9 ms·

I never thought that the case for patents was to increase productivity. I always thought that their purpose was to eliminate trade secrets: people are going to

I never thought that the case for patents was to increase productivity. I always thought that their purpose was to eliminate trade secrets: people are going to invent stuff whether you want to or not, but without patents, they'll take the secrets to the grave. My favourite patent of all time is Michael Jackson's smooth criminal patent: https://mashable.com/2015/03/28/michael-jackson-shoe-patent/#AdzmmRRZPmqY https://mashable.com/2015/03/28/michael-jackson-shoe-patent/... This is the patent system working as intended. A magician revealing his tricks. Tell us how you did it, and in exchange we promise to not compete with you for a while. The patent system could work if patent clerks had the time to be a lot more discerning and reject any patent that is obvious or already known. But it takes a lot of work to do that and the deluge of patents makes it impossible to give each application sufficient consideration. I think the general idea of patents is good, but perhaps it's impossible to execute it correctly. Edit: Should have read the damn article. They address concern further below and also argue that patents are not revealing any secrets because pattent attorneys are not qualified to judge the merit of patents and because the way patents are written makes it almost impossible to reproduce their results, and sometimes patents are granted for utter nonsense, such as moving "through the fifth dimension".

enriquto 8y ago
> I think the general idea of patents is good Also, there may be a fundamental difference between physical inventions, where patents seem to make sense, and mathematical theorems (and computer algorithms), where patents do not make sense at all.
- api 8y ago
  The difference is between a specific invention vs. a law of mathematics or nature. The latter absolutely should not be patentable. The negative opinion of patents around here comes from the fact that most people here are programmers and the vast majority of software patents are trash. I've seen very few patents on software that are not "obvious to anyone skilled in the art," patents on abstract mathematical concepts, or patents for which there is ample prior art in the literature or open source code repositories. Telling the difference requires a great deal of conceptual thinking. Patent examining is "strong AI hard." I think it is possible to expand on these ideas to some useful extent. Obviousness for example could be explained as first-order connectedness is the evolutionary/learning state-space graph. Something is obvious if getting from current art to the new idea requires only one "hop" through state space or traversal of a densely connected region of state space. Non-obvious means some kind of great leap across state space or long journey through many indirections was required. It's basically a measure of creative labor. The RSA patent is often cited as an example of a decent software patent: original, a specific well-bounded invention for a specific purpose, and non-obvious in the extreme. Another example would be if whomever Satoshi is had patented proof of work block chains since that would also meet those criteria. Of course patents on cryptographic systems tend to result in them not being used, but that's a separate issue.
  - enriquto 8y ago
    > The difference is between a specific invention vs. a law of mathematics or nature. The latter absolutely should not be patentable. I do not understand the difference between the two. As a mathematician with an interest in philosophy, I would say that about 50% of mathematicians would agree that algorithms and theorems are discovered, not invented. Thus, your example of RSA as a "good" patent sounds preposterous to me. The RSA algorithm was discovered, and it is a law of nature as much as the Euclidean algorithm to find the gcd. I find it honestly absurd that it can be patented (not angry, merely surprised).
    - rayiner 8y ago
      That's a bit of reductio ad absurdum. The RSA algorithm is a "law of [mathematical] nature" as much as a log cabin is natural. The composition of natural building blocks into something that performs a function is a work of human ingenuity. That said, RSA itself probably would not be patentable today. The Supreme Court's decision in Alice probably precludes pure algorithm patents. You need an algorithm in the context of a specific use case. For example, a cell phone that uses RSA to encrypt keys to protect data in flash memory.
      - pnw_hazor 8y ago
        A modern RSA patent application would get past Alice. It is a protocol that could be described/claimed in the context of communication between computers. Where the protocol improves the performance of a computers or communications because X,Y, or Z and improves the security of the session.
snarf21 8y ago
I agree. I think that all patents aren't created equal. We'd have less issues if software patents were only valid for like 5 years. You still have protection but you get rid of a lot of the trolls and the IBMs of the world having 10K patents. For more physical things, the protection can be longer since it takes longer to recoup your investment.
- wolfgke 8y ago
  > I think that all patents aren't created equal. We'd have less issues if software patents were only valid for like 5 years. And we'd have even less issues if software patents were only valid for 0 years (i.e. not exist).
  - tracker1 8y ago
    I would rather they front load the cost for the application fee, reduce the actual license fee. And then hire reviewers with a mindset to reject if it doesn't clearly describe how to recreate a process properly, or isn't unique or non-obvious. I think another extension to the standard for patent-ability should be does this patent make something less effective (as is often the case with extension patents).
- jordigh 8y ago
  There often isn't even any "investment" in patents. Richard Feynmann would describe how his superiors would just walk around the labs and say, so! who has some ideas we could patent? And Feynmann's fellow scientists would say, well, you could grab this nuclear reactor and stick it into a submarine to power it, but that's pretty dumb and obvious. "Nonsense!" replied the superiors. "Patent that right away!" I don't know how true it is that a lot of patents come just from employees sitting around navel-gazing in their free time and aggressive management telling them to write down those ideas so they can file more patents, but it seems believable that it would happen at least sometimes.
  - snarf21 8y ago
