Most debates about government funding of science and technology revolve around the extent to which the government should support research, usually measured in dollars. Those against government funding will point out crowding out effects and the dangers of government picking winners, while those for it will argue that because scientific discovery produces positive externalities it will be underproduced by a private market. What’s left out of these debates is the effect of how the government chooses to support R&D. Patents have been a standard method for centuries but their distortionary and monopolistic effects work to offset their benefits. Prizes can offer all of the benefits of patents without the costly second order effects.
Properties of Technology
Evaluating the effectiveness of prizes and patents requires understanding the value of what they are trying to promote: technological growth. There are two elements to technological growth: Zero-to-one improvements and one-to-n improvements. Both are required for it to have large effects on human welfare. Zero-to-one improvements are new discoveries that extend our understanding and control of the universe. The Wright Brother’s first flight at Kitty Hawk was a zero-to-one moment. One-to-n improvements take the ideas and prototypes of zero-to-one and commercialize them. The growth of aviation from a scientific marvel to mass-market product was one-to-n. Mass production, vertical integration, and learning by doing lower the costs of new technologies and bring them to a wide audience. One-to-n improvements are the link between technological discovery and the impacts on human welfare that we actually care about.
Another important aspect of technological growth is that ideas are infinitely copy-able. This is unlike all other physical goods. If one wishes to share a pie they made, they must take slices away from their own portion. Once an idea is discovered, it can be spread to every mind without requiring anything to be taken from the originator. Physical goods are divided among people, but ideas are multiplied. This property of ideas implies that we are losing out whenever ideas are not maximally spread. Since the marginal cost of producing new copies of an idea is basically zero, if there is anything to be gained from spreading an idea, it should be done.
Patents vs Prizes
These two properties of technological growth usefully frame the properties of prizes vs patents. Both strategies of government support target zero-to-one technological growth. Prizes do this directly by simply paying for new discoveries, while patents reward inventions with the promise of monopoly profits. Where they begin to differ is in their effects on the one-to-n growth.
The promise of monopoly profits incentivizes individuals and firms to invent. Since invention produces a positive externality, this government incentive pushes the supply of new ideas closer to the social optimum. However, several costs either push against the supply increase promoted by patents or dilute the value of the ideas that are protected by them. Patents reward idea creators by restricting the spread and use of the idea after its creation. Since ideas cost nothing to spread once they are created, restricting the number of people who can use them after conception is a waste of resources. To maximize the social impact of an idea, it should be spread and productized as quickly as possible, but this restriction is required to produce monopoly profits. Patents require resources to enforce. Tens of billions of dollars are spent by inventors and firms each year litigating patent infringement in courts, and the average case costs more than three million dollars. The time and money spent on litigation dissipates the incentive that patents provide for invention. Patents can also harm future innovation by attaching licensing fees to combinations of and incremental improvements to patented ideas. All of invention is building off of previous ideas to an extent. Since patents add costs to iteration and the spread of new ideas/technology, they work against their original goal of subsidizing invention.
Prizes also incentivize invention but in a simpler way. Instead of promising to enforce monopoly profits, prizes simply pay the inventor upfront. For any given patent there is a quantifiable value of monopoly rights which depends on the marginal cost of production and demand for the product. Monopolies can sustainably charge a price above what it costs to make their product (including opportunity cost) because no other firm is allowed to enter the market and undercut them.
A prize could be awarded equal to the value of these monopoly profits. This transfer would have exactly the same effect on the supply of innovations as the patent, but without incurring the costs of monopoly. The rewarded idea could be spread openly, commercialized by anyone, and built upon by other inventors.
Take Elon Musk’s X-prize for carbon capture as an example.
He is offering $100 million dollars to anyone who can demonstrate a carbon capture technique that sequesters a thousand tons of carbon a year at the lowest possible price. This invention incentive could be translated into monopoly rights over a unique carbon capture technology for some number of years. The incentive to invent the technology might be the same, but we would end up removing far less carbon from the atmosphere since the patent holder would have to restrict the quantity produced to get their monopoly profits, thus offsetting the benefit we got from producing this positive externality good in the first place.
Prizes’ Economic Calculation Problem
Although the theory behind prizes is simple, there are numerous calculation problems facing anyone who tries to allocate prizes. First, inventions and their original inventors are often difficult to define. Multiple discovery is surprisingly common, the development of ideas is often incremental, and timelines may not be known. How should a prize for the development of the steam engine be split between Thomas Savery and James Watt? Second, the social value of an invention is difficult to foresee.
This means that the closer a prize is to the inception of a technology, the more likely it is to misestimate its value. Finally, there is the unknown-unknown of the all important future technologies that no one has even thought of yet. It is important to promote the development of these technologies but impossible to set up a specific prize for them in advance.
In the first of these economic calculation problems, prizes have the clear advantage over patents. Both prizes and patents have to decide which party to reward for inventing something, and both have to decide how to handle incremental improvements to already existing technologies. Both strategies lack clear bright-lines for these decisions, but the consequences for getting it wrong are much larger with patents. The example of the steam engine is useful again. The British crown bestowed a patent to Thomas Savery for his steam pump, despite several similar earlier inventions. Savery then got this patent extended to total of 35 years. The patent covered all pumps that raised water by fire and he pursued infringement vigorously. This not only prevented other inventors from improving on his idea, but it also wasted the time of Thomas Savery himself. Instead of pursuing new ideas, he spent his time seeking rents for his existing one. If Savery was simply given a prize for his trouble, he and everyone else could have continued their work and the world would be better for it.
