Explanations vs. predictions

If curiosity did, in fact, kill the cat, then felines are likely the only creatures for which that particular psychological trait is a bad thing.

If curiosity did, in fact, kill the cat, then felines are likely the only creatures for which that particular psychological trait is a bad thing. Certainly, being unable to resist the urge to find out what’s over the next hill, down the next river, or in the next galaxy is a big part of why—for better or for worse—we are the planet’s dominant species.

Our curiosity can usefully be seen as inquiries into two sorts of questions: Why/how did that happen? And, what will happen next? The former captures our desire to explain events; the latter addresses our need to predict them.

Different scientific disciplines address these two questions to varying degrees. As ever, physics is the gold-standard science that scratches both itches especially well. The Large Hadron Collider in Europe seeks to provide an explanation for pesky phenomena such as mass by nailing down the details of the Higgs boson. That will be tremendously satisfying and essentially complete the standard quantum model. We will know—really know—what matter is. And the predictive power of the explanations that physics provides manifests itself through practical application in fields such as engineering.

Medicine probably falls on the “prediction” end of the spectrum. Although certainly built upon an attempt to explain why disease strikes and how it behaves in order to better combat it, the need for effective countermeasures is so great that we settle for effective interventions even if we don’t fully understand them. In other words, we accept useful predictions despite the lack of correct explanations. If a drug works in 40 percent of cases like yours but no one can explain why or how or if it will work for you, if that’s the best we have, you’ll probably take the drug.

So where does management science fall in the explanation/prediction space?

Here’s how I’ve seen it play out with the Disruption theory of innovation. Disruption posits that new innovations have the best chance of success when they have materially different performance profiles, appeal to customers of relatively little interest to dominant incumbents, and, when launched by an established firm, enjoy substantial strategic and operational autonomy from the mainstream business. In contrast, attempts to enter markets dominated by successful incumbents—no matter how well-resourced—will typically fail.

This is precisely the kind of statement that managers need, because it prescribes what they should do. And it turns out to have at least some predictive power. In a series of experiments at Harvard Business School, MIT Sloan, the Ivey Business School in London, Canada, MBA students using Disruption theory were nearly 50 percent better at picking winners than were seasoned  venture capitalists using their own idiosyncratic approaches. Specifically, the VC folks had a success rate of about 10 percent, while the disruption- savvy MBA candidates enjoyed a success rate of about 15 percent.

Whether or not this is enough of an improvement in predictive power to make a difference is a topic for another day—I happen to think it’s pretty significant. Still, it’s a long way from perfect. Take the iPhone, for example. This was Apple’s attempt to succeed in the phone business by offering a better product than that of successful incumbents (Nokia and BlackBerry, to name only two). Disruption theory would have predicted failure. Yet the iPhone has been a huge hit.  This leads me to think that Disruption theory is more like medicine than physics: material but imperfect predictive power based on incomplete explanations of causal mechanisms still poorly understood.

Unfortunately, creators of management theory typically respond to the inevitable imperfections of other theories by addressing the wrong problem: seeking to improve explanatory power, adding additional variables, new frameworks, and often impenetrable jargon.  Improving the explanatory power of theories solves the wrong problem, because what we really need is better predictive power. This often means settling for relatively limited advances, but so long as we remember that “better” can often be enough, we exhibit the lesser-known second half of the inquisitive cat aphorism, the “satisfaction that brought him back.”