Now More Than Ever, We Need Steel-Manning

When considering an opposing argument, a “steel man” is much more productive and effective than a “straw man”

By Vaughn B. Baltzly

The early days of COVID-19 will be remembered, among other things, for their shortages—from TP to PPE. But one thing the coronavirus hasn’t depleted is our tendency to prognosticate: in this case, how things will look “when all this is over.” We’re awash in claims that we’ll never go back to “business-as-usual”; that this or that feature of everyday life will be permanently abandoned or altered; and that we’ll all have to adjust to some version of a new (and very different) normal.

At the risk of increasing this surplus, let me add one more prediction to the pile. Mine is not a claim about what the post-COVID world might look like in the longer term but the medium term—once the light at the end of the tunnel appears, when we have effective treatments or a vaccine or both and begin settling into the post-COVID era. During this time, it seems inevitable that there will be a tremendous amount of Monday morning quarterbacking: endless second-guessing and relitigation of many of the crucial, early-stage, high-stakes decisions made by our local, national, and international institutions. There will be withering critiques and passionate defenses, and few relevant leaders or institutions will be exempt.

When that day inevitably comes (and increasingly, it seems already to be upon us), we will all do well to practice the art of “steel-manning.” Steel-manning is of comparatively recent vintage—perhaps first appearing online less than a decade ago—and so it may be unfamiliar to many readers. But even those encountering it for the first time can likely easily work out its meaning, both from context and from its affinity with the more famous phrase from which it’s derived: straw-manning. Straw-manning, of course, refers to the intellectually dubious activity of constructing intentionally weak caricatures of opponents’ arguments for the express purpose of easily knocking them down. By contrast, steel-manning is the deliberate effort to construct, as a prerequisite of ever deigning to engage an opponent in argument, the strongest possible defense of that opponent’s position.

The phrase seems to have initially gained traction within the “Rationalist Movement,” a Bay-Area-centered community of programmers, AI-researchers, and other assorted types centered around (among other things) the causes of learning and disagreeing more efficaciously. But the term deserves—and now seems to be attaining—wider currency. For instance, Conor Friedersdorf introduced it to his readers in an Atlantic essay in 2017.

Some people conflate steel-manning with a number of related ideas, including the Principle of Charity, which is the standard of good-faith intellectual debate enjoining a person to take the strongest possible understanding of an opponent's position. The Principle of Charity applies in cases where you’re engaged in conversation with an interlocutor. In such circumstances it’s usually best practice to ensure that you’re interpreting your partner’s arguments with maximal fairness.

By contrast, steel-manning is a technique less to be practiced in dialogue and encounter and more to be practiced in reflection and preparation. You don’t steel-man this or that person’s arguments or positions in particular; you steel-man a given position or argument in general. “I don’t think I agree with Position Z,” you say to yourself, “but just to make sure—and certainly, before I embark on a crusade to discredit Z and its adherents, whether verbally or in print—I will first construct, and consider, the strongest possible version and defense of Z.” Mixing the two techniques, though—trying to steel-man an interlocutor’s position during live debate—might yield mixed results. For one thing, you risk sounding patronizing and insulting your conversation partner.

Another related notion is the Ideological Turing Test, first articulated in 2011 by George Mason University economist Bryan Caplan. You pass this test when you’re able to articulate and defend your opponents’ position in ways that are perfectly acceptable to them—in ways that perhaps might even convince them that you truly endorse their position. But again, and as with the Principle of Charity, there is an interpersonal dimension to this notion that need not apply in the case of steel-manning. While your steel-manning endeavors might certainly profit from proper engagement with real-life proponents of rival views, you needn’t necessarily do so. In any event, you certainly needn’t convince any such parties that you sincerely share their views.

Though the term itself may be of recent vintage, the basic idea underlying it has a respectable and familiar pedigree. Most prominently, invocations of steel-manning and its associated notions (like Caplan’s Turing Test) typically quote the famous counsel of British philosopher John Stuart Mill from his book, On Liberty: “He who knows only his side of the case knows little of that.”

Turning back to present-day applications, though, it seems nearly inevitable that the Monday morning quarterbacking of our institutions’ responses to this pandemic will swiftly harden into a variety of fixed, antagonistic, rival camps. Notwithstanding this seeming inevitability, I am still moved to urge all concerned that before attacking an opponent or a rival position, we please try a little steel-manning first!

Let me briefly sketch how this might work out with respect to some divisions that have already arisen and begun to calcify. An easy place to begin is the ongoing debate between those who are skeptical of the need for lockdowns and other restrictions to “flatten the curve” and those who are ardent defenders of these measures. The skeptics might begin their steel-manning with an acknowledgement that, in its early days, the virus did utterly overwhelm healthcare systems in two different parts of the world: Wuhan, China and Northern Italy. So, erring on the side of caution, for fear that this virus was far more deadly and/or contagious than similar diseases at the center of other recent outbreaks (SARS, MERS, H1N1), lockdowns and similar measures seemed reasonable. But defenders of these actions should acknowledge that the public was, in an important sense, “hectored” into accepting the wisdom of curve-flattening measures, even in spite of the evident failures to fully explain them.

For example, Few (if any?) versions of that near-ubiquitous “one simple chart that makes it obvious why you need to stay at home,” had any units on the x-axis. Without units on the x-axis, readers of the graph couldn’t tell how long the “curve-stretching” exercise they were being asked to accept would last. (The corollary of the curve’s being “flattened” relative to the y-axis, which measures the level of hospitalization-worthy infections, is its being “stretched” on the x-axis, which measures time.) Would the curve-stretching exercise last weeks, months, or years? Likewise, no end-game for the stay-at-home orders was ever proffered—even though it was obvious to many people, from the beginning, that the stated timelines (a couple weeks’ duration, e.g.) of the initial orders were implausibly short.

If you pursue a thoroughgoing policy of first steel-manning any potential opponents before engaging them, you’ll likely find that your resulting counter-arguments are more convincing, and you might even find a receptive audience among your opponents. This is steel-manning’s overt agenda.

But you also just might find that, having steel-manned an opponent’s position, you regard this position with a little more sympathy, and your opponent with a little less suspicion. This opponent might seem after all to not quite be so much a rival as an interlocutor—a partner in truth-seeking. And this is steel-manning’s covert—and dare I say true?—purpose.