Jim Manzi Makes the Case for Doing Whatever We’re Doing Right Now — Or Nothing
I have to start this post by saying how much I like Jim’s recently-bruited notion (and coinage): “causal density.” I’ve been sharing it with my friends. In my words:
An event in physics — a ball being bit by a bat and landing in center field – has very few causes, so it’s pretty easy to deduce and explain those causes. Chemistry is significantly more complex, but still largely likewise.
Get into biology, and you’re very quickly into a lot of causes (what causes diabetes?), and those interacting with each other in extremely complex ways. It’s much harder to deduce causation, or reduce it to a simple or even straightforward formula or algorithm — or to make predictions based on particular interventions.
Jim suggests that the social sciences have even higher causal density. My example: What caused the decades-long decline in crime rates? Better policing techniques? Shifts in law-enforcement and judicial budgets? Harsher sentencing laws? Demographics? Roe v. Wade? Update: unleaded gasoline?
Despite yet more decades of regression analyses trying to “control” for all those factors and many more, controlling for those factors’ interactions makes the quest for cause deucedly difficult, perhaps even Quixotic. This especially when you consider all the factors that might not have been considered or controlled for: cf. Roe v. Wade, until Levitt came along, or the effect of genes on character development, before the explanations of evolutionary psychology — especially pre-Trivers.
I think he overstates his case some (economics does have some quite good predictive abilities in particular areas — way better than we had a century ago), but still I agree that it’s a difficult issue. (Though I give a tentative and perhaps only somewhat useful response here.) But what concerns me most is what I discern to be the unstated implication of his posts:
Since we can’t predict the results, we shouldn’t do anything.
This thinking is rooted in a fundamental (and not totally crazy) belief among conservatives: that we’ve spent decades, centuries, millennia working out our social, economic, and cultural systems, so they should get extra credence in deciding what to do next. (I’m pretty sure I’ve heard Jim make this argument, but if not we’ve all certainly heard it from other conservatives.) The best bet, probabalistically, is to maintain the status quo. (This closely related to a basic dictum of efficient-market theory: absent any new information, the most likely tomorrow price for something is … today’s price.)
Here are a few ways that I have a problem with that:
1. According to Jim’s own thinking, where we happen to be at any given moment is the result of an infinitely complex mix of interrelated causes and effects — some of which we’ve had control over, but many of which (especially the interactions) were and are completely beyond our ken. So our current position is to a great extent the result of luck, chance, fortune: an accident of history. Saying otherwise is to invoke a rather Panglossian and teleological belief system.
2. Which moment in history embodies that Panglossian moment? Today? This is directly at odds with the conservatives’ stated beliefs: that things were much better before the New Deal. Who gets to choose the conservative moment? Who decides that the status quo of the current moment (or some other) is the best yet, and that we should cling to it?
3. Favoring the status quo is directly at odds with America’s (humanity’s?) greatest virtue and capability: trying new ideas and approaches in an effort to progress and make things better.
4. The current moment can be very far from Panglossian — think Soviet Russia, or the radical, free-market- and government-capture-driven wealth and income inequality in America today. Since we don’t know with certainty what policies will fix that broken moment, should we do nothing about it? (But then again, I think we — with the exception of Republicans in their self-delusion — do know with reasonable certainty many of the answers for American inequality and its drag on growth. We know them largely because of research in economics — particularly econometrics.)
5. (Added Nov. 22) Doing nothing is doing something: it’s actively choosing whatever — to some extent by historical accident — happens to be in place at the moment. Who is to say that the unintended consequences of that choice will be superior to the unintended consequences of some other choice? We have to make a choice; the only question is what we should use as the grounds for that choice. Should systematic, rigorous economic analysis — despite all its failings — be considered in making that choice?
We’ve faced a decidedly non-Panglossian series of moments over the past decade: health-care costs going through the roof, and projected to continue, threatening to shatter the American way of life. And in the meantime the Republican party had its blank piece of paper — an almost total carte blanche — for six years. The “unintended consequences” and “history is our best guide” reasoning led to exactly what Jim’s posts suggest and imply: doing nothing. I don’t think he would argue that it was the optimal policy approach.
Now Jim might say that he’s not implying that we should cling to the status quo or do nothing. But he rather leaves us hanging. If the implication that I deduce is not what he intended to suggest, what was his intention? Simply to debase economic analysis as an input to our judgments, in favor of — for instance — casual lifelong observations?
It seems likely. He says that he came to his belief in the superiority of free-market solutions “through some combination of historical reasoning, introspection, practical experience and so forth.”
I’m completely agog. He doesn’t even mention systematic (or statistical) analysis — by himself or others. I find it hard to believe that Jim has never read an econometric analysis, or that he has been completely uninfluenced by those he has read. He’s not that kind of guy.
Understand: Jim is co-founder, chairman, and managing director of Applied Predictive Technologies. To quote Wikipedia (wonder who wrote this…), “APT’s software takes a statistically rigorous test and learn approach…”
Now of course it’s true: we can’t design and implement controlled tests of economies, which makes his company’s work of a decidedly different order. But does that mean that econometric analysis has no place — at all — in informing our judgments? That we should simply ignore it? That’s what Jim seems to be saying. But again: I really can’t believe that he does that.
If we review hundreds of systematic, rigorous econometric studies looking at many different data sets using many different analysis methods — and even have the benefit of systematic, rigorous reviews of all those studies — and we find that X seems to have no correlation with Y, should we draw the conclusion that changing X probably isn’t going to change Y in the future? Or should we ignore that finding, and rely instead purely on “historical reasoning, introspection, practical experience, and so forth”? In other words, casual, non-systematic observation and surmise? Should we do so even if that surmise is contradicted by all the systematically rigorous analysis?
Manzi admirers want to know.