What inspired you to explore the relationship between industry publishing and scientific novelty?
We often think of scientific research as something that happens primarily within universities and public research institutions. However, industry has a long history of contributing to science. Companies such as IBM and Bell Labs have produced influential scientific publications for decades, and more recently, firms like Google, Amazon, and OpenAI regularly publish in leading academic journals and conferences.
At the same time, firms do not participate in scientific publishing for the same reasons as universities. Their research agendas are often shaped by commercial objectives. Firms may even influence the direction of academic research. For example, studies have shown that industries such as alcohol or tobacco have funded research in ways that shift attention toward certain explanations or interventions while downplaying others.
This made me interested in a broader question: when firms become more active participants in scientific publishing, does this change the kind of knowledge that gets produced? Since novelty is central to scientific progress, I wanted to examine whether stronger industry presence in a scientific field is associated with different patterns of novelty.
In simple terms, how would you explain “novelty in science” to a general audience?
Novelty in science is difficult to capture completely. No single conceptualization or operationalization can fully account for all forms of novelty, because newness can emerge in many different ways. However, one useful way to understand novelty is through the metaphor of LEGO bricks. Think of scientific knowledge as a giant set of LEGO bricks. Each brick represents an existing idea, concept, method, or finding. Novelty happens when a scientist takes bricks that have not usually been connected before and snaps them together to build something new.
For example, multiplication can be understood as building on the idea of repeated addition. In a more advanced sense, many scientific breakthroughs happen when disparate ideas are brought together. Similarly, Prospect Theory emerged by combining insights from psychology and economics to challenge traditional assumptions about how people make decisions.
So, novelty is not always about inventing something entirely from nothing. More often, it is about making a cognitive leap: seeing a connection that others have not seen before and using old ideas in a new combination.
How does industry involvement influence the way new scientific ideas are generated and combined?
So, the core argument is that greater industry publishing does not simply increase or decrease novelty in a uniform way. Rather, it may relate to different kinds of novelty differently because of different underlying mechanisms. First, industry participation can increase the idea pool within a scientific field. Because industry research is often shaped by practical, commercial, and technical problems, it can introduce new kinds of questions, constraints, data, methods, and application-oriented insights into scientific conversations. These become additional “LEGO bricks” that researchers can combine with existing academic knowledge.
Second, the problems introduced by industry often require interdisciplinary research. Many industry-relevant problems are complex and cannot be solved from within a single knowledge domain. This can encourage researchers to draw on a wider range of theories, methods, and fields, thereby increasing the scope of novel recombinations, or what I call novelty breadth. In this sense, industry involvement may push scientists to combine more diverse knowledge elements than they otherwise would.
However, industry involvement can also constrain novelty. Greater industry publishing may be associated with stronger commercial influence within a field, which can encourage secrecy, delays in publication, or the movement of important knowledge disclosure from publications to patents. These dynamics may limit the free flow of knowledge and reduce the diversity of ideas available for recombination.
At the same time, a stronger focus on industry-relevant problems may steer academic research away from more exploratory, curiosity-driven, blue-sky research toward safer, more immediately useful directions rather than risky, curiosity-driven combinations of intellectually distant ideas. Industry influence can also narrow research agendas, encourage secrecy, delay publication, or shift disclosure from publications to patents. Together, these mechanisms may reduce the novelty distance of scientific publications within a field.
Your study highlights concepts like novelty breadth and novelty distance. Could you explain these?
I use three related but distinct measures of novelty in my study: novelty occurrence, novelty breadth, and novelty distance. Novelty occurrence captures whether a publication contains any novel combination of prior knowledge at all. In simple terms, it asks: does this paper connect ideas that are not usually connected? It is like asking whether someone at a dinner party introduced a new conversation by bringing together people or topics that normally do not interact.
Novelty Breadth captures the overall scope of novel recombination in a paper. A paper has greater novelty breadth when it introduces multiple new combinations, particularly when those combinations connect ideas that are relatively far apart. Using the dinner party metaphor, novelty breadth is like how many unusual conversations are happening across the room, while also considering how different the people in those conversations are. A party where a chef talks to a musician and an economist talks to a climate scientist has greater breadth than one with only a single unexpected interaction. And the breadth becomes even greater when the conversations connect people from very distant worlds — like a medieval historian and a rocket scientist.
Novelty distance captures how far apart the combined ideas are intellectually. Talking to a chef about food and nutrition may involve related ideas, so the distance is relatively small. But connecting ideas from medieval history and rocket science would involve a much larger intellectual jump. So, novelty distance is about how unfamiliar or distant the ideas were before they were brought together.
In short, novelty occurrence asks whether a novel combination happened, novelty breadth asks how wide the combination is, and novelty distance asks how far apart the combined ideas are. Together, these measures allow us to capture not just whether novelty occurs, but also what kind of novelty is being produced.
What was the most surprising or impactful finding from your large-scale analysis of scientific publications?
One of the most surprising findings from the large-scale analysis was that greater industry publishing does not simply increase or decrease scientific novelty overall. Instead, it appears to shape different kinds of novelty differently.
Specifically, fields with greater industry publishing activity, on average, showed higher levels of novelty occurrence and novelty breadth, but lower levels of novelty distance. In other words, greater industry involvement in scientific publishing may encourage more novel recombinations, particularly those involving a broader scope of ideas and interdisciplinary connections, while simultaneously discouraging riskier, blue-sky combinations across intellectually distant ideas.
This is particularly impactful because it adds on to the conversation that novelty is not captured through a single concept or measure. The broader implication is that understanding scientific creativity requires going beyond simply asking whether research is “novel.” We also need to ask what kind of novelty is being produced.
How can your findings help researchers, institutions, or policymakers make better decisions about research collaboration and funding?
The findings have important implications for researchers, institutions, and policymakers designing research collaboration and funding strategies. The key takeaway is that greater industry publishing does not influence all forms of novelty in the same way. While greater industry publishing may encourage more novel recombinations and broader knowledge integration useful for complex problem-solving, it may also reduce more distant, exploratory, blue-sky recombinations. So, policymakers need to think beyond a single measure of novelty and pay attention to different dimensions of novelty. This means research funding systems may need to balance two goals simultaneously: encouraging industry engagement and preserving curiosity-driven science. If policymakers want fields with strong industry presence to also produce more exploratory breakthroughs, complementary mechanisms may be needed, such as long-term funding for basic research, institutional support for high-risk projects, or protections from short-term commercial pressures.
Another important finding is that top–ranked research institutions are relatively better positioned to retain the benefits of industry publishing while still supporting more exploratory forms of novelty. This suggests that policymakers may consider institutional research ranking when promoting university–industry engagement, while also strengthening the broader research capacity of other institutions to support long-horizon, curiosity-driven research.
