Using the impact of the Soviet Union’s collapse on the performance of theoretical mathematicians as a natural experiment, we attempt to resolve the controversy in prior research on whether specialists or generalists have superior creative performance. While many have highlighted generalists’ advantage due to access to a wider set of knowledge components, others have underlined the benefits that specialists can derive from their deep expertise. We argue that this disagreement might be partly driven by the fact that the pace of change in a knowledge domain shapes the relative return from being a specialist or a generalist. We show that generalist scientists performed best when the pace of change was slower and their ability to draw from diverse knowledge domains was an advantage in the field, but specialists gained advantage when the pace of change increased and their deeper expertise allowed them to use new knowledge created at the knowledge frontier. We discuss and test the roles of cognitive mechanisms and of competition for scarce resources. Specifically, we show that specialists became more desirable collaborators when the pace of change was faster, but when the pace of change was slower, generalists were more sought after as collaborators. Overall, our results highlight trade-offs associated with specialization for creative performance.

Prior research suggests that academic scientists who collaborate with firms may experience lower publication rates in their collaborative lines of work due to industry's insistence on IP protection through patenting or secrecy. The main empirical challenge of examining the effect of industry collaboration on scientific productivity is that research projects that involve industry collaborators may be qualitatively different from those that do not. Hence, any difference in subsequent output of academic scientists who collaborate with industry may be driven by differences in the nature of research projects that attract industry collaborators. To address this issue, we exploit the occurrence of simultaneous discoveries where multiple scientists make roughly the same discovery around the same time. Following a simultaneous discovery, we compare the follow-on research output of academic scientists who collaborated with industry on the discovery with that of academic scientists who did not. We find that academic scientists who collaborated with industry produce more follow-on publications and fewer follow-on patents on their collaborative research lines than their academic peers who did not collaborate with industry. Our results suggest that research lines with both scientific and commercial potential provide an opportunity for a productive division of tasks between academic scientists and their industry counterparts, where the former focus on exploiting the scientific opportunities and the latter focus on the commercial ones. We also show that these effects are particularly salient when the industry partner is an established company rather than a startup.

Teamwork is an increasingly important aspect of knowledge production. In particular, factors influencing team formation relative to the composition of expertise are crucial for both organizational performance and for informing policy. In this paper, I draw attention to technology access as a highly influential factor impacting expertise in team formation. I examine the hack of Microsoft Kinect as an exogenous event that suddenly reduced motionsensing technology costs to find that great reductions in technology costs substitute for ex-ante optimal involvement of area specialists and facilitate involvement of outside area specialists through collaboration with researchers with broader knowledge – generalists. In other words, technology costs influence the composition of expertise in teamwork, with suficiently large reductions leading to knowledge creation that combines more broadly across knowledge areas. These findings suggest implications for organizations and policy makers in crafting incentives for more diverse knowledge creation through strategic investments that lower technology costs and influence team formation.

The fall of the Iron Curtain led to an influx of new mathematical ideas into western science. We show that research teams grew disproportionately in size in subfields of mathematics in which the Soviets were strongest. This is consistent with the knowledge burden hypothesis that an outward shift in the knowledge frontier increases the returns to collaboration. We also report additional evidence consistent with this interpretation: (i) The effect is present in countries outside the United States and is not correlated with the local population of Soviet scholars, (ii) Researchers in Soviet-rich subfields disproportionately increased their level of specialization.

We consider the role of the individual researcher in absorptive capacity and examine the impact of individual-level knowledge diversification on identifying and integrating novel knowledge in innovation. Our study is motivated by a paradox where impactful discoveries are the result of diversified knowledge input, yet achieving diversification is increasingly difficult as knowledge accumulation leads to division into narrower knowledge areas and increased specialization. We build on recent findings suggesting an important role for diversified researchers as coordinators in increasingly large teams of specialists, to consider their role in identifying and integrating novel knowledge in innovation. Using a natural experiment, we find evidence that diversified researchers have a higher propensity to engage with novel knowledge, and do so in a manner that produces more highly cited output. The effect is more pronounced when the novel knowledge is distant from the researcher’s past experience and requires exploration rather than exploitation.

We examine how an increase in the availability of IT-based research technology influences the production of research ideas and the mobility of research in ideas space. To do this, we leverage the unanticipated and substantial expansion in access to motion-sensing research technology that occurred as the consequence of the introduction and subsequent hacking of the Microsoft Kinect system. To estimate whether this technology induces mobility in ideas production, we employ novel measures based on machine learning (topic modeling) techniques as well as traditional measures based on bibliometric indicators. Our analysis demonstrates that the Kinect shock increased the production of ideas and induced researchers to pursue ideas more distant to their original trajectories. We find that this holds for both incumbents and entrants in motion-sensing research, with stronger effects among entrants. Importantly, the boost in entrants’ ideas mobility extends to projects outside of motion-sensing, suggesting that the automation of research tasks can act as a conduit towards unexplored knowledge trajectories.

Startups rushing to market with potentially disruptive technologies have the potential to reduce the documented negative effect of private return incentives for scientific discovery (basic research). While it is optimal for startups of disruptive technologies to enter the market to validate their technology in order to enable options for other market transactions, this entry strategy influences industry’s innovation incentives as the revealed technology threatens the incumbents’ trajectory. Furthermore, given the uncertainty of such technologies - ambiguity regarding technology’s robust scientific likelihood to improve to reach its disruptive potential - innovations efforts unfold predominantly in the realm of scientific discovery, leading to an increase in the rate of basic research in the knowledge area of the technology. I provide evidence from efforts to develop and commercialize quantum computing.