In my last column I noted how Political Science, along with most social sciences, has a bigger problem with prediction than seems to be generally acknowledged. This is of course hardly unique to members of this particular tribe; the media are even worse. Take for example the US media’s treatment of Bode Miller in the Winter Olympics. For those of you who have been living in a cave for the past month, Miller was the ‘sure thing’ for the US ski team. After all, he had his own set of sponsored ads, videos, and an interactive website from Nike. Miller was competing in five events and was, according to the US media, the front runner to lift possibly all five gold medals. Quite why this was the case was a mystery to me. Sure, he’s a damn good skier, but if you looked at his world cup results you would see that he was hardly head-and-shoulders above the competition, and in particular events he was well below the top rank.
Now, consider that each event Miller participated in at the Olympics was hardly an independent event due to the psychological impact of each result on the next, and that he is fully entitled (like the rest of us) to have a bad day at the office. Well, he did. He missed in all five events. Needless to say the media are now picking his corpse clean for defying their predictions. As the New York Times put it after his first ‘failure’ – “He is paying the price for misplacing career priorities.” Quite how the writer of this piece knows exactly where Miller left his priorities is unclear. The fact that he did not win is insufficient evidence, and you can bet your last dollar that had he won his next event such concerns would have been completely erased. Moreover, the last time anyone won five gold medals at a Winter Olympics was Eric Heiden in 1980 for speed skating. As far as I am aware, no one has ever won five medals in an Alpine event. Why Miller didn’t win could be a surprise only if one was deliberately ignoring much relevant information. Indeed, taking a select few data points and projecting them forward as an inevitability almost always produces disappointing results.
Two things stand out for me from this nonsense. First, why is anyone surprised that Miller did not win any gold medals, let alone five, when no one has ever done so? Second, and more interestingly for the non-skiers out there, why do people have a tendency to take two or three data points and project them into the future as an inevitable trend? Beyond the hype associated with US contenders, and the sheer myopia of the US media to the possibility of ‘foreigners’ actually beating the home-grown talent, such a tendency has consequences far beyond the Winter Olympics.
Consider Condoleezza Rice and Hamas’ electoral victory. The Secretary of State noted after Hamas’ victory at the polls that “I don’t know anyone who wasn’t caught off guard by Hamas’s strong showing.” In fact, me, my wife, my cat, and The Economist Newspaper all knew this was coming down the track. Why then didn’t the Secretary of State, with all the resources at her disposal, not have an inkling that such a thing was going on? Perhaps what might be called ‘the Bode Miller problem’ was at work here too?
Consider that on issues as disparate as the invasion of Iraq, Social Security privatization, and energy policy, the Bush administration has never been one to let mere facts get in the way of a good ideology. Disconfirming evidence is screened out and only confirming evidence is admitted. A few supporting data points are projected as a trend while everything else is ignored.
In the case of the election of Hamas, while Fatah had recently done what the US has wanted in terms of halting suicide attacks, holding elections, and playing nice with Israel (all of which was acknowledged (trended) by the US), what Israel had done to Fatah over the past few years, in particular, bombing the PLO’s governing infrastructure into the ground thus cutting off their all important patronage network, was, like the totality of Miller’s results, totally ignored. The trend-line predicting Fatah’s victory was projected forward since only confirmatory data were being examined, and everything that didn’t fit the trend was ignored. Consequently, when a Palestinian voter said at an exit poll “Fatah hasn’t done anything for us,” this seemed to come as surprise; despite it being manifestly obvious to anyone who wanted to look at the totality of the data. Simply ignoring data because it does not fit with a preconceived model can be justified if the data is randomly distributed and constitutes clear ‘outliers’ from the observed trend. But to ignore a clear trend in the data and simply focus on what you want to see is pretty much guaranteed to end up producing a nasty surprise, pace Hamas.
Now this tendency to see trends, ignore data, and pointlessly project into the future is not only sadly common among the media and the political classes, (remember the US government not so long ago predicting budget surpluses into infinity on the basis of three data points?) it has determinate effects on likely future outcomes. When Hamas won the election the reaction of the US, Israel, and even the normally placid Europeans, was swift and condemnatory, and who could be surprised by this? After all, the Hamas Charter of 1988 does call for the destruction of Israel and cites The Protocols of the Elders of Zion and the current “Nazi-Tartar” invasion by the West as reason enough. Indeed, there are undeniably a lot of data points out there pointing to actions by Hamas consistent with that interpretation and those ends. But even here there may be a ‘Bode Miller problem’ at work in that even here the past may prove no real guide to the future.
Consider that until into the 1990s the main body of the Irish Republican Army believed and proclaimed (quite seriously) that the UK government was holding the six counties of the North hostage as part of a colonial struggle, despite the exercise costing the rest of the UK millions of pounds each month with nothing in return except mainland bombings and death. Indeed, some breakaway Republican groups still adhere to the same beliefs. Yet, in order to believe such things one has to filter out massive amounts of data and project the few points that fit the preferred theory into the eternal and unchanging future. But when is the future ever eternal and unchanging? I am sure that much of Hamas is quite capable of continuing to believe in the forgery of the Protocols and act violently towards Israel, but let’s remember that one could have made the same projections about the IRA a decade ago, and yet they changed fundamentally, and quite unexpectedly.
Filtering the data to see only one trend negates potential futures. Seeing Hamas as a trend that cannot be stopped inevitably leads one to conclude that isolation and punishment is the only way forward. But Hamas has only ever known isolation and punishment. As such, proposals to cut-off aid in order to encourage capitulation is to fundamentally misread the data. True, there has been no IRA-like change yet, but to address the situation as an inevitable conflict preordained in the data will surely bring about such a conflict since we are blind to other possibilities.
So is expecting Bode Miller to win five gold medals the same as expecting Hamas to never change? Yes, but with one difference. Whereas Miller ‘failed’ on his own terms given the competition and the randomness of the day (after all, he might win six world cup races in a row in 2007), Hamas may only really ‘fail’ in the eyes of the Palestinians if the West and Israel are seen to make them fail. Key to the West and Israel doing this is to pick the data points they want to see (Hamas as unchanging and violent due to the trend line of the data) and project it forward.
Now, I freely admit that I know more about skiing than I know about the intricacies of Middle Eastern politics, but it does seem to me that, as the millions of people who read their astrology every day attest, humans like patterns and can see them in almost anything. Add to this ‘the Bode Miller problem’ that we can ignore much of importance in order to see much of irrelevance since it reinforces the patterns that we want to see, and perhaps it is better to let Hamas run the schools’ budget rather than deprive them of it. After all, something new in the data might be the start of a new trend, both for Bode Miller and Hamas.