Agree on government funding being necessary for R&D. Here’s a toy model which will help prove this concept analytically:
1. Let there be a coin toss game which costs $1 to play, with the probability of heads as 1/n and payout for getting heads $ n+ε
2. The first coin is always beneficial to flip, the second may need at least 2 flips, and to break even on the 10th coin you will ideally want to play the game much longer. In fact, you will find that the capital needed or number of coin flips you want will scale quadratically
3. The “further out” research is, the greater the risk and the payoff. As the timescales increase, you go from small company to large monopoly, to national labs or government program
4. The state being the largest reservoir of capital, it alone can flip enough coins by funding a plethora of R&D projects to make this a reliable investment
Its not obvious to me that "so long as there is unextracted consumer surplus, there will be inventions that would make the world better to create, but are not made" or at least it doesn't track because, while the market is certainly better than government at allocation, it may be thousands of years before it is perfectly efficient because so much data is required. So I think even with perfect price discrimination, I don't think all of these inventions would be created...
The fundamental advantage of patents over prizes lies in avoiding government winner-picking. Under patent systems, inventors receive compensation proportional to the value they actually generate, while useless inventions—which represent the majority—receive no public funding whatsoever.
This market-driven selection mechanism proves far superior to bureaucratic evaluation. Government officials lack both the information and incentives necessary to identify genuinely valuable innovations ex ante. Patents allow the market itself to determine value through actual usage and licensing demand, creating automatic feedback loops that reward genuine utility.
The contrast with government arts funding illustrates this principle perfectly. Public arts agencies consistently fund elitist, derivative productions that generate minimal public interest, while private markets protected by copyright produce entertainment that masses actually enjoy. The difference reflects the superior information-processing capacity of decentralized market signals versus centralized bureaucratic judgment.
Government prize systems force administrators to predict which research directions will prove valuable before results emerge—an impossible task given the inherent uncertainty of innovation. Patent systems eliminate this prediction problem by rewarding success after it demonstrates market value.
Prizes can serve complementary functions alongside patents, particularly for basic research where commercial applications remain distant. However, replacing patents entirely would recreate the fundamental knowledge problems that make central planning ineffective across all economic sectors.
I actually believe certain patents, especially materials science inventions, merit longer protection periods—perhaps 50 years. Materials innovations often require decades of development before commercial applications emerge, and shorter patent terms may inadequately reward the enormous upfront investment required for breakthrough discoveries.
However, I support implementing Harberger taxation for patents to enhance allocative efficiency and reduce patent trolling. Under this system, patent holders would self-assess their patent's value and pay taxes accordingly, while anyone could purchase the patent at the declared price. This mechanism would force patent holders to honestly value their intellectual property while preventing warehousing of unused patents purely for litigation purposes.
The optimal innovation policy combines strong patent protection with market-based mechanisms for efficient allocation, avoiding the twin dangers of inadequate innovation incentives and bureaucratic resource misallocation.
Is there really no literature comparing no patents to yes patents? My prior on this is that patents probably increase innovation, but with diminishing returns on the breadth and/or length thereof; this by analogy to copyright law, where I know more. There we do have literature on the introduction of the system, see Giorcelli and Moser (2020), which finds the expected theoretical effect. But of course, that doesn't imply that the increase from 50 years pma to 70 would do much of anything.
Nicholas, thank you for writing this. You're right that there are no clear answers. I am leaning toward concluding that patents should be severely curtailed, if not abolished entirely. I have explored this topic in several essays that you may enjoy:
Like you, I like this idea, but I don't find it practical. As I noted,
"In my view, the problem with Kremer’s system is that it’s vulnerable to manipulation. Since the government is buying the vast majority of patents at a significant markup, there is the risk that bidders and inventors will collude, driving up patent valuations and forcing the government to pay excessive prices. Kremer recognizes this and suggests that the government step aside on random auctions, creating risk for colluders. Even with safeguards, however, it is too easy for three CEOs to meet on a golf course and arrange a “patent pumping” scheme that would net billions in profit, even if they would take occasional losses."
Grant = "So you'll want to do research to discover X, cool, fill these forms explaining why you need the money and if we like your idea enough we'll give you the money you need. Otherwise, you'll have wasted maybe a dozen hours of work filling the paperwork (but you can submit the same paperwork next funding cycle)"
Prize = "Whoever discovers X first gets a prize. If you sank millions but someone beated you to it, tough luck"
I hope it's clear why grants are more common. In practice prizes work only when one unambigously define what success looks like, and even there very few entities are willing to sink millions hoping to maybe get a result. Prizes obviously do not enable coordination, the incentive for researchers is to keep the results secret until they are sure they can claim the prize, thus very much hindering the first-order benefits of research, creating publicly available knowledge.
Delightful article for a Thursday evening. Truly.
