Dynamics of scientific evolution: How do luminaries influence the vitality of science?

Knowledge accumulation and the dynamic processes by which new research builds up from prior one are matters of central importance to scientific progress. In the BSE Working Paper (No. 857), “Does Science Advance One Funeral at a Time?” Pierre Azoulay, Christian Fons-Rosen, and Joshua S. Graff Zivin study the way in which distinguished scientists influence the direction of scientific progress in the subfields of life science research in which they are particularly active.

Institutional context and empirical design

Azoulay et al. study the influence eminent scientists may have in the type of research undertaken in their scientific fields. In particular, they examine how the death of 452 star scientists, who passed away while at the peak of their scientific abilities, affected the direction of scientific knowledge.

The basic approach in this article consists in conceptualizing death of star scientists as shocks to the structure of the subfields in which they worked prior to their death. The authors use the passing of “superstar” scientists as a tool to estimate the extent to which this event affected publication rates and funding flows in their fields, decomposing between “insider” (i.e. previous co-authors of the superstar) and “outsiders”.

Effects of superstar extinction

The first set of results presented in the article pertains to the main effect of superstar death on the growth of their subfield, measured both by publication rates and funding flows. The authors find that overall, publication activity slightly increases following the death of a star scientist, but the magnitude of the effect is not large, about 2%.

However, this finding hides an interesting pattern that the authors uncover by differentiating between publications by collaborators and non-collaborators. On one hand, the decline of publication activity of previous collaborators of the star is of the order of 40%. Crucially, they find that publication activity by non-collaborators in the subfield more than offsets the previous reduction in scientific activity. In the case of the funding flows the patterns are very similar to those obtained in the case of publication activity. These results are shown in detail in the table below.

Table: Main Effect of Superstar Extinction

Furthermore, the authors study the dynamic effects of the superstar’s death on the output in the subfield. They find that the effects are increasing over time, and there is no indication that the effects are only transitory. Moreover, five years after the star’s death, the increase in publication activity by non-collaborators is large enough in magnitude to fully offset the decline in activity by collaborators. This finding is graphically represented in the figure below.

Figure: Effect of Star Scientist Death on Subfield Growth and Decline

The mechanisms behind the results: barriers to entry

To examine how important the additional contributions by non-collaborators are, the authors decompose them into different percentiles based on impact factors. The effect is driven strongly driven by high impact contributions by non-collaborators.

The authors also explore the effect the passing of the star has on the intellectual direction of the subfield; they do so by examining the cited references contained in each related article. They find that subfield growth is mostly due to articles from non-collaborators who do not build on the work of the star. While the additional contributions do not represent a radical departure from the subfield’s traditional questions, they draw from newer and different sources of knowledge.

Articles published by non-elite members of the profession account for much of the relative growth. These results indicate that the researchers driving the growth in publication activity following the star’s death are largely outsiders, and that stars might be able to regulate entry into their field while alive.

In order to delve deeper into the dynamics of knowledge accumulation, Azoulay et al. explore some possible barriers to entry:

  • Goliath’s Shadow
    Outsiders are simply deterred by the prospect of challenging a luminary in the field. They find that stars that are important to the field, in terms of research output, appear to be an important deterrent to entry.
  • Intellectual Closure
    Entry into the field, even after it has lost its star, may be deterred if the subfields appear unusually coherent to outsiders. The authors find that when researchers in the subfield agree on the set of questions, approaches and citations, this will tend to dissuade outsiders from entering.
  • Social Closure
    A field might be perceived as socially coherent when the researchers active in it form a tightly knit clique, often collaborating with each other.
  • Incumbent Resource Control
    Stars do not appear especially well positioned to directly block entry through the control of key resources, however it is possible that those resources can be controlled indirectly through the influence on collaborators.  It appears that important coauthors, in their effort to keep the star’s intellectual flame alive, erect barriers to entry into those fields that prevent its rejuvenation by outsiders.


The death of an elite scientist has a negative and seemingly permanent impact on the productivity of their coauthors. In contrast, the productivity of non-collaborators within the same fields appears to increase, at a rate that more than offsets the decline experienced by collaborators.

The increase in productivity and in funding flows appears to come mainly from outsiders who tackle the mainstream questions with newer ideas originated in other domains. In the light of the results presented by Azoulay et al. (2015) one could conclude that the loss of an elite scientist in a field signals those outside that the cost-benefit on putting forward original ideas has changed, thus encouraging them to enter the subfield.

Finally this article has important policy implications. As the presence of a big star scientist in a field appears to inhibit participation from outsiders, policies conducive to ease the access of less established investigators become relevant. Examples of such policies include caps on the mount of funding a single laboratory is eligible to receive, or bonus points for first-time investigators in funding programs.