Many people use arguments as soldiers.

Just because you're a pacifist doesn't mean there isn't a war on.

Even if you aren't personally in the trenches lobbing grenades, you can still affect the war -- even if you don't mean to. You need to be careful who you support, or else you might accidentally support people who commit war crimes.

Support for a given faction can come in many forms. It can mean nitpicking opposing arguments, or steelmanning, or teaching people how to debate smarter and more respectably, or any of countless other forms.

In isolation these things can be good, but in context it matters if there's a bias as to which arguments and argumenters receive these benefits. If you give aid and comfort, charity and courtesy, to only one side of the debate, then that makes them more likely to win.

(You are a consequentialist, right? If the actual results of your actions are that bad guys win, it doesn't matter if you were following all the rules of courteous debate.)

If you want to support peace, you have to acknowledge that there are more ways to support war than firing a gun.


So what does this mean in practice?

  • Jargon. When you use the memes and register of a faction, you signal agreement with that faction -- even if you're saying that you disagree with them. Don't signal agreement with someone you think is wrong, and remember that just because someone is wrong doesn't mean their main opponents are right.
  • Timing. When you partially disagree with someone who's mostly right, or at least one of the least wrong people in the debate, make sure you only bring up the quibble in such a way that it won't seem to bystanders like a rebuttal to their main point. A major part of this is where and when you bring it up; as far as possible, try to keep quibbles separate from the main discussion.
  • Time and weight. Choosing to focus on a given position -- even to disagree with it -- tends to give it a degree of legitimacy. After all, the reasoning goes, you wouldn't put so much effort into debunking it unless you had been seriously considering the possibility that it might have been true.

Go forth, and sin less on an expected-value basis.