Big data that’s wrong is still wrong.
In the opinion pages of the Sunday New York Times—Sunday Review in the dead-tree version—we have one article that mentions on-line dating sites as effective users of Big Data techniques, and another suggesting that the matching algorithms used by these sites don’t work.
The first article, “The Age of Big Data” includes this:
Online dating services, like Match.com, constantly sift through their Web listings of personal characteristics, reactions and communications to improve the algorithms for matching men and women on dates.
The second, “The Dubious Science of Online Dating” has this:
One major problem is that these sites fail to collect a lot of crucial information. Because they gather data from singles who have never met, the sites have no way of knowing how two people will interact once they have been matched. Yet our review of the literature reveals that aspects of relationships that emerge only after two people meet and get to know each other — things like communication patterns, problem-solving tendencies and sexual compatibility — are crucial for predicting the success or failure of relationships. For example, study after study has shown that the way that couples discuss and attempt to resolve disagreements predicts their future satisfaction and whether or not the relationship is likely to dissolve.
When it comes to data, size isn’t everything.