    Yeah, that seems like the IBM (and others approach). The problem with 0 software patents is that any small player no longer has a window to be a first mover. Google and FB have thousands of engineers sitting around and if they could just duplicate the new idea without consequence there would be no reason to do their own research and no relief for the little guy.
rayiner 8y ago
Suggesting that they’re “impossible to execute” correctly is a bit odd, as all US technology arose in the context of the patent system (the first Patent Act predates the Constitution).
- derf_ 8y ago
  You haven't given any evidence that it arose because of that system or in spite of it. There are plenty of examples going back to Singer sewing machines [1] and the Wright Brothers [2] that demonstrate that the surest way to delay the widespread deployment of any transformative technology by many years is to patent it. [1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1354849 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1354849 [2] https://en.wikipedia.org/wiki/Wright_brothers_patent_war https://en.wikipedia.org/wiki/Wright_brothers_patent_war
  - rayiner 8y ago
    You can also find dozens of examples where investors pour billions of dollars into developing technology because they know they can patent it and won’t immediately be undercut by free riders. I’m typing this on a machine that has CPU technology licensed from ARM, GPU technology licensed from Imagination Technologies, and cellular technology licensed from dozens of companies. Those companies did, in fact, invest in all of those technologies in reliance on patent protection for the resulting developments. One can imagine alternative ways to monetize that R&D. Unfortunately, the main alternative historically was proprietary products and vertical integration. But one can imagine there is something better.
    - derf_ 8y ago
      The claim wasn't that the system is "unworkable" in the sense that it's causing world hunger or something, but that the system doesn't "work" in the sense that, in aggregate, it's not "promoting progress of the useful arts." People will certainly use patent protection where it is available to them, but that doesn't mean they wouldn't have invested in those technologies solely on the basis of first mover advantage [0], and the lack of any observed effect of strengthening the patent system and increasing the scope of what is patentable on R&D spending tends to back that up. [0] ARM would need a different business model, but it's unclear to me that other business models (i.e., Intel's) wouldn't have been just as successful, and wouldn't come with the social deadweight loss of patent trolls highlighted in the article.
renox 8y ago
> I always thought that their purpose was to eliminate trade secrets: people are going to invent stuff whether you want to or not, but without patents, they'll take the secrets to the grave. Except that if you look at the history of inventions you'll notice that quite frequently several people invented the same thing independently at almost the same time (the phone for example). So the risk of losing a 'trade secrets' isn't that high..
- CalChris 8y ago
  Two people can hold the same (independently discovered) trade secret. If I slave away and re-discover the Coke recipe, I can keep it and use it as a trade secret. But if I intern at Coke and steal it, I can go to jail for trade secret theft. A patent incents me to slave away and then requires me to teach the patent in exchange for a limited monopoly. Trade secret law only punishes me for stealing but incents me to keep secrets. The trouble with trade secrets is that I have to be able to exploit my invention and still keep it a secret. I probably can't do that with the phone. So I wouldn't slave away working on that problem if the best protection I could get is a trade secret.
matz1 8y ago
Trade secret is becoming less and less issue these days. Nowdays information spread and leak like crazy.
- DannyB2 8y ago
  And can be disassembled, decompiled and studied.
- rhino369 8y ago
  In some ways they are more powerful than ever since you can hide behind an API.
- dnautics 8y ago
  It's actually better in many cases. For processes, in order to enforce the patent, you have to get a warrant (how can you prove from the product that the process was used?). A judge will generally refuse to grant a warrant to search a competitors premises because that would become an open door for anyone to disrupt their competitor's work.
wtdo 8y ago
Not that this would be an efficient use of anyone's time but... Assuming the purpose of patents is 1) to eliminate trade secrets, 2) that patents must be novel, and 3) that patents must be non-obvious to someone skilled in the art, then perhaps a solution would be: take a handful of people skilled in the art, tell half of them to implement the patent and tell the other half only the end goal and see if any of them come up with the same process. The first group would help to see if the patent is written well enough. If they can't implement it then your patent is automatically rejected. If anyone from the second group comes up with the same process, then that's an automatic patent rejection. Even if they don't, it's still helpful information for the patent examiner. For example, if they can solve the problem, just not in the same way as the patent application proposes, perhaps the patent examiner will see that the patent isn't necessary. Go ahead and have your trade secret, the public can obviously solve that problem with or without that knowledge.
- jpadkins 8y ago
  the costs of your system are probably way more than the value created for society via patents.
  - AmIFirstToThink 8y ago
    The cost of creating artificial exclusivity, enforced by the state against natural human learning/discovering capabilities, is even higher.
    - pnw_hazor 8y ago
      Patents cover stuff that is already invented and put into public use. How does that go against natural learning/discovery? Patents may stop someone from on purpose or accidentally copying someone else invention, but that isn't stopping discovery.
      - AnthonyMouse 8y ago
        People spend resources trying to discover things because they want to use them. If it turns out that when you go to use your invention, a patent troll jumps out from under the bridge to shake you down, people will be less inclined to spend resources trying to invent new things to use and instead just stick to the status quo.
philipkglass 8y ago
I mostly read patents about chemistry, materials, and electronics. I find a lot of older patents to be straightforward and informative. Occasionally I will run across a 21st century patent that is as readable, but they're a lot rarer now. If I had to estimate when the transition happened, I'd guess around the 1980s, maybe late 1970s. Weirdly, this is also about the time that academic publications appeared to decline in information-per-page and reproducibility-per-procedure. The societal disclosure-value of the patent seems to be dropping off over time, both because reverse engineering has improved and because the body of a typical patent is becoming less informative about how to construct a useful embodiment of the patented invention.
