Stop arguing over data and try Adversarial Collaboration

Methodology

I see it all the time in my line of work. Two people look at the same set of facts and come to completely opposite conclusions, then spend three years writing papers about why the other person is wrong. It is a waste of time. Most of the time, the problem isn't the data; it is the way the study was set up to support a specific narrative. There is a better way called Adversarial Collaboration. Instead of fighting after the fact, researchers with opposing views sit down and co-design a single study. They agree on the methodology and, more importantly, they define exactly what the results need to look like for one side to be proven right. Here is how it works in practice: First, the opposing parties agree on a specific, testable hypothesis. Second, they design the experiment together. This stops the usual arguments about sample sizes or weak controls that happen once the results are already out. Third, they pre-register the success criteria. They decide, before any data is gathered, that if X happens, Party A is right; if Y happens, Party B is right. Fourth, they run the study and agree to honor the outcome regardless of their personal feelings. It turns a shouting match into a structural problem. It is like two contractors agreeing on the plot stakes before they start digging. Once the stakes are in the ground, you stop arguing about where the house goes and just start building.

4 comments

Comments

SkepticalMike·2 hours ago

This assumes a compromise design doesn't dilute the effect size. If both parties agree to a middle ground methodology, you risk a type II error where neither side is proven right.

LurkingLorraine·2 hours ago

pre-registration of the power analysis prevents that.

HotTakeHarvey·2 hours ago

Why are we still talking about collaboration when the real issue is the incentive structure of tenure? Will a university actually promote a researcher who admits their rival was right in a co-authored paper?

CuriousMarie·2 hours ago

maybe there's a way to credit both authors for the result regardless of who won... could a new type of collaborative resolution citation exist?