The Infidelity of Models

The Infidelity of Models

Last week, my Foreign Policy colleague Marc Lynch made a strong case that the theoretical lesson to be learned from the Ryan Braun steroid saga is to trust the generalists: "Braun’s case was unique … until it wasn’t. Everybody thinks that their own case is unique, and of course it’s true that every case is unique in a million little different ways. But there are also broad patterns and mechanisms which seem to play out repeatedly across lots of those unique cases."

As a generalist myself, I am deeply sympathetic to Lynch’s argument — and indeed, the latest developments in the Biogenesis scandal do confirm certain truisms (like: A-Rod is a jerk). That said, I would argue that this summer’s crop of movie duds offers a cautionary tale to balance my colleague’s argument that foreign policy analysis could do with some more "Nate Silver-ing."

Why are so many summer blockbusters doing so badly? In a brilliant meta-essay at FP‘s sister publication Slate, Peter Suderman explained that one problem was that too many Hollywood films were adhering too closely to a single homogenous formula. Too many screenplays followed the beats that the author of Stop! Or My Mom Will Shoot spelled out in his guide to writing a successful Hollywood movie. The result is that, consciously or not, the movie-going audience can see every beat ("false victory," "all-is-lost moment," "dark night of the soul") before it happens. It’s comforting, but literally formulaic, and eventually the viewer gets bored.  

What does this have to do with foreign policy analysis? All too often, world affairs commentators try to fit a new situation into a prefabricated narrative. It’s how our monkey-brains work — we see patterns, we fit patterns, we predict patterns. Lynch argues that these patterns hold some predictive power. Yes, predictability is good — if the underlying situation is actually predictable. But any reader of The Signal and the Noise will tell you that there are hard limits to predictive models. Echoing Harvard political scientist Gary King, Silver observed that there remains an awful lot of phenomena where the amount of unexplainable variation remains pretty large. If analysts try to over-apply known narratives to unclear situations, they risk looking even more foolish than the prognosticators in Philip Tetlock books. Indeed, as first David Hume and then XKCD has pointed out, just because something hasn’t happened before does not mean it won’t happen in the future.

To see what I mean, consider the question of whether China is rising or falling. Up until, oh, about two months ago, it was not hard to find author after author explaining in laborious detail how China had finally perfected a growth model that looked very different from the Washington Consensus. The "Beijing Consensus" or "capitalism with Chinese characteristics" or "China model" of development included a prominent role for a bevy of state institutions, including national development banks, sovereign wealth funds, and state-owned enterprises. China’s impressive rates of growth fuelled a narrative of a country’s inexorable rise — so much so that U.S. representative Michele Bachmann praised Chinese policies, while Chinese leaders talked about the importance of re-reading Thucydides.

Then a funny thing happened — China’s growth model started to show its flaws, and they’re pretty serious. It turns out that state-led investment is not always the most productive way of boosting economic growth. Corruption and environmental degradation are pretty serious issues. Over the past few months, the fifth generation of leadership has taken steps to rein in out-of-control spending and borrowing. This has triggered its own narrative that China will now follow the path of Japan into fading challenger to U.S. hegemony. 

So, China is rising because of its economic model, or China is flailing because of its economic model? Here’s an alternative way of thinking about it — what if China’s new leadership is honestly trying to reform its model? Based on statements to date, it might be that China’s leaders are willing to take a short-term hit in lost growth to retool the economy more toward domestic consumption. This would mean slower but more sustainable growth in the future. The odds of such reforms being successful are unclear — President Xi Jinping is a fan of economic liberalization but quite opposed to any kind of political liberalization, and one wonders how the latter will constrain the former — but China’s development path to date has defied prediction so far. It would not be surprising if that continued.

That’s simply the most obvious example of a simple narrative trying to following examples where life does not exactly fit what people are saying about life. There are others:

The easiest prediction to make as a foreign affairs pundit is that Israeli-Palestinian talks will crash and burn. Maybe, just maybe, however, this is the one thing that Secretary of State John Kerry can accomplish in a world where American foreign policy is run by the White House.

The Obama administration’s belated decision to arm the Syrian rebels followed laments about how Assad’s forces were winning. Except that every time one side in this conflict seems to have the upper hand, the other side finds a way to rebound.

Over the past five years, the easiest pundit trick in the book is to claim that global governance has failed: see, in this calendar year alone, Mark Leonard, Mohamed El-Erian, Martin Indyk and Robert Kagan and, most recently, Clive Crook. The truth is actually far more complicated: in a lot of crucial areas, global economic governance has muddled through surprisingly well.

My point here isn’t to proclaim An End to Theory — indeed, that would be just another All Powerful Narrative that wouldn’t explain much. As Slate’s Ben Kenigsberg has pointed out, even in this season of box-office duds, some movie formulas still have some juice left. Lynch is correct to point out the power of political science models — but I’m equally correct to point out their limitations.