- marcosdumay 8y ago
  At the 70's - 80's our hyperoptimization cycle started. Before that society was focusing on expensive growth, since there was a lot of "empty space" created by the wars and recent science progress.
- gowld 8y ago
  That's because the PTO is granting garbage patents, even in the case of legitimately patentable inventions. It a main problem with patents today (along with the lifetime of a patent which was originally chosen during a much slower time for humanity's intellectual advancement).
  - philipkglass 8y ago
    You inspired me to look for US patent grants by year, which brought me to this table: https://www.uspto.gov/web/offices/ac/ido/oeip/taf/us_stat.htm https://www.uspto.gov/web/offices/ac/ido/oeip/taf/us_stat.ht... In 1963 the USPTO granted patents to 54% of applications. In 2015 the rate was actually slightly lower (52%). But as "The Case Against Patents" notes, the total volume of applications has soared. I kind of expected to see some sort of upward inflection around 1980 (Bayh-Dole Act) but didn't spot one by eye. The growth rate just in this century is astonishing; there were "only" 288,811 applications in 1999. Total applications per year were below 500,000 as recently as 2009. They're now well over 600,000/year.
jandrewrogers 8y ago
The decline of practical enforceability in some patent areas has created adverse incentives with respect to how you use the patent system, leading to much of the mess you see today. As a practical consequence, less R&D is being patented since disclosure is not adequately protected, and patents are instead used strategically to increase the cost of reverse engineering a trade secret. This leads to the large number of patents designed with enough specificity to deny competitors access to technology without enough critical implementation detail to replicate the technology from said patent. Even trade secrets often only buy you a small amount of time these days before a competitor reverse engineers your technology, so increasingly elaborate (and often futile) measures are being taken to extract an advantage from R&D that justifies the investment in the first place before your competitors use it against you. Especially in computer science, as much as many people complain about algorithm patents, an enormous amount of state-of-the-art research is being solely done as trade secrets now and will never be patented or published. I suspect many people that strongly favor banning computer science patents would not be happy with the reality that this drives advanced computer science research to never be published at all. There is no longer a viable path for recovering investment in computer science research that involves publishing it, which should be cause for concern. (People that want to monitor the advancement of AI tech should find this reality particularly distressing, as it severely hampers their efforts.)
- philipkglass 8y ago
  "Even trade secrets often only buy you a small amount of time these days before a competitor reverse engineers your technology" -- this tends to undermine the idea that good algorithms would remain secret without software patents. Someone else suggested that you can hide some kinds of algorithms if you only provide software-as-a-service. That way there's never even a binary available for reverse engineering. The same applies for software that is used just internally, like if an aircraft manufacturer has their own geometric modeling kernel that aircraft buyers never get near. But even with NDAs for employees and no external distribution of software, good ideas leak as people change jobs and find inspiration from systems they built/used before. Sometimes the main thing you need to know is that a particular line of research was fruitful; exactly reproducing the original implementation is less important. I tend to think that it would usually take less than the 20 years of patent exclusivity for a good, intended-to-remain-secret algorithm to diffuse out of a for-profit company, or to be independently invented again by somebody who will publish regardless of patentability. Pondering a world without software patents, is it worth it to learn about a good idea 10 years "late" (assuming secrecy is attempted), in exchange for being able to freely use that idea 10 years "early"? I say with complete confidence: maybe.
thayne 8y ago
But the reason to eliminate trade secrets is to increase innovation, because other innovators can build on the released information to make new innovations. But in practice, at least in the us, that isn't what happens. Patents are often difficult difficult to reproduce They are in effect so long it isn't worth creating derivative works unless you have significant negotiating power (usually as a corporation), because then licensing your patent requires also licensing the original patent. And in some ways it is worse than trade secrets, because even if you reverse engineer a product, you can't make your own version without paying royalties. I suppose one benefit of patents is to reduce incentives for corporate espianage.
jnmd 8y ago
Glad to see your edit. They also point out the theoretical backing for patents not having this functionality: if a trade secret is expected to stay a secret longer than a patent lasts, companies don't file patents (why would they?) They themselves admit some holes here, but cite a general consensus that patents fail to act this way, both empirically and theoretically, in most cases.
bdowling 8y ago
Patents and trade secrets are tools for two different types of IP. If you've got a thing you can keep hidden, then you keep it as a trade secret. The quintessential example is a secret recipe like the formula for Coca-Cola or Google's PageRank. If you've got a thing you can't keep hidden, for example because it gets shipped to a customer and can be reverse-engineered, then you can apply for a patent on it. Regarding a patent revealing trade secrets, an applicant really has no incentive to do so. A patent only needs to enable a person of ordinary skill in the art to practice the invention. That has been interpreted to mean without undue experimentation, meaning some experimentation may be necessary. [0]. And there is a requirement that a patent set out the "best mode" contemplated by the inventor, but a failure to do so doesn't result in the invalidation of the patent. [1]. [0] https://www.uspto.gov/web/offices/pac/mpep/s2164.html https://www.uspto.gov/web/offices/pac/mpep/s2164.html [1] https://www.uspto.gov/web/offices/pac/mpep/s2165.html https://www.uspto.gov/web/offices/pac/mpep/s2165.html

