Tuesday, June 27, 2006

Prediction will eat itself

Since early May, I have been working with the Long Now Foundation, a nonprofit organisation dedicated to encouraging long-term thinking and responsibility, as research fellow on the Long Bets project. Long Bets is based on an ingenious, simple idea. If people want to make predictions about matters of long-term social or scientific importance, let them put their money where their mouth is. Stakes placed on bets, involving two contrary views of the future running head to head, are held in a long term investment fund, and when the matter is settled, perhaps years, decades or more down the track, the money goes to the winner's chosen charity. Money is also put up in support of predictions (one person's statement about the future, no opposing view necessary), which goes towards maintaining the service. The point is that all this behooves participants to think carefully about how posterity will judge their prognostications, and their audience benefits from the availability of relatively well thought-out, succinct supporting arguments.

Four years after the project was launched, the key to what I've been trying to do at Long Now recently is to take what is clearly a long-standing human interest in prophecy -- call it a predilection for prediction -- and leverage it into a more complex and nuanced appreciation of multiple possibilities. (Hence my preferred title for the field I'm in: futures. The "s" is for plurality, multiplicity, unpredictability. This condition is what makes choice possible.) Why would I want to do a thing like that? Because I believe prediction, without more, to be an inherently limited, and often erroneous, if not dangerous, stance towards the future.

A London Financial Times article (subscription required) was sent around at Long Now by Board of Directors member Peter Schwartz last week, on a recent book called Expert Political Judgment. Author Philip Tetlock is a psychologist "who has spent 20 years asking pundits to predict who will win elections, what countries will acquire nuclear weapons or enter the European Union and how the first Gulf war would end. He has tested 30,000 predictions from 300 experts against outcomes. Mr Tetlock finds that his respondents are not very good. They do better than a chimp who answers at random, but not much, and worse than simple forecasting rules based on extrapolation" (John Kay, "The world needs more foxes and fewer hedgehogs", Financial Times, 20 June 02006; see also this excellent review of Tetlock's book by Louis Menand, "Everybody's an Expert", The New Yorker, 5 December 02005).

For those who expect the future to be predictable, assuming that in principle it can be known but that they simply don't quite have the expertise, tools or funding to get at it, this conclusion may come as a surprise. GBN's Jay Ogilvy has made the point well: "modern planners have often assumed that, with enough Enlightenment science in the budget, the future can be predicted and/or controlled. Scenario planning is based on a contrary assumption, that the future cannot be predicted..." (Creating Better Futures, 02002, p. 12). And it's not just "modern planners" who hold this mistaken view; in my experience it's far more widespread than that. Yet, for many of the things that are most important to us -- social, human-level things like our families, communities, workplaces, and countries -- the future is in principle and by definition unpredictable; partly because the predictions themselves feed into the change process.

Tetlock's findings, which indicate that even experts aren't much good at prediction -- that those who ostensibly know the future best, in fact know no more than the rest of us -- are, however, very much in accord with one of the main lessons taught by a long-time academic futurist, Professor Jim Dator at the Hawaii futures program, where I study. Dator's First Law of the Future is, "The future" cannot be "predicted" because "the future" does not exist.

Now, if you're convinced that the future is predictable, you're not going to change your mind here (...I, er, expect). Fine. But stay with me for a moment. Assuming for the sake of argument the basic contention here is sound, what would this mean for Long Bets, and predictive discourse generally? Based on many, many conversations I've had on this point, at this stage quite a few people start to wonder what the hell is going on with a self-proclaimed futurist who believes there's no future to study. As well they may. What we look at, since the future itself is perpetually unavailable for comment, is images of the future. These are the beliefs, hopes, fears, desires, intentions, fantasies, expectations and, yes, actual images of one sort or another, that people carry around in their heads, and express in their art, stories, and above all their decisions; thereby bringing "the future" into being from among the countless possible futures that might have been. Returning to the question above, what this means for predictive discourse is that predictions can make a useful, interesting point of access into futures-oriented discussion (constituting, as they do, a major example of images of the future, albeit a conspicuously overconfident variety). It also means that predictions should not be mistaken for statements of fact about things that have not happened yet. Simply put, what people predict is a useful guide to what they believe and how they plan to act, but one of the dangers of prediction is that, while you're arguing over what the future will be, you're missing the more important discussion about what it could be, and other people may be making the crucial decisions without you.

What this means for Long Bets, as I see it, is that there's an important challenge in diverting some proportion of the prodigious amount of energy that people seem to devote to figuring out what the future "will be", and redirecting it into the exploration, invention and pursuit of what they would like it to be. Dator again: "The point is not to try to predict a better future, but to strive to create one." Similarly, sci-fi writer Bruce Sterling once said in an interview, "Future is not a noun, it's a verb." Long Bets version 2.0 should, then, be a useful broad-based resource featuring all sorts of different images (predictive expectations) of the future, which at first exploits the apparently timeless allure of "the future will be thus and so", but which deepens the discussion unexpectedly and problematises the predictive stance. People may come expecting a noun, but will end up getting verbed.

No comments: