Small teams disrupt? Maybe it was never about size

You’ve heard the line even if you never opened the paper. Small teams disrupt, large teams develop. It comes from Wu, Wang and Evans in Nature in 2019, and since then it has been quoted in grant panels and on conference stages every time someone argues that science has grown too big to be brave. Small and scrappy breaks new ground. Big and well funded refines it.

A new preprint says the size part was mostly standing in for something else. Bili Zheng and Jianhua Hou, in “Synergy, not size”, argue that what drives disruptive work is not how many people sign a paper but how the collaboration is wired. It went up on arXiv in September 2025 and has not been peer reviewed, so read it as a strong claim, not a settled result.

What Zheng and Hou actually did

They looked at more than 14 million papers across 19 disciplines from 1960 to 2020. Instead of counting heads, they built what they call a synergy factor, a way to score the cost and benefit of adding people to a team. Then they ran a mediation analysis, which is a formal way of asking a simple question: when team composition predicts disruption, is size doing the work, or is something that size is merely correlated with doing the work? Their answer is that synergy, not team size alone, accounts for 75 percent of the link between who is on the team and how disruptive the paper turns out to be.

The best team size is not one number

The result that stuck with me is that there’s no single best size. Physics peaks at medium sized teams. The humanities reach their highest synergy through individual scholarship, one author alone. So “small teams disrupt” really means that the right architecture depends on the field, and in some fields the right architecture is a single person.

Star authors help, but not the ones you would rank highest

Two more numbers. Papers that include an exceptional researcher show, by their measure, 561 percent higher disruption. And a twist that should make anyone who reads CVs squirm: high-citation authors were linked to less disruptive potential, while authors with a track record of breakthroughs were linked to more. Being cited a lot and doing something new are not the same signal. Zheng and Hou sort teams into four modes: elite-driven, baseline, heterogeneity-driven, and low-cost.

So was “small teams disrupt” wrong?

Not really. This is not a refutation. It is a rewrite of the caption. Size was never the cause. It was a visible proxy for how a team combines skills and viewpoints. Small teams disrupt more, on average, because small teams are often wired for synergy by default. Build a large team that keeps that wiring and, in principle, you keep the disruption. That is a more useful claim than “stay small,” because wiring is something you can actually decide.

What I would not bank on yet

Three caveats, in order of how much they bother me. First, it is a preprint. No peer review, no independent replication that I’ve seen, and a 561 percent effect is exactly the kind of number a reviewer tugs on to see what falls off. Second, it is observational. Fourteen million papers is a mountain of correlation and zero experiments. Nobody randomly assigned scientists to teams. Third, and this is the one I keep circling back to, “disruption” here is a citation-pattern score, not a verdict on importance. The disruption index asks whether later work cites you instead of the things you built on. That captures something real, it also misses plenty, and the index has its own critics. When the paper calls a team disruptive, it means a shape in the citation graph, not a Nobel.

If you are choosing who to work with

Here is why a working researcher should care. If you’re picking co-authors, the size heuristic is easy and mostly wrong. This preprint’s version is harder and more honest: ask what each person adds that the others cannot, and whether adding them raises synergy or just raises the author count. A collaborator with one real breakthrough behind them may matter more than three with fat citation totals. And if you work in a field where solo work still disrupts, the reflex to bolt on co-authors may be quietly costing you the exact thing you were trying to buy.

I would hold all of it loosely until it clears review. But it does reframe a line I keep hearing repeated as if it were settled physics. Maybe the question was never how many people are in the room. Maybe it was always what happens once they are.