Two things seem to be true: more science leads to more technological progress, but only a minority of new technologies directly rely on science. So what kinds of technology benefit most from a scientific foundation to draw on?
One way to conceptualize the difference between science and technology, is that scientific knowledge tells us about how the world works, while technology is about capturing and orchestrating regularities in nature to do something we think is useful. But by definition, a new technology has to step into the unknown and try something new - either relying on novel natural processes, or orchestrating existing ones in novel ways. Since science gives us knowledge about how natural processes work and interact, it can provide an imperfect map of this unknown terrain, helping inventors step wisely. We can see this is indeed the case with a set of papers, each of which looks to patents as a measure of invention, but which take varied approaches to measuring the “reliance on science” and the degree to which the terrain an inventor is exploring is “unknown.”
To start with a quite intuitive application, science is useful for inventors who are changing fields. Arts and Fleming (2018) looks at inventors who initially patent in one technological domain, and then switch to another. Looking at US patents over 1975-2002, they find the patents of people who are new to a domain seem to be less valuable, as compared to the patents of people with more experience in the domain. Here, they are measuring “value” by how many citations the patent receives, or how likely it is to be renewed (you have to pay non-negligible fees to keep patents active, so renewal is a common proxy for the value of a patent - it suggests the patent-holder thinks the patent is worth the fees).
Using science helps inventors avoid this penalty when they enter a new field. Arts and Fleming use the citation of scientific articles in a patent to measure reliance on science. In general, patents that cite science receive more citations and are more likely to be renewed. But this effect is especially strong for the patents of people changing fields. Compared to experienced inventors in the same field, people for whom the field is an unfamiliar one seem to disproportionately benefit from having science there to guide them while inventing.
(As an aside, Arts and Fleming also find these patents by outsiders seem to exhibit more creativity, as measured by the number of unusual combinations of technology they make. A finding to dig into some other time)
Science also seems to help navigate technological domains that are very “far” from any current work. Testing that notion requires a way to measure when a technology is “near” or “far” from technologies that already exist. Kneeland, Schilling, and Aharonson (2020) do this by relying on a set of about 10,000 technology classifications used by the US Patent and Trademark Office to help patent examiners search for relevant prior art. Basically, the patent office slaps labels on patents for every major kind of technology present in a patent - whether those are for lithium batteries, sewing machines, or neural networks. Kneeland, Schilling, and Aharonson measure the distance of a patent from another by the number of these classifications that differ (most patents are assigned more than one such classification). For example, consider the following three patents, assigned technology classifications A, B, C, D, and E.
Patent #1: ABC
Patent #2: ABD
Patent #3: CDE
In this example, Patent #1 and Patent #2 are one step away from each other, because you just need to make one change to turn one into the other (change “C” to “D” or vice-versa). Patent #3 is farther away, since you would need to change at least two classifications in patent #1 or #2 to go from either of them to patent #3. Kneeland, Schilling, and Aharonson focus their attention on the 8% of patents that differ by two or more steps from any other patents granted at the time of filing.
Kneeland, Schilling, and Aharonson (2020) find two different measures of reliance on science are correlated with being one of these outlier patents. Patents that cite more scientific articles are more likely to be outliers. So are the patents invented by people with a university affiliation. In other words, patents for inventions building on science (measured either by citations to science or a university affiliation) are more likely to occupy more unexplored corners of the technological landscape.
It’s not only unexplored regions of the technological landscape that benefit from science though. There may also be regions that are well-trod, but treacherous. In technologies that orchestrate a set of components interacting in complex or non-linear ways, small changes to the design of a technology can have large impacts on whether the technology is useful or not. In these realms of very “fussy” technologies, which need to be just right to work, science can usefully guide inventors to understand what can be changed, and in what way, and what cannot be changed.
Measuring this is not easy, but Fleming and Sorenson (2004) gives it a shot. Like Kneeland, Schilling, and Aharonson (2020), this paper also relies on the technology classifications assigned to patents (though not the exact same ones). They attempt to create a measure of how “fussy” each technological classification is based on how well it seems to play nice with other technologies (fussy is my term, not their’s, but I think it captures what they’re going for in a colloquial way). If a technological classification is frequently attached to a patent alongside a wide range of other classifications, they’re going to say this isn’t a very “fussy” technology. It can be used in plenty of diverse applications. On the other extreme, if a classification is only ever assigned to a patent with one other classification, then we’re going to assume the technology is very sensitive and very fussy. It only works well in a very specific context.
While this measure is a bit ad-hoc, Fleming and Sorenson also do a survey of inventors and they show this measure is correlated with inventors self-assessments of how sensitive their own inventions are to small changes, and that this measure is not merely picking up how novel or new the technology is; it’s picking up something a bit different. Their main result is to show that patents primarily composed of these very sensitive/fussy technologies seem to disproportionately benefit from science (again measured here by citation of a scientific article).
