Among many of Jeff Bezos’s wisdoms, his two-pizza rule is one of the most famous. It states that every internal team should be small enough that it can be fed with two pizzas–to make meetings more effective and the teams more innovative.
When it comes to scientific research, the world seems to disagree with Bezos. As large-scale scientific collaboration become increasingly popular, research groups grow bigger and bigger. A world record seems to belong to a physics paper published in 2015: 5,154 authors. Only the first nine pages of the 33-page article describe the research itself; the rest (24 pages) lists the authors and their affiliation. Just imagine how many pizza parlors would have to work overtime to provide a single lunch for this tight-knit group of collaborators!
The proponents of “the bigger, the better” approach can point to at least one argument in support of their position: a positive correlation between a team’s size and a citation impact of a product that this team had produced. This correlation holds not only for STEM research, but also for social sciences, art and humanities, and patents. The bigger the team of collaborators, the greater the buzz their work is generating.
And yet, Bezos might have had the last laugh—as usual.
A group of authors led by James A. Evans of University of Chicago designed an advanced, more nuanced citation-based index capable of discriminating between “disruptive” and more conventional (“consolidating”) research contributions. Their logic was as follows: when future citations to a given article also reference a substantial proportion of that article’s own references, then the article can be seen as consolidating its scientific domain. However, when future citations do not acknowledge the article’s own references, the article can be seen as disrupting its domain.
Evans and his co-authors analyzed more than 65 million articles, patents, and software products generated over 1954–2014. They show that smaller teams tend to come up with more new ideas and opportunities (disruptive contribution), whereas larger teams tend to develop existing ones (consolidating contribution).
The authors also show that work from larger teams often builds on more recent and popular developments, so that attention to their work comes immediately. By contrast, contributions by smaller teams go more deeply into the past and, if successful, project further into the future.
My first impulse is to apply the result of Evans and his collaborators to the incremental vs. disruptive innovation dichotomy. When pursuing incremental (so tempting to say, consolidating) innovation, organization would seem to benefit from creating larger teams that could rapidly expand on recent product development gains. On the other hand, achieving disruptive innovation goals would be more plausible by establishing many small groups pursuing diverse projects—mimicking essentially the approach used by VC investors.
I also wonder if eating a particular type of pizza—say, Neapolitan vs. New York-style—would benefit a specific type of innovation.
 For example, the index shows that articles directly contributing to Nobel prizes tend to exhibit high levels of disruptiveness; at the other extreme, review articles tend to be highly “consolidating.”