A Dynamic Network Approach to Breakthrough Innovation

This paper outlines a framework for the study of innovation that treats discoveries as additions to evolving networks. As inventions enter they expand or limit the reach of the ideas they build on by influencing how successive discoveries use those ideas. The approach is grounded in novel measures of the extent to which an innovation amplifies or disrupts the status quo. Those measures index the effects inventions have on subsequent uses of prior discoveries. In so doing, they characterize a theoretically important but elusive feature of innovation. We validate our approach by showing it: (1) discriminates among innovations of similar impact in analyses of U.S. patents; (2) identifies discoveries that amplify and disrupt technology streams in select case studies; (3) implies disruptive patents decrease the use of their predecessors by 60% in difference-in-differences estimation; and, (4) yields novel findings in analyses of patenting at 110 U.S. universities.

💡 Research Summary

The paper introduces a dynamic network framework for studying breakthrough innovation, treating each discovery as a new node that adds links to an existing technological citation network. Traditional metrics—impact (citation counts) and distinctiveness (distance from prior art)—capture either the volume of use or the novelty of the combinatorial structure at the moment of invention, but they ignore how an invention reshapes the future use of earlier technologies. To fill this gap, the authors develop two continuous, time‑varying measures: disruptiveness and radicalness. Disruptiveness quantifies the extent to which a patent, after its issuance, reduces the subsequent citation rate of the prior patents it builds upon; a high disruptiveness score indicates that the new invention supplants or hides its predecessors. Radicalness multiplies disruptiveness by the patent’s own citation impact, highlighting inventions that are both disruptive and widely adopted.

Using U.S. utility patents issued between 1976 and 2010, the authors construct a citation network and compute yearly disruptiveness scores via a sliding‑window approach. They demonstrate that patents with similar raw citation counts can have markedly different disruptiveness, revealing a dimension of innovation quality invisible to standard impact metrics. Three case studies—touch‑screen smartphones, lithium‑ion batteries, and CRISPR gene‑editing—illustrate how high‑disruptiveness patents cause sharp declines in citations to earlier technologies, effectively re‑routing subsequent research.

A difference‑in‑differences analysis provides causal evidence: after a highly disruptive patent appears, the average citation rate to its cited predecessors falls by about 60 %, confirming that disruptive inventions actively diminish the relevance of earlier work.

The framework is then applied to the patent portfolios of 110 U.S. research‑intensive universities. Results show a striking divergence: higher industrial R&D funding correlates with lower radicalness (fewer disruptive patents), whereas greater federal R&D support correlates with higher radicalness. This suggests that industry tends to fund incremental, “amplifying” innovations, while federal programs foster more fundamental, “disruptive” breakthroughs.

Overall, the study offers a novel methodological contribution by embedding innovation assessment in a dynamic network perspective. The disruptiveness and radicalness indices capture second‑order effects—how new inventions alter the usage patterns of prior knowledge—while remaining continuous, dynamic, and valenced. Empirically, the measures discriminate among patents of equal citation impact, validate the disruptive impact through causal analysis, and reveal policy‑relevant patterns in university patenting. The work thus provides scholars, firms, and policymakers with a richer toolkit for evaluating and steering the evolution of technological change.