One final lesson from “Freakonomics”, and then we’re done with Sumo wrestling. In the last post, we chatted about Sumo wrestlers who lost matches to help fellow wrestlers. The Japanese didn’t believe that Sumo wrestlers could cheat. Betraying their profession, their religion and their nation just didn’t seem possible. Yet, the Freakonomics staff provided a compelling analysis showing that once wrestlers won enough matches to secure their own ranking, in the late season they could strategically “give” matches to an underperforming wrestler, allowing them to maintain their rank and status. But data analysis, insider leaks, whistle-blowers, even murders to prevent further investigations all failed to change public opinion..
Wait a minute! Isn’t crime in Japan almost non-existent and the police force nearly 100% effective in arresting criminals? Shouldn’t the police be interested in what whistle blowers had to say? In fact, we learn that a documentary was produced, with testimonies from Sumo managers that cheating was widespread; this led to a public outcry… that the whistle blowers should be jailed for slandering Sumo. Hmmm… tough crowd. Then, two of the most prominent whistle blowers scheduled a press conference, but were both murdered the day before. Any interest by the police now? Nope, none at all. Now, in the past I’ve had to work with groups that are just a wee bit resistant to self-examination, but this is ridiculous! How could the most effective police force on the planet not see that something was wrong?
Of course they could. It all goes back to incentives. One former police officer, summed it up for the Freakonomics staff. Japanese police only open a case if they know who committed the crime; this is partially because of the demand for success, but the more important police need to maintain law AND order (the same as in the US). However, the Japanese uniquely interpret this to mean that taking on problematic cases will leave criminals at large, undermining the public’s confidence in the law, eventually undermining the public order. By not prosecuting all cases the arrest rate is nearly 100%, assuring the public that no one can escape the law, making crime an undesirable proposition, preventing crime from being committed in the first place. Crime prevention trumps crime resolution. Strange thinking, but I’ve heard an echo of the same logic from QC and reporting groups that want to protect the firm from bad statistics, don’t want to find problems that they won’t be able to fix, etc. Almost any group that collects and reports metrics has more than one directive. In fact we’re all eternally trying to solve the old Good /Fast / Cheap puzzle, which means focusing on different elements. You may have even had some frustrating meetings where you’re staff eventually says, “Can’t you just choose one?”
These are all foundational issues. We all find (or should find) strange anomalies in our numbers, different interpretations of our goals, and moments when it looks like your managers were asleep at the wheel. But more often than not, the correlations that you see are results of incentives that were accidentally put in place. Causality is usually more complicated than what you see on the surface. And that is definitely, my Niccolls worth for today!