Tuesday, April 21, 2020

What We’re Not Talking About


By Kevin D. Williamson
Tuesday, April 21, 2020


Jason Richwine writes: “All predictive models have uncertainty, but it follows that the more uncertain the model, the less useful its predictions are. That should not be controversial.”

It would not be controversial . . . if that’s what the controversy were about.

Jason here is raising our old friend Bonini’s Paradox, which may be of some interest to readers: As a model becomes more complete, it becomes as difficult to understand as the underlying phenomenon it is intended to represent. I like Paul Valéry’s version: “Everything simple is false. Everything complex is unusable.”

When you are making a map, you leave out a lot of details because including all that complexity makes the map less useful rather than more useful — an accurate and comprehensive map of Chicago would be Chicago-sized.

The problem is that some details matter a great deal. “Go north on this road for 25 miles, and you will reach your destination” may be all you need to know. Some details are true but unnecessary and unhelpful (“There’s a right turn after the Walmart — you don’t want that. If you come to the Dairy Queen, you’ve gone too far”) and some aren’t. If there is a bridge out somewhere in the next 25 miles on that road, then that matters. Directions that leave that detail out aren’t necessarily untrue — the point you seek really is 25 miles north on that road — but they are not useful.

People bring their passions to their map-reading. They are very skeptical of claims that they do not want to be true and very generous toward those they do want to be true. Consider the myth that “all the polls were wrong” in the 2016 presidential election. In fact, the polls were for the most part pretty solid, and the results in most cases ended up within the margin of error estimated by the pollsters themselves. (Interesting stuff on polling accuracy here.) “The polls are always wrong!” is something you hear from people who don’t want the polls to be right.

One stream of our political discourse right now pits self-aggrandizing progressives who are not themselves scientists but wish to associate themselves with the prestige of science (“We believe in science!” etc.) against right-leaning populists who abominate what they call “elitism” and who wish to lower the status of certain elites and elite institutions (Silicon Valley, Wall Street, the Ivy League, etc.), a face-off that produces very little more than a mutual and self-reinforcing sneer. Our friends on the Left can turn anti-expertise just as quickly — see Alexandria Ocasio-Cortez on why it is more important to be “morally right” than “factually correct.”

I remember a few years ago a headline in the New York Times that read something like: Why are the jails so full when crime rates are going down? The answer from those who advocate more aggressive sentencing and law enforcement was: Crime rates are going down because the jails are so full. Criminal justice measures at the time were relatively aggressive, and crime rates were going down — everybody agreed on the facts, but there was no broader agreement. Many people believe very deeply that New York City’s crime rates went down dramatically at the turn of the century because of the policies of Rudy Giuliani, who used to be a mayor before he became whatever it is he is today. But crime also went down in similar cities around the country and around the world, including in places where no similar policies had been pursued. We have a great deal of information about that era and some plausible stories about why that wondrous change happened in New York City. But the fact is that nobody really knows, and what people “know” is mostly determined by cultural and political affiliations, not by knowledge.

Mostly this is a status game, hence Joseph Vinetz’s sneering that Alex Berenson should “stick to his novel writing.” Berenson is a former New York Times reporter. “That writer who writes things should stick to writing things instead of writing things” is a very odd line of criticism to make. I am sure that Dr. Vinetz is excellent at what he does, but there isn’t anything about an M.D. that necessarily qualifies one as a public-policy authority, as opposed to a medical authority.

Many of Berenson’s claims are not medical in nature at all, e.g. that “the companies that run the tests say they already have spare capacity.” That’s a question that can be answered (a claim that can be verified), and answering questions like that is journalism, not medicine. I myself take a much more risk-averse view of this epidemic than Berenson does, but conversations like the one above are not about risk-mitigation at all. Those conversations are about turf-defending, and once the conversation has devolved into that, then it is rare for any useful exchange to happen at all.

And that is where Paul Valéry has it wrong: Grotesque oversimplification can be useful — not if your goal is understanding the world, but if your goal is holding political power.

No comments: