May 20, 2013

Not an optimist. Not a pessimist. A hopeful scientist.

In a world that's filled with bad news, a constant barrage of current outrages, and coming disasters, I am generally unworried.

People who listen to these reports and believe them are pessimistic, and it's easy to see why. When I tell them that I am not pessimistic also, they accuse me of optimism, as though this was some sort of character defect. But, I explain, the fact that I'm not pessimistic does not mean I'm optimistic.  
Is the glass half empty or half full? The pess...
Is the glass half empty or half full? A pessimist would pick half empty, while an optimist would choose half full. An engineer, the story goes, would say that the glass was twice as large as it needed to be. A scientist would report only what was testable: that the 300 ml glass contained 150+/-3 ml of water with 99% confidence.(Photo credit: Wikipedia)
Unless I see the predicted doom as inevitable--and there is very little that is inevitable--then I am hopeful, which is often taken for optimism. But it's not. I can be pessimistic and hopeful or  optimistic and hopeful.

My degree of optimism or pessimism is based on my assessment of the probability of either outcome, my degree of certainty about the probability statement, and whether I believe the likely outcome will ultimately be bad or good. In most cases the probabilities are uncertain. In the minority of cases where the probabilities run strongly for a bad outcome, then I will be pessimistic (though hopeful). If they run strongly for a good one, then I will be optimistic (though guardedly so; things can always go wrong.)

But in most cases I can't make a prediction primarily because there's no good basis for making a prediction, and secondarily because while first-order effects can sometimes be predicted, there are often second-order effects that may offset them. For example, one might consider a reliable prediction in the price of oil prices to be a bad thing, but one consequence of such an increase might be more investment in sustainable energy, a good thing.

For me to take any prediction seriously, it has to be scientifically credible. That means the there must be a well defined method for making such a prediction, and the method must have been tested many times, very thoroughly and have been found accurate. Note that what must be predicted is the future, and not the past: there is a vast difference between successfully predicting a future outcome, and successfully demonstrating that had the method in question could have successfully the predicted the past, from a still earlier past time that viewed that past as its future.

The "ask the expert" prediction technique is testable: see what that has expert predicted in the past, and measure how accurate have those predictions have been. Sadly, the answer, found by Philip Tetlock, is that most experts are not very good.

His Expert Political Judgment: How Good Is It? How Can We Know? (2005) describes a twenty-year study in which 284 experts in many fields, including government officials, professors, journalists, and other, and with many opinions, from Marxists to free-marketeers, were asked to make 28,000 predictions[1][2] about the future, finding that they were only slightly more accurate than chance, and worse than basic computer algorithms.
The world is uncertain, and there's not a lot in the future that I am certain about. But as long as it's possible, even if improble, that things will work out I will remain hopeful.
Enhanced by Zemanta

No comments:

Post a Comment

Pages