Are you capable of adversarial collaboration?

That is the question. The answer is “maybe not.”

But let’s not jump to conclusions. First you may want to know what adversarial collaboration is, and second why you would want to be capable of it. So let’s start there.

1. What is adversarial collaboration?

Adversarial collaboration is when two or more people seek to resolve disagreements about an issue X by defining a mutually satisfactory way to learn more about X together, and then documenting what they’ve learned. Those people are often professional academics or scientists, but there’s no reason why they couldn’t be you and your mad Uncle Angus. The issue X is often a matter of some controversy in the scientific literature, but again there’s no reason why it couldn’t be something like the birthplace of the 44th president of the United States, or recent changes to the rate of monthly job creation in your state. The beauty of the thing is that it is available to us all, princes and paupers alike, for our mountains and our molehills. All it takes is two willing people.

The crux of adversarial collaboration is an openness to learning from your adversary. That may be asking too much. I am typically unwilling to learn from my friends. To learn from my adversaries? That is two bridges too far. But maybe you are better. Even if you are not open to your mad Uncle Angus, could you be open to your cousin, your mother, or your significant other?

Don’t answer that.

Fortunately, you do not need the mindset of openness to get started. You can take the first steps inflated by the conviction that your adversary will be the one doing the heavy learning. They have so much to learn, they just don’t know it yet.

Of course, they are welcome to hold the same conviction about you.

The first part of the collaboration is agreeing up front how you might learn more about the matter in question (e.g. looking up the definition of “spy” on Wikipedia; or designing a randomized controlled trial that tests an treatment against a control group). The second part of the collaboration is going and doing what was agreed upon. The third part is documenting what each side learned from the shared experience so that it can be offered as a testimony to others.

2. Why would you want to be capable of adversarial collaboration?

Because it will help to kill your ego, which is causing you and those around you so much pain and frustration. The enemy of your ego is your friend. 

Because you will learn things that you would never learn on your own. This will make you a more interesting person, more open to the world around you. It is trying to teach you things, if you have ears to hear and eyes to see.

Because the alternative is to continue your life (a) avoiding difficult conversations, or (b) forcing your way through them unproductively. If you feel lots of resentment, you may have a habit of avoidance. If you feel righteous anger, you may have a habit of aggressive assertiveness. There’s no need to choose one or the other. A long train of under-reactions can trigger an over-reaction. You can have bitterness and bile too.

Adversarial collaboration offers a middle way between “flight” and “fight.” You want to be capable of walking that line, both for your sake and for those that depend upon you.

3. Are you capable of adversarial collaboration?

And this brings us to the initial question. And the answer there is “maybe not.”

How would you know?

The question is whether you have the ability to overcome the challenges that wait for you at every step on the journey: (a) the awareness challenge, (b) the heroism challenge, (c) the planning challenge, (d) the execution challenge, and (e) the learning challenge.

a. The awareness challenge. To begin, you must be aware that it can be hard to know things.  If you believe that it is easy to know things, then you will infer that anyone who disagrees with you is ignorant, stupid, greedy or evil. Once they resist your attempts to educate away their ignorance, you’ll infer they’re stupid, greedy or evil. And they will feel that from you.

If you are aware that knowing can be hard, then it becomes more likely that (i) your views are incorrect or incomplete in some ways, (ii) your adversary’s views may be more correct or complete in some ways.

But it’s not enough to be aware of this in abstract. That part is easy. It is recovering that awareness in the moment of disagreement that is hard. When a disagreement rears its ugly head, can you think “O lovely, an opportunity to learn!”? Maybe not. But with practice, you can.

b. The heroism challenge. Knowing things may be hard, but saying things is easy. It is often intoxicating to assert truths pulled from – ahem – the foggy recesses of our minds. Many of us, myself included, have a lot of fog to pull from. It’s great fog, the best fog really. It can be intoxicating to tell a hard truth. It feels heroic.

And this makes it addictive. Or, shall we say, habit-forming. A short-cut to heroism has opened up before us. I carry the Sword the Truth! Now is the time to use it! And now! Also, now!

What would happen if you set down the Sword of Truth? Could you cut back on all that heroism and still survive? Maybe not. But with practice, you could.

c. The planning challenge. Even if you’re able to restrain all that bold truth-telling, there comes the challenge of aligning on a plan to (i) define the issue, (ii) surface areas of agreement and disagreement, and ideally (iii) test or reconcile some of those disagreements. There will be many ways to do each of these, so you will have meta-disagreements to iron out. What’s the scope of the thing? All of the southeast or Florida only? Since 1980 or 2000? What are the relevant sources of facts and/or insight? Does it have to be randomized controlled trials or would observational studies suffice?

If you are not expecting any meta-disagreements, you may throw your hands up in despair at this point. It’s just disagreements layered on disagreements layered on disagreements! How very exhausting. Are you capable of agreeing on process and method? Maybe not.

But with practice, you could be. A lot disagreements look similar at the meta-level, so you can walk a path you (and others) have walked before, no matter how unique the particulars.

In fact, working out these disagreements may give you more confidence in one another’s good faith.

d. The execution challenge. Even if you get through the planning, there are the challenges of execution. The challenge may be inside the work, e.g., a key source was too costly to obtain, or a key resource was unavailable for a scheduled experiment. Or, more likely, it may be outside the work: the competing commitments for your time and attention. This may delay or kill the collaboration altogether. This challenge is not unique to adversarial collaboration, of course. It causes the delay and death of projects far and wide without regard for their stated objectives, their process steps, their hopes and dreams.

e. The learning challenge. Even if the work gets done, there may be emotional resistance to conceding – even to yourself – that you have learned anything. This is where the ego makes its last stand. If you are determined not to learn, it is amazing the things you can not learn for yourself!

The more specific you were about your expectations in the planning phase, the easier it is to be surprised by the results. Surprise is an emotional marker of learning. Are you capable of being surprised? Are you capable of being wrong?

Of course, there are many defenses we can marshal to put bounds on just how wrong we were. There are the variations on “I was almost right” as well as variations on “I was wrong but made the right mistake.” As Phil Tetlock has documented, many of us are happy to entertain counterfactuals when we are “almost right” that we would or do not when we are “almost wrong.” The more of these defenses you seek to invoke, the more your collaborator may seek to invoke as well. The slow back-and-forth of the write-up offers opportunities for mutual disarmament.

If agreement on the lessons learned was somehow required, the conversation could easily grind down into a bitter stalemate. But agreement is not required. Each side can offer their own synthesis of their lessons learned, and propose what additional evidence or testing would change their minds.

In some ways this last challenge is bringing you back to the origin, to the openness to learning from your adversary.

Unaided, this can feel like trying to leap over a 10-foot wall. Now, however, the scaffolding of the collaborative process has reduced the emotional leaps required to learn. You can take it step by step, knowing that no step will ever be more than a foot or two above the previous one. You can do it!

This post was inspired most immediately by the Adversarial Collaboration contest (also the update here) that Scott Alexander is organizing at Slate Star Codex. My longer term interest in adversarial collaboration was inspired by Philip Tetlock’s work on expert knowledge and superforecasting, and Daniel Kahneman’s adversarial collaboration with Gary Klein.



Leave a Reply

%d bloggers like this: