Matt Steele

Marginal Revolutions

26 Mar 2017

Weekend illnesses are a great opportunity to work through an Instapaper queue. I'll blame the head cold, but I see connective tissue across several articles.

First, Scott Alexander makes a compelling case that Trump supporters aren't immune to logic and truth. One presumes they are, because when's the last time you debated a Trumpist and changed their minds?

But that's because these debates usually entail a handful of drive-by "CHECKMATE ATHEISTS" encounters. Rather, one should focus on a longer-term, convince-one-person-over-five-years approach:

Am I saying that if you met with a conservative friend for an hour in a quiet cafe to talk over your disagreements, they’d come away convinced? No. I’ve changed my mind on various things during my life, and it was never a single moment that did it. It was more of a series of different things, each taking me a fraction of the way. As the old saying goes, “First they ignore you, then they laugh at you, then they fight you, then they fight you half-heartedly, then they’re neutral, then they then they grudgingly say you might have a point even though you’re annoying, then they say on balance you’re mostly right although you ignore some of the most important facets of the issue, then you win.”

Instead, we throw up our hands and proclaim we're in a post-truth era, because the longer-term approach is hard:

Improving the quality of debate, shifting people’s mindsets from transmission to collaborative truth-seeking, is a painful process. It has to be done one person at a time, it only works on people who are already almost ready for it, and you will pick up far fewer warm bodies per hour of work than with any of the other methods.

Next, in The Heroism of Incremental Care Atul Gawande argues that our health care valorizes specialties with obvious, immediate results (think: surgery, orthopedics) over longer, more incremental approaches (internal medicine, immunology). And yet, the longer-term approach saves more lives:

In the United Kingdom, where family physicians are paid to practice in deprived areas, a ten-per-cent increase in the primary-care supply was shown to improve people’s health so much that you could add ten years to everyone’s life and still not match the benefit.

The internists' secret weapon: access to way more data:

“It’s the relationship,” they’d say. I began to understand only after I noticed that the doctors, the nurses, and the front-desk staff knew by name almost every patient who came through the door. Often, they had known the patient for years and would know him for years to come. In a single, isolated moment of care for, say, a man who came in with abdominal pain, Asaf looked like nothing special. But once I took in the fact that patient and doctor really knew each other—that the man had visited three months earlier, for back pain, and six months before that, for a flu—I started to realize the significance of their familiarity.

Finally, my hometown of Omaha is looking to solve its public transit problems by building a new $150M streetcar system. It's a massive, big-bang solution to public transit woes, which is odd, given that Nebraska spends less on public transit than any other state.

Similarly, the city has let its road infrastructure crumble so much over the years that it's converting some roads back to gravel, and declaring them "unmaintained".


In each of these fields, the more successful approach is one that emphasizes slow, incremental changes. No one doctor appointment resolves a chronic condition. But the aggregate of routine visits captures more data, and solves more problems, than the alternative.

You see it in software, too. Most orgs celebrate the dramatic rewrite of an aging codebase. But to accomplish it they'll move resources away from longer-term projects, like making testing and continuous integration more consistent across the enterprise.

Making those long-term investments is hard. And it's an uphill battle. When short-term solutions work they're more dramatic, are easier to measure, and are more fun. So of course folks prioritize them.

But it's not just that the longer-term approach is more effective. It's that the alternative is often actively harmful.

If you don't think you can rationally convince a Trumpist they're wrong, then what's your alternative? Sophistry, emotional appeals, or shutting down discourse a la the antifa folks? But these approaches can be wielded by both sides; your side will only win by coincidence of having more effective storytellers. And Trump's camp sure seems to have more expertise in this area.

So try embracing the slow burn. This post might not convince you of its efficacy, but I'll leave it to you to find the one weird trick that does.