As you can see above, patents that cite science always tend to receive more citations (from other patents) than those that don’t. But the gap between the citations received by patents that cite science and those that don’t is wider for patents that rely on these more fussy technologies.
So, across three different settings, inventors venturing out into unfamiliar terrain - either because it’s new to them, new to everyone, or very treacherous - benefit disproportionately from a scientific map of the region. But one additional paper provides further complementary evidence from a different vantage point.
All of the papers considered so far are about inventors using science to help them find valuable ideas. Among other things, Arora, Belenzon, and Suh (2021) instead look at how science helps people assess the value of patented ideas. They specifically look at patent reassignments, which occur, for example, when an inventor sells their patent to a firm. For these kinds of transactions to occur, obviously the buyer must be able to form some idea about the value of the patented invention. So, in principle, this can provide another line of evidence for how science helps navigate unknown terrain - in this case, by putting a dollar value on a newly patented invention. One thing Arora, Belenzon, and Suh do is show that these trades are more likely to occur when the patent is based on science.
They measure a patent’s links to science in two ways. The first is familiar: does the patent cite a scientific article? But the second is a nice new complementary measure. They use natural language processing to measure the similarity of the patent’s text to the text of pre-existing academic paper abstracts- the more the language in a patent resembles the language in science articles, the more likely it is to build on scientific ideas. That helps them identify patents that are heavily influenced by science, but which do not cite science (perhaps because they instead cite patents that, in turn, cite science), as well as patents that cite science but don’t actually seem to use any of the ideas. In either case, they find patents that are more reliant on science are more likely to be traded.
So scientific patents are more likely to be bought and sold. Is that because links to science reduce the uncertainty about the quality of the invention for the buyer? For example, suppose a team of inventors relies on science to create a primitive quantum computer (and they get a patent). They can show it really is a quantum computer, but it can’t actually do anything very useful yet. But a buyer who understands the theory might quantum computing technology has enormous future applications and be willing to spend a lot to acquire the patent.
On the other hand, maybe science just helps inventors create new technologies that are demonstrably better than those not reliant on science? For example, suppose a team of inventors uses science to create a sophisticated quantum computer. They show investors that it can do amazing and useful things right now. In this case, buyers will be interested, even if they don’t understand the science of how it works. They just see that the technology works and does new and useful things.
But no. While it’s true that patents reliant on science seem to be more valuable, Arora, Belenzon, and Suh show that scientific patents are still more likely to be traded, even after you account for the value of the patent in various ways. So, if you have two patents that seem to be equally valuable (by some imperfect proxy), the one based on science is still more likely to be traded than its peer.
Arora, Belenzon, and Suh also find some results that echo what we’ve seen above. Just as Arts and Fleming found the patents of experienced inventors entering an unfamiliar field disproportionately benefited from a scientific “map” of the region, Arora, Belenzon, and Suh find technological domains where it is more common to cite science are also more likely to have a greater share of inexperienced inventors (i.e., those who have not previously held patents).
Like Kneeland, Schilling, and Aharonson, they also find evidence that science is more useful in domains that are unfamiliar to everyone. Patents that cite newer scientific articles are more likely to be traded. And the impact of science is also stronger for patents that are more novel - that is, more distant from what has come before. Here, novelty is measured in a clever new way, as the dissimilarity of the patent text from the text of existing patents. Patents that are more unusual (their text doesn’t look like the text of patents already out there) get an extra “boost” from citing science, in terms of their probability of being traded. Similarly, patents that are given unusual combinations of technology classifications by the patent office also see a disproportionate increase in trade if they cite science.
So science accelerates technological progress in a few ways. First, it helps inventors find their way in unfamiliar places. But second, it also helps scientists convince others that what they have found is valuable. That contribution shouldn’t be underestimated either - new technologies can’t have much of an impact on society if they aren’t made widely available, and it is often not the inventor who has the capability of making that happen.
New articles and updates to existing articles are typically added every two weeks. To learn what’s new on New Things Under the Sun, subscribe to the newsletter.
Arts, Sam, and Lee Fleming. 2018. Paradise of Novelty - or Loss of Human Capital? Exploring New Fields and Inventive Output. Organization Science 29(6): 1074-1092. https://doi.org/10.1287/orsc.2018.1216
Kneeland, Madeline K., Melissa A. Schilling, and Barak S. Aharonson. 2020. Exploring Uncharted Territory: Knowledge Search Processes in the Origination of Outlier Innovation. Organization Science 31(3): 535-557. https://doi.org/10.1287/orsc.2019.1328
Fleming, Lee, and Olav Sorenson. 2004. Science as a map in technological search. Strategic Management Journal 25(8-9): 909-928. https://doi.org/10.1002/smj.384
Arora, Ashish, Sharon Belenzon, and Jungkyu Suh. 2021. Science and the Market for Technology. NBER Working Paper 28534. https://doi.org/10.3386/w28534