Monday, September 17, 2012

recency bias

Recency bias is the tendency to think that trends and patterns we observe in the recent past will continue in the future. Predicting the future in the short term, even for highly changeable events like the weather or the stock market, according to events in the recent past, works fine much of the time. Predicting the future in the long term according to what has recently occurred has been shown to be no more accurate than flipping a coin in many fields, including meteorology, economics, investments, technology assessment, demography, futurology, and organizational planning (Sherden, The Future Sellers).

Doesn't it strike you as odd that with all the intelligence supposedly going on that such things as the breakup of the Soviet Union, the crumbling of the Berlin wall, the former head of Sinn Fein meeting with the Queen of England, the worldwide economic collapse of recent years, the so-called Arab Spring, the recent attacks on U.S. embassies in several Muslim countries, and a host of other significant historical events were not predicted by the experts? Wait, you say. So-and-so predicted this or that. Was it a lucky guess or was the prediction based on knowledge and skill? If the latter, we'd expect not just one correct prediction out of thousands, but a better track record than, say, flipping a coin. Find one expert who's consistently right about anything and we still have a problem. How can we be sure that this sharpshooter isn't just lucky? If thousands of people are making predictions, chance alone tells us that a few will make a right call now and then. The odds in favor of prediction success diminish the more events we bring in, but even someone who seems to defy the odds might be the one a million who gets lucky with a string of guesses. You flip the coin enough times and once in a while you will get seven heads in a row. It's not expected, but it is predicted by the laws of chance. Likewise with predicting how many hurricanes we'll have next year or what stocks to buy or sell  this year.


Recent events and trends are easier to remember and discern than either events in the distant past or unknown events that will occur in the future. Rather than do the hard work of studying the past or accepting the fact that despite our best efforts at predicting events far into the future for such things as the weather, technological advances, and population trends, these and many other areas of human interest are beyond our ability to predict at much better than chance levels or naive estimates. (A naive estimate of weather prediction uses either today's weather to predict tomorrow's weather or it uses a seasonal average to predict this season's weather.)

Some of the most amazing technological advances that have occurred during my lifetime weren't predicted by any of the experts: the Internet, the personal computer, the smart phone, digital music, to name just a few. And I'm still waiting, along with millions of others, for my jet pack.

1 comment:

  1. Theoretically speaking, if someone knows all the factors behind a phenomena, he will accurately predict the phenomena itself. If someone knows the pressure, the force and all physics factors behind the coin that has been flipped his prediction will be accurate. The science works with probability because there is small chance for getting all the causes that create the phenomena that has been observed. Predicting the act of every single human is almost impossible so predicting the future politics, wars etc... is impossible for now. But I personally doubt that some of the things that we enjoy in our modern time weren't guessed before they happened. Starting with Jule Verne and his submarine stories and ending with Leonardo da Vinci and his ideas of flying machines... Whether this is a coincidence or not I cannot tell for sure.

    Hari Kacanovski

    ReplyDelete