is an oversimplification so gross that some may find it completely unjustifiable. 

Having taken this premise, the author generalises. Let me take a typical example. While discussing the factors that may affect the 'turnaround of a failed state', he concludes, among other things, that democracy is not one of the factors.

There are two immediately perceptible problems here. First, democracy is an amorphous concept. North Korea calls itself the Democratic Republic of Korea; India is a democracy; and so is the United Kingdom.  

Second, this overall statistic may mean nothing in a particular case. 

Let's go off elsewhere to understand this. 

Let's say we want to know, from data on donors to a charity that funds children's programmes, if a person's place of residence influences his propensity to give. And we come to the conclusion that it doesn't. A citizen of Madras, picked at random, is just as likely to donate as a citizen of Bombay (picked at random too). 

This tells the charity that when, say, it's mailing out appeals, there is no reason to prioritise mailing to any city (or neglect any one). 

Yet the conclusion may be completely wrong on in the particular, where place of residence may be decisive. Let's say you live next to a well-run school for poor children; while I live next to a remand home that is hell on earth. You believe giving money will help run similar schools; I believe anything I give will go into more such hell-holes. You are very likely to give; I am very likely not to. 

Both of us are probably wrong, but that's not the point. The point is that in our cases, residence mattered.

In an ideal world, this detail, and every other, would be known to the fund-raiser, and he'd be able to act with minimal risk. Of course, he has nothing remotely comparable to omniscience, and does the best he can. 

That's probably OK too, because data comes at a price. One has to do a cost-benefit analysis, even if that's intuitive. The 'general' will, at some stage, triumph over the 'particular'.

Now, Collier is not talking about thousands of donors on a mailing list; he is talking about millions of people mired in seemingly hopeless poverty, whose lives can well depend on interventions by Western powers. Is statics of his sort a good tool to decide matters of that magnitude?

"But wait a moment," you'd say, "nowhere does Collier claim that his conclusions will hold in every case. He is simply giving an overall picture. So what's the problem?" 

The problem is that his picture is accessible. It will be used by many, directly or indirectly (by those who know Collier from, say, columnists who quote him), to judge their governments, without going into the details and caveats which, surely, exist in Collier's academic papers.  

So, let's say you are a policy-maker who must decide if a particular third-world country should get badly needed aid. It is at present under an old-fashioned deranged dictator. Further, your government is not particularly bothered about democracy, choosing instead to worry about economic progress. Nonetheless, in this case, you know that nothing will improve unless the poor country gets something that can be logically called a democracy and free press. 

Logically, you should make aid conditional to the dictator's departure. But... you can't. Because, thanks to Collier, everyone knows that democracy doesn't matter. 

Obviously, Collier never drew that conclusion. Inevitably, that conclusion, and similar ones, will be drawn.  

No comments: