If social scientists are able to make accurate predictions by crunching big data numbers, but there's no theory or understanding behind those predictions - are they doing social science?
Science, knowledge and understanding are fundamentally useful to the degree that they make predictions. Some say human intelligence, at its core, is nothing more than the ability to have a handle on the future (what valley you're most likely to find ripe fruit, what gal is most likely to make a good wife - and agree to ever marry you, what stocks to invest in, etc.).
In the past, in order to be a good predictor you had to have good theories (or at least a few rules of thumb), because there was no way a human brain could calculate all the factors affecting future events. A good theory therefore, was touted as one that could explain a lot with as little input as possible.
But today, super computers are getting better and better at making predictions with no "understanding" whatsoever. Merely by scanning for past patterns and relations, Watsons can churn out questions to Jeopardy answers and Nate Silvers can predict presidential election results. (Nate's not actually a super computer, but his algorithms are pattern hunters that are more interested in predicting how people will vote rather than why they vote the way they do.)
Here's another prediction: as computers churn ever larger seas of data, "theory" will one day become as quaint a notion as "philosophy" or "god."
No comments:
Post a Comment