enriquto 8y ago

> I think the general idea of patents is good Also, there may be a fundamental difference between physical inventions, where patents seem to make sense, and mathematical theorems (and computer algorithms), where patents do not make sense at all.

api 8y ago
The difference is between a specific invention vs. a law of mathematics or nature. The latter absolutely should not be patentable. The negative opinion of patents around here comes from the fact that most people here are programmers and the vast majority of software patents are trash. I've seen very few patents on software that are not "obvious to anyone skilled in the art," patents on abstract mathematical concepts, or patents for which there is ample prior art in the literature or open source code repositories. Telling the difference requires a great deal of conceptual thinking. Patent examining is "strong AI hard." I think it is possible to expand on these ideas to some useful extent. Obviousness for example could be explained as first-order connectedness is the evolutionary/learning state-space graph. Something is obvious if getting from current art to the new idea requires only one "hop" through state space or traversal of a densely connected region of state space. Non-obvious means some kind of great leap across state space or long journey through many indirections was required. It's basically a measure of creative labor. The RSA patent is often cited as an example of a decent software patent: original, a specific well-bounded invention for a specific purpose, and non-obvious in the extreme. Another example would be if whomever Satoshi is had patented proof of work block chains since that would also meet those criteria. Of course patents on cryptographic systems tend to result in them not being used, but that's a separate issue.
- enriquto 8y ago
  > The difference is between a specific invention vs. a law of mathematics or nature. The latter absolutely should not be patentable. I do not understand the difference between the two. As a mathematician with an interest in philosophy, I would say that about 50% of mathematicians would agree that algorithms and theorems are discovered, not invented. Thus, your example of RSA as a "good" patent sounds preposterous to me. The RSA algorithm was discovered, and it is a law of nature as much as the Euclidean algorithm to find the gcd. I find it honestly absurd that it can be patented (not angry, merely surprised).
  - rayiner 8y ago
    That's a bit of reductio ad absurdum. The RSA algorithm is a "law of [mathematical] nature" as much as a log cabin is natural. The composition of natural building blocks into something that performs a function is a work of human ingenuity. That said, RSA itself probably would not be patentable today. The Supreme Court's decision in Alice probably precludes pure algorithm patents. You need an algorithm in the context of a specific use case. For example, a cell phone that uses RSA to encrypt keys to protect data in flash memory.
    - pnw_hazor 8y ago
      A modern RSA patent application would get past Alice. It is a protocol that could be described/claimed in the context of communication between computers. Where the protocol improves the performance of a computers or communications because X,Y, or Z and improves the security of the session.

snarf21 8y ago

I agree. I think that all patents aren't created equal. We'd have less issues if software patents were only valid for like 5 years. You still have protection but you get rid of a lot of the trolls and the IBMs of the world having 10K patents. For more physical things, the protection can be longer since it takes longer to recoup your investment.

wolfgke 8y ago
> I think that all patents aren't created equal. We'd have less issues if software patents were only valid for like 5 