This is my distillation of rationality community wisdom on how to disagree. For this audience I fear this is all obvious or has been said better elsewhere. But I've been extolling the benefits of articulating insights in one's own words, so I'm taking my own advice. The eightfold path is as follows:

1. Collaborate on characterizing the disagreement
2. Clarify your cruxes
3. Subject yourself to ideological Turing tests
4. Steelman your opponent's position
5. Empathize and seek common ground
6. State what you've learned from your opponent
7. Rebut and criticize only after all of the above
8. Celebrate mind-changing

The overarching theme is seeking truth rather than a debate victory. From the first part of the first step — collaboration — we're cultivating the scout mindset. You want to literally love being wrong. David Heinemeier Hansson, of Ruby on Rails fame, in a blog post titled "I love being wrong", puts it like this:

Being wrong means learning more about the world, and how it really works. It means correcting misconceptions you've held to be true. It means infusing your future judgements with an extra pinch of humility. It's a treat to be wrong.

A debate in which both sides are eager to understand each other and genuinely excited about the possibility (however small it is) of being wrong feels amazing.

Alright, so let's discuss each of these steps in more detail.

First, you'll want clarity on whether you're disagreeing about facts, definitions, predictions, or values. Many disagreements immediately dissolve just by doing this. Take the classic debate about a tree that falls in the forest without anyone hearing it. Does it make a sound? There may be a fun debate to be had there but you'll want to start by clarifying that it's a debate over definitions. Should "sound" refer to acoustic vibrations or an auditory experience?

Step two, clarifying your cruxes, means articulating what would make you decide you are wrong. For example, most of my nerd friends think Daylight Savings Time is a transgression against Nature and that Benjamin Franklin^[1] has the blood of generations on his hands. I think DST is an ingenious hack that solves a coordination problem with no other viable solution. But I can outline a cost/benefit calculation with traffic deaths and programmer hair loss on one side and how much people would collectively pay to hit the snooze button on the sun setting in the summer on the other side. If the costs outweigh the benefits, then I'm wrong. In general, be explicit about things that would hypothetically/counterfactually change your mind. A debate with anyone who can't do this is quite futile!

(Of course you get extra bonus points if you can state your crux as a wager. As the saying goes, betting is a tax on bullshit.)

Ideological Turing tests, step three, are amazing, just as a concept. You should understand your opponent's position so well that you can pass as a believer. It's tragicomic how abysmally people commonly fail this. The loudest people in the abortion debate in the US, for example, seem to believe their opponents have a bloodlust for babies, or are seizing on an excuse to oppress women, respectively. 

But you don't need to formally administer an ITT (though I'm not saying not to — it sounds super fun). Here's how Daniel Dennett described what I'm calling step three of the eightfold path — the first of Rapoport's rules for ethical debate:

You should attempt to re-express your target's position so clearly, vividly, and fairly that your target says, "Thanks, I wish I'd thought of putting it that way."

Number four, steelmanning, takes that even further. Remember the overarching theme: collaboration and truth-seeking. And recall that the strawman fallacy is when you argue against a weak or stupid version of your opponent that's easy to knock down. You want to do the exact opposite. Can you improve on their arguments? You want to be wrong, after all. If there's a version of your opponent's argument that defeats yours, that's the version you want. Collaborate with your opponent to try to find that version.

Empathizing and common-ground seeking (step five) are self-explanatory. Both do wonders for setting the tone of a debate and keeping everyone in scout mindset. Don't just perfunctorily list things every sane person agrees on. Uncover the most surprising or uncommon points of agreement you can find.

Telling your opponent what you've learned from them is step six. And if you can't find anything to learn from them, well, what are you even doing? How is the debate worth your time? The part about actually telling them what you've learned is, like step five, about fostering scout mindset and collaboration.

After all that, we get to rebuttals and criticism in step seven. Phew. Again, the point is to conscientiously hit all six previous steps before you allow yourself to rebut and criticize. I'm basing this part in particular on the Rogerian model of argument.^[2]

Finally, rounding out the eightfold path is a major theme of Julia Galef's book (The Scout Mindset — and I'll link again to Scott Alexander's excellent review of it): Never try to catch someone out on their old views. Help cultivate norms in which changing one's mind is something to be truly proud of. Learn to love being wrong.

1. ^{^}

   UPDATE: So embarrassing, I didn't think to check this till after hitting publish, but apparently Benjamin Franklin didn't really propose Daylight Savings Time. It seems he just made a snarky comment about how Parisians should wake up earlier to save money on lamp oil.

   FURTHER UPDATE: I mean, not embarrassing, I was wrong and that's amazing go me.

2. ^{^}

   Thanks to Theo Spears for pointing me to the Rogerian model.