The second and third problems are where patents have an advantage, at least over a traditional prize system. As is often the case in economics, Adam Smith said it best:
Thus the inventor of a new machine or any other invention has the exclusive privilege of making and vending that invention for the space of 14 years by the law of this country, as a reward for his ingenuity, and it is probable that this is as equal an one as could be fallen upon. For if the legislature should appoint pecuniary rewards for the inventors of new machines, etc., they would hardly ever be so precisely proportioned to the merit of the invention as this is. For here, if the invention be good and such as is profitable to mankind, he will probably make a fortune by it; but if it be of no value he also will reap no benefit.
With patents, the central authority need not attempt to calculate the economic value of an invention, because the market will aggregate everyone’s demand for said invention and award the inventor proportionally. Since patents don’t have to choose a goal to incentivize ahead of time, they also promote investment in a wide range of inventions, even those that almost no one has thought of or those that most people think are useless.
What Smith leaves out is that governments still must proportion the reward from patents to the merit of the invention by choosing how long they are protected for. If 14 year patents are good, would 15, 50, or 5 year patents be better? In the US, design patents still get 14 years but the most common type, utility patents, get 20. There are exceptions for pharmaceutical products and the majority of newly issued patents receive some sort of extension over the 20 year mark. Additionally, although patents are more flexible than some targeted prizes, like the Royal Society’s prize for a method of calculating longitude at sea, prize funds can easily be open to an unspecified range of entries like the Macarthur Genius Grants, which have rewarded select researchers from poetry to physics with a million dollars for over 30 years.
So in practice patents are subject to the same economic calculation problems as prizes, and they have more severe consequences for getting it wrong. Now the question becomes how best to allocate budgets between funding scientific prizes and other public goods, and how to allocate prizes among scientists and inventors.
What We’ve Already Tried: Grant System
While patents are the most important lever of government support for commercial innovation, something like a prize system already dominates more basic scientific research.
The National Science Foundation (NSF) and National Institutes of Health (NIH) allocate fifty billion dollars to scientific research every year. This fact is a double edged sword for prizes. On one hand, it supports the main point that restricting the spread of an idea is a bad strategy for promoting progress even though it does subsidize idea creators. If scientists were rewarded with exclusive rights over research topics they discovered, progress in new fields of science would be much slower even though the pecuniary and prestige rewards for new discovery would be larger. The spread of new ideas across many minds and labs is essential to the progress of science, so rewarding discoverers by restricting spread is counter productive.
The other edge of the sword comes from a growing realization that this grant system is not all that good at producing impactful science. High and rising time costs of applications, and bureaucratic incentives for conformity coincide with falling research productivity. Half of all scientific papers were published in the last 12 years, but much less than half of all scientific progress has happened in that time. Simply funding the research of many scientists instead of handing out monopoly rights is not sufficient to effectively fund innovation.
Debates over budget appropriations for these national scientific organizations ignore the huge range in impact per dollar between the most and least effective public goods funding strategies.
In the field of scientific progress this cost-effectiveness range is exemplified best by Fast Grants. With just a few million in funding and fewer than a dozen employees this tiny organization funded some of the highest impact research on Covid-19 while the NIH was revising font sizes and waiting for ethics board review. The bottleneck on our funding of research and technology as public goods is not how much we spend but how we spend. Since patents and grants have been tried and found to have huge problems, what’s the best way to fund science and technology?
Talent Curation: Scalable model?
Most people never invent anything new. A tiny proportion of the population of individuals and organizations produce the vast majority of new inventions and discoveries. Even among scientists, impactful work is power-law distributed. This implies that increasing the production of impactful scientific discovery and technological progress means selecting a group of highly productive individuals and organizations to fund, not funding an industry of incremental discoveries like the NIH and not funding specific inventions like patents. Organizations like Fast Grants, Emergent Ventures, and Y-Combinator have found success with this model, but their scope is small compared to the desired economy wide effect of government supported science. They also rely heavily on tacit knowledge and unilateral decision making from their founders which cannot be easily scaled into an institution.
How can we connect the ‘man on the spot’ knowledge of the world’s best talent curators with the blind, fat pocketbooks of our government? This is a difficult question that I do not have a full answer to. A method of efficiently distributing government money that is immune to strategic play and rent seeking has yet to be found, so I don’t think anyone else has a full answer either.
A first approximation would be to devolve some of the decision making that takes place within federal funding agencies like the NSF and NIH to individual academics. Instead of accepting applications for narrowly defined grants, these organizations would distribute their funding among all STEM professors at R1 universities. The only requirement is that researchers must use a portion of their allocation to fund projects of their peers and students. This would eliminate several layers of middleman from the science funding process, and it would allow brilliant talent curators to use their tacit knowledge and discretion without requiring the approval of a federal committee. The status quo approval requirement drastically changes the research that our top scientists pursue, often pushing them towards narrow, intra-field incrementalism rather than ambitious, multi-disciplinary theory crafting. Keeping the Federal government’s funding but removing their say in what gets researched will allow more speed, ambition, and cooperation. Anyone familiar with the state of academic science today knows that this is dearly needed.
This idea is aimed at reforming academic research, but a similar system could be put in place for firms performing more applied R&D. A Federal agency would hand out open-ended grant money to some subset of firms, requiring only that the money be used for R&D and that some of it must be angel invested into other firms.
Implemented as simply as I’ve described it, this system is probably vulnerable to fraud and rent seeking. Further thought and experimentation is needed. The central point remains that our current mechanisms for funding invention and scientific progress backfire and create counterproductive incentives. Therefore, at the current margin, the way in which we fund research is far more important than how much we spend on it.