Friday, July 9, 2010

Of Black Swans and Octopuses: Brazil, Soccer and Why We Just Can’t Predict

By Dr. Suhaib Riaz.

I was in Rio de Janeiro for the Academy of International Business conference when Brazil played Chile in the soccer world cup and won 3-0. Everyone at Copacabana beach, where the FIFA Fan Fest Screen showed the game to over 10,000 people, had such confidence in the team that “goal” shouts erupted on each occasion before the actual goals happened. And yet Brazil is out. And so are other favorites, such as Argentina and Germany. Two teams that have never won the soccer world-cup before will battle in the final. Why can’t we predict outcomes from such a controlled environment, where the regulatory framework (rules) are well known, the resources and capabilities of teams are well known, and the only strategy dynamics are how to move a handful of players to push a ball into a goal post? And if we can’t predict even these simple outcomes, how about more complex real world business environments?

By sheer coincidence I’ve been reading and ruminating on Nassim Nicholas Taleb’s “The Black Swan” (the second edition is out with a new section on robustness and fragility). Taleb doesn’t mince words, calling it a ‘scandal of prediction’ in most disciplines, particularly in the social sciences: “We are suckers for those who help us navigate uncertainty, whether the fortune-teller or the “well-published” (dull) academics or civil servants using phony mathematics.”

Such suckers in fact, that our latest prediction-obsession (fueled by the media) is with an Octopus named Paul in Germany that seems to predict Germany’s soccer outcomes better than any human judgment. Octopus Paul apparently has picked 6 out of 6 correct outcomes for Germany in the world cup. Any explanations?

Could it be that it’s simply drawn to the red on the flags, which would easily account for 4 picks of German wins against non-red flags, and two picks of losses against flags with more red than the German flag? I have no idea at all if Octopuses can even see that much colour, but here’s what Taleb might say: the real comparison would be if we consider a large number of random octopuses and see how many of them get it right: perhaps one or two will, just due to randomness, and Paul happens to be that one: “The reference point argument is as follows: do not compute odds from the vantage point of the winning gambler, but from all those who started in the cohort.” He calls this “the problem of silent evidence”, and it is widespread in our disciplines as various forms of survival bias – we only look at the ones that got it right, or survived, not at the failures.

Taleb’s larger point is of course that “we just can’t predict”, particularly given the presence of outliers or Black Swans in all social science phenomena. Particularly vexing is his refusal to accept a clear demarcation between Risk and Uncertainty – to him, the idea that risk is the situation where we can calculate probabilities means nothing. He seems to suggest that in social sciences there are no real world situations where you can calculate probabilities – and thereby predict risks; there is just uncertainty.

Which brings me back to uncertainty in the business world. Can we predict how the Brazil economy will look in 10 years? Can we predict what the leading companies will look like? I doubt it. Even a short visit there reveals the complexity of predicting such phenomena. Brazil’s institutional framework is under transformation. There are innumerable socio-economic variables that could change (say, for example a different type of political leader; or a new generation from the Favelas that is now apparently sent to school 7am to 7pm to keep them away from the drug lords); firms and their capabilities that are typically embedded in such complex environments could see sudden changes; Brazilian firms’ interactions with firms from the rest of the world could thus change; in all, the role that Brazil and its companies would have in the global economy is largely unpredictable. Only with a lot of “ifs” and “buts” can we say something, and most likely we would have to later blame our wayward predictions on factors “exogenous” to our analysis that we obviously couldn’t have seen coming.

Taleb, or NNT as he refers to himself in the book, has many more gems, but more on that later – here’s just one on a favourite theme:
“Don’t ask the barber if you need a haircut- and don’t ask an academic if what he does is relevant.”

Incidentally, I was presenting a narrative on the debt crisis, mostly after the events of the financial crisis, as is the bane of social science academics. The paper drew a surprisingly positive response for what I thought was really a work in progress. Perhaps because even at this point in the 'post-crisis crisis', there were few works on even trying to explain the largely unpredictable events that happened.

The question for us is: What kinds of organizations and strategies can we conceive of, for a world where we just can’t predict? Any takers? Or should we get an Octopus as CEO?  

No comments:

Post a Comment