Friday, December 6, 2013

Another Cycle of Faulty Predictions and Forecasts is Upon Us

This is the time of year where we start seeing lots of stories about “the year ahead.” At least some of the “what will happen in 2014” predictions will prove reliable, especially those which simply extrapolate from trends already in place.

Far harder are predictions about what might happen in five, ten or 20 years. And yet we are compelled to keep making predictions, sometimes for unworthy reasons, such as when trying to prove to others “how smart we are.”

But lots of firms make a business about predictions, because there is a market for such things, even when the forecasts are unreliable. Just how unreliable is the issue.

Philip Tetlock's Expert Political Judgment: How Good Is It? How Can We Know? found that “specialists are not significantly more reliable than non-specialists in guessing what is going to happen in the region they study.”

Sam L. Savage’s The Flaw of Averages points out that plans based on average assumptions are wrong on average, because uncertainty in life is much more pronounced than people generally assume to be the case.

Nassim Talib’s The Black Swan likewise deals with the powerful impact of unpredictable and unexpected developments.

In fact, some would go so far as to say that forecasts always are wrong. That isn’t necessarily a bad thing, as minor fluctuations along a predicted trend line nearly always happen. That is true of most economic forecasting, some argue.

What is important is the trend line, not the actual results that fluctuate around a trend line.

The bigger problem is when a forecast trend fails to materialize at all, has far-lower magnitude than expected or fails to conform to a forecast timeline.

In other words, there are bigger problems when a forecasted series of events fail to happen at all, represents a much-smaller change than expected, or takes significantly longer to occur than expected.

In such cases, whole companies or industries fail to develop, fail to have big consequences and impact or happen too far in the future to have immediate practical consequences.

Forecasts nearly always become less accurate the farther into the future one predicts. One might suggest a logical reason for that failure: we tend to extrapolate present trends into the future, when history suggests different assumptions of unclear dimensions will operate in the future.

Granted, there arguably is a difference between “forecasting,” the attempt to identify ranges of possibilities and “prediction,” the attempt to describe with certitude a future event or set of events.

In practice, the dividing line is likely arbitrary and porous. The future is nearly impossible to predict, and at risk of great, sometimes fundamental error, even when forecasting.

All that said, get ready for the “2014 predictions.” And that isn't a prediction! It's an observation!

No comments:

FCC Says 5G Just as Safe as Other Gs

Are 5G or other mobile phones safe to use? The scientific evidence so far suggests the clear answer is “yes.” And after a new review in li...