Innovation in Teams

Patents, papers, code, and comics

Between 1980 and 2018, the number of inventors per US patent doubled (see figure, calculated from PatentsView data). Is the rise of teams good or bad for innovation? Against the notion that individual brilliance generates the best ideas, a series of papers find that bigger teams tend to produce higher impact innovation.

Singh and Fleming (2010) compares the patents of individuals and teams over 1986-1995. Compared to the patents of individual inventors, they find patents with teams of inventors are:

• more likely to be in the top 5% most heavily cited

• less likely to receive zero citations

Ahmadpoor and Jones (2019) extend this result to papers (and a lot more patents). For teams of 2-5 people, the bigger the team the higher the citations the paper/patent receives (though the extent varies by field).

The figure below gets at this: it’s the distribution across fields of the citation bump received relative to an individual (the horizontal axis), after controlling for the quality of the team. Papers left, patents right.

As we go from 2-person teams (black line) to 5-person teams (green line), the distribution is pulled to the right: lots more fields where the citation bump is bigger.

Turning to a completely different creative domain, Taylor and Greve (2006) look at the market value of comic books published between 1972 and 1996. Comics with a bigger team of writer/artists were more valuable on average.

What makes big teams so good? One possibility is the most talented innovator sets the agenda and direction of research for the whole team. Maybe talented innovators attract more people eager to work with them, and end up on big teams more often? Alternatively, if teams are assembled at random, the expected max talent is higher for a larger team.

But no; Ahmadpoor and Jones develop a flexible way to decompose the performance of the team into the performance of individual members. They find it’s not the most “talented” person, but rather the least talented one that has the most weight in determining how a team project gets cited.

(What do we mean by talent here? It’s actually “individual citation impact.” They’re looking at how highly cited a researcher’s output is as they are part of different teams - some people are usually on projects that are highly cited, others on projects that tend not to be.)

This figure for two-person teams makes the point well. The greater the difference between the individual citation impact (the thing I’ve loosely called “talent”) the lower the citations received by the project. Papers left, patents right.

If the least “talented” person disproportionately determines the outcome of a project, then the act of teaming up itself must be what brings something to the table (otherwise, why would the most talented ever team up?). The obvious candidate is different knowledge and expertise. Maybe high impact ideas require lots of knowledge and it’s easiest to get that by bringing lots of people together.

But this poses a complication. One way to think about creativity is that it’s about making connections between previously disconnected ideas. This process is often believed to occur inside an individual’s brain, at the unconscious level (called “incubation” in the psychology of creativity).

But what if the ideas are split across different people’s heads? Does this still work? Maybe not so well.

Taylor and Greve (2006) try to measure the creativity of comics by the variance of comic book valuations (lots of hits and duds). When they look at how many genres a the comic-book’s creator(s) have previously worked on, they find this is positively correlated with creativity for individuals, but not for teams.

Much more systematically, Wu, Wang, and Evans (2019) look at how the size of team is related to how “disruptive” patents, papers, and software code are. Here, disruption is an index based on how much your work is cited on it’s own, and not in conjunction with the stuff your work cited. (Intuitively, your contribution rendered the previous work obsolete)

Like the previous work covered, they also find big teams get cited more (purple line below). But looking at their disruption measure (green line), individuals and small teams produce more disruptive work than big ones.

The trouble is all this work is descriptive. If anyone knows of research that looks at creativity of teams with an experimental design, please let me know. I suspect there is a world of psychology research I’m not familiar with that does this.

Until then; my reading is that disruptive and highly creative work may not be served well by big teams, but not all work needs to disruptive and on the whole teams do very good work.

Thanks for reading!