Asymmetric Weapons Aren’t Always on Your Side

Some time ago, Scott Alexander wrote about asymmetric weapons, and now he writes again about them. During these posts, Scott repeatedly characterizes asymmetric weapons as inherently stronger for the “good guys” than they are for the “bad guys”. Here is a quote from his first post:

Logical debate has one advantage over narrative, rhetoric, and violence: it’s an asymmetric weapon. That is, it’s a weapon which is stronger in the hands of the good guys than in the hands of the bad guys.

And here is a quote from his more recent one:

A symmetric weapon is one that works just as well for the bad guys as for the good guys. For example, violence – your morality doesn’t determine how hard you can punch; they can buy guns from the same places we can.
An asymmetric weapon is one that works better for the good guys than the bad guys. The example I gave was Reason. If everyone tries to solve their problems through figuring out what the right thing to do is, the good guys (who are right) will have an easier time proving themselves to be right than the bad guys (who are wrong). Finding and using asymmetric weapons is the only non-coincidence way to make sustained moral progress.

One problem with this concept is that just because something is asymmetric doesn’t mean that it’s asymmetric in a good direction.

Scott talks about weapons that are asymmetric towards those who are right. However, there are many more types of asymmetries than just right vs. wrong—physical violence is asymmetric towards the strong, shouting people down is asymmetric towards the loud, and airing TV commercials is asymmetric towards people with more money. Violence isn’t merely symmetric—it’s asymmetric in a bad direction, since fascists are better than violence than you.

This in turn means that various sides will all be trying to pull things in directions that are asymmetric to their advantage. Indeed, a basic principle in strategy is to try to shift conflicts into areas where you are strong and your opponent is weak.

For instance, people who are good at violence benefit from things getting violent. People who are locally popular benefit from popularity contests. People who have lots of free time benefit from time-consuming processes. People who are better at keeping their composure benefit from discourse norms that punish displays of emotion.

Developing asymmetric processes that point towards truth is a good idea, and I’m all for it. But in practice there are also asymmetric processes that point towards error, or merely asymmetric processes that point towards what’s currently popular or faddish. Those processes are, if anything, just as likely to have people trying to promote them than the pro-truth ones—perhaps more likely!

That doesn’t make the people promoting those ideas “anti-truth” or whatever—they may not even be aware of what they’re doing—but even so, people tend to respond to incentives, and those incentives may well pull them towards norms and methods that are asymmetric in their favor independent of whether those norms and methods promote truth.