My Account

Warning: You are using a browser that does not support angularJS. Some site functionality will not be available to you. Please consider updating to a newer version.

Algorithms Can't Pick College Football Champions or Predict Economies

William L. Anderson

At the end of this week we will know who will play for the national championship of major college football. Will it be Oregon? Auburn? Boise State? Texas Christian University, or another team? We won’t know until a mathematical equation tells us.

That’s correct. A mathematical algorithm determines who plays for the Bowl Championship Series trophy. True, the teams that play will have had outstanding seasons, and maybe had there been a playoff the same two teams would have reached that final game. It’s something we never will know, but there are some things we can know about the use of mathematics.

A mathematical formula never would have placed Butler University within one rimmed-out shot of beating Duke University in the NCAA basketball finals last April. A mathematical formula never would have given us Boise State’s exciting overtime win over Oklahoma in the Fiesta Bowl a few years ago, complete with a hook-and-ladder play for a touchdown and a Statue of Liberty play to score the winning points.

Predicting the Economy’s Future

Likewise, a mathematical formula cannot tell us what the U.S. economy (or any other economy) is going to do next year; nor can an algorithm tell us exactly how much revenue a new tax will collect, no matter what the Congressional Budget Office and Paul Krugman tell us. (A computer spits out the results, but it only operates according to the formula someone programmed into it.)

I am not denigrating mathematics per se; there is a place for math. However, there are legitimate reasons that using math in the way economists currently use it will result in failure. The first is that our economic future is not based on risk for which there can be understood probabilities. No, what we face is something entirely different: We face uncertainty, which economists like Frank Knight and Ludwig von Mises understood as a range of outcomes that are unknown until they happen.

The second reason is that economists cannot place entrepreneurial insight into an equation. Entrepreneurship is not quantifiable; one cannot subject it to probabilities or mathematical rigidity.

For example, laws of probability could not tell us that two college dropouts would invent a personal computer, build it in a garage, and then have the entrepreneurial vision to turn that invention into a line of products under the name Apple. Furthermore, one cannot put the freedom of enterprise (that quickly is disappearing in this country) into an equation and then predict an iPod or iPhone with it.

Economists once understood this point, but the lure of mathematics was too great. Paul Samuelson at mid-century wrote that unless economics adopted the mathematical analysis of the physical sciences, economists would not be true scientists, which would place their work into the undesirable category Mises called “metaphysics.” While Mises, F.A. Hayek, and Murray Rothbard pointed out the folly of trying to fit the square peg of economic analysis into the round hole of math, other economists derided them and declared that mathematics should dominate economics because it “fits the market test. (I deal with the “market test” issue in the Quarterly Journal of Austrian Economics [pdf].)

There are legitimate reasons why a math formula could not have seen Butler almost win the NCAA championship and why no mathematical economist could have foreseen Apple. An algorithm can be based only on what we already know and by definition cannot deal with uncertainty. It cannot predict an injury to a key player, a critical missed foul shot, a slip by a defender, or a fingertip catch in the end zone. In fact, none of us can know these things – until they happen.

The popular 1960s cartoon The Jetsons featured futuristic home computers that looked like, well, 1960s mainframes. The creators could not have envisioned what actually would exist just a few decades later, not hundreds of years in the future. Likewise, the algorithm, although useful in building rockets and bridges, cannot tell us what we need to know in economic analysis – or give us a national collegiate champion in football.

Related Articles


{{}} - {{relArticle.pub_date | date : 'MMMM dd, yyyy'}}