Agree on government funding being necessary for R&D. Here’s a toy model which will help prove this concept analytically:
1. Let there be a coin toss game which costs $1 to play, with the probability of heads as 1/n and payout for getting heads $ n+ε
2. The first coin is always beneficial to flip, the second may need at least 2 flips, and to break even on the 10th coin you will ideally want to play the game much longer. In fact, you will find that the capital needed or number of coin flips you want will scale quadratically
3. The “further out” research is, the greater the risk and the payoff. As the timescales increase, you go from small company to large monopoly, to national labs or government program
4. The state being the largest reservoir of capital, it alone can flip enough coins by funding a plethora of R&D projects to make this a reliable investment
Its not obvious to me that "so long as there is unextracted consumer surplus, there will be inventions that would make the world better to create, but are not made" or at least it doesn't track because, while the market is certainly better than government at allocation, it may be thousands of years before it is perfectly efficient because so much data is required. So I think even with perfect price discrimination, I don't think all of these inventions would be created...
On Patents, Prizes, and Innovation Incentives
The fundamental advantage of patents over prizes lies in avoiding government winner-picking. Under patent systems, inventors receive compensation proportional to the value they actually generate, while useless inventions—which represent the majority—receive no public funding whatsoever.
This market-driven selection mechanism proves far superior to bureaucratic evaluation. Government officials lack both the information and incentives necessary to identify genuinely valuable innovations ex ante. Patents allow the market itself to determine value through actual usage and licensing demand, creating automatic feedback loops that reward genuine utility.
The contrast with government arts funding illustrates this principle perfectly. Public arts agencies consistently fund elitist, derivative productions that generate minimal public interest, while private markets protected by copyright produce entertainment that masses actually enjoy. The difference reflects the superior information-processing capacity of decentralized market signals versus centralized bureaucratic judgment.
Government prize systems force administrators to predict which research directions will prove valuable before results emerge—an impossible task given the inherent uncertainty of innovation. Patent systems eliminate this prediction problem by rewarding success after it demonstrates market value.
Prizes can serve complementary functions alongside patents, particularly for basic research where commercial applications remain distant. However, replacing patents entirely would recreate the fundamental knowledge problems that make central planning ineffective across all economic sectors.
I actually believe certain patents, especially materials science inventions, merit longer protection periods—perhaps 50 years. Materials innovations often require decades of development before commercial applications emerge, and shorter patent terms may inadequately reward the enormous upfront investment required for breakthrough discoveries.
However, I support implementing Harberger taxation for patents to enhance allocative efficiency and reduce patent trolling. Under this system, patent holders would self-assess their patent's value and pay taxes accordingly, while anyone could purchase the patent at the declared price. This mechanism would force patent holders to honestly value their intellectual property while preventing warehousing of unused patents purely for litigation purposes.
The optimal innovation policy combines strong patent protection with market-based mechanisms for efficient allocation, avoiding the twin dangers of inadequate innovation incentives and bureaucratic resource misallocation.
Is there really no literature comparing no patents to yes patents? My prior on this is that patents probably increase innovation, but with diminishing returns on the breadth and/or length thereof; this by analogy to copyright law, where I know more. There we do have literature on the introduction of the system, see Giorcelli and Moser (2020), which finds the expected theoretical effect. But of course, that doesn't imply that the increase from 50 years pma to 70 would do much of anything.
Nicholas, thank you for writing this. You're right that there are no clear answers. I am leaning toward concluding that patents should be severely curtailed, if not abolished entirely. I have explored this topic in several essays that you may enjoy:
I discuss many of the drawbacks and problems with patents, which you undoubtedly already know here: https://risknprogress.substack.com/p/patent-failure?r=8frpw
Here, I explored the relative merits of patents and prizes. What's interesting is that, when you dive into this, prizes aren't really all they are cracked up to be: https://risknprogress.substack.com/p/patents-or-prizes?r=8frpw
I discussed Kremers' patent buyout system and suggested some alternatives here: https://risknprogress.substack.com/p/the-ideas-anticommons?r=8frpw
Like you, I like this idea, but I don't find it practical. As I noted,
"In my view, the problem with Kremer’s system is that it’s vulnerable to manipulation. Since the government is buying the vast majority of patents at a significant markup, there is the risk that bidders and inventors will collude, driving up patent valuations and forcing the government to pay excessive prices. Kremer recognizes this and suggests that the government step aside on random auctions, creating risk for colluders. Even with safeguards, however, it is too easy for three CEOs to meet on a golf course and arrange a “patent pumping” scheme that would net billions in profit, even if they would take occasional losses."
What are the differences between prizes and research grants? Why do grants seem to be so much more common than prizes?
Grant = "So you'll want to do research to discover X, cool, fill these forms explaining why you need the money and if we like your idea enough we'll give you the money you need. Otherwise, you'll have wasted maybe a dozen hours of work filling the paperwork (but you can submit the same paperwork next funding cycle)"
Prize = "Whoever discovers X first gets a prize. If you sank millions but someone beated you to it, tough luck"
I hope it's clear why grants are more common. In practice prizes work only when one unambigously define what success looks like, and even there very few entities are willing to sink millions hoping to maybe get a result. Prizes obviously do not enable coordination, the incentive for researchers is to keep the results secret until they are sure they can claim the prize, thus very much hindering the first-order benefits of research, creating publicly available knowledge.
In practise, patents are a scam that takes wealth away from ordinary people and gives it to big corporations.
there's more to life than sleeping in and getting high with you