Tuesday, February 24, 2009

On information and judgements

I didn't think I'd ever find myself arguing in support of Tony Blair, but Martin Weller, via Phil Greaney, has made me.

Martin's argument, based on ideas about the wisdom of crowds, is that in the run up to the Iraq war, Tony Blair had so much information that it prevented him from making a proper choice about the war. The ordinary people, though, with much less information to go on, were able to "see the salient features of the war, and ... instinctively judged it to be 'wrong'".

Alan Cann disagrees, and I am more on his side than theirs. The point Alan makes is that there is no such thing as too much information; there is just inadequate filtering. I think that he is right in that, but also that he is wrong, as the others are, to treat this as purely an issue of information.

When decisions have to made, information is only sometimes your friend. Take the issue of going to war in Iraq as an example, and let's say there are two clear choices, either to go to war, or not to. If we are lucky, the information we have makes the decision for us. Either it is clear that Saddam Hussein has been neutralised, and we do not need to go to war. Or it is clear that he is still as madly aggressive as always and needs to be stopped by military means. In either of those two cases, the information we have makes the decision easy. Tony Blair was not in that position. He was in that awful in between state where the information doesn't tip you one way or the other. People put in this situation often in fact go seeking more information, in the hope that new information will make the decision clearer. Alas, in politics that is rarely the case. I don't believe, in fact, that Tony Blair had too much information. What he had was equivocal information, and he did what he was being paid to do in such circumstances - he made a decision. To be frank, I respect him for that. He did not beat around the bush, pardon the pun, and make half a decision, or some lame compromise, he followed it through properly.

The other side of the coin is about what crowds do. I like the thesis of the wisdom of crowds, but again I think you have to put it into the context of decision making. Nobody who was protesting against the decision to join the Americans in the invasion of Iraq actually had to make that decision. Martin makes the point that Blair might say what if they knew what I know, and counters it by suggesting that Blair should have asked himself "What do they know that I don't?" Again I think that misses the point. It's often easier to be on one side or the other when you don't have to make the decision. But suppose someone else were put in Blair's shoes - in the sense of having to be responsible. It suddenly becomes your job to decide who will live and who will die. All of a sudden, that information that Blair has will become very precious to you, as you search for anything, any tiny clue that will tell you which direction you should be pointing in. Somebody's blood will be on your hands whichever decision you make. You might well end up making the opposite decision to that which Blair made. But you will realise that it was not nearly as clear cut as you thought it was.

2 comments:

AJ Cann said...

Nice post Rob, but I did say in my post "Leaving aside political considerations, for example..." as scienceoftheinvisible.blogspot.com is about education and technology and I didn't want to get sucked into a political argument about "the legacy of Blair", so I restricted my remarks to information processing.

Unknown said...

Hi Rob,
like AJ I wasn't really interested in the political angle, it was a good example of where the crowd had one view, the holder of the info another, and in restrospect I think most people would say the crowd were right. It's interesting just to think why that was so. Could just be luck.
I think something you say is key: "What he had was equivocal information" - yes exactly, and adding more equivocal information doesn't help, it only confuses the issue. The crowd had less equivocal info, and I think, could thus see what the important features were. This is the same as AJ's filtering argument - he's right in that what you need are better filters, but in uncertain environments those filters don't exist.
But maybe this example is too tied up with political issues to be used as a crowds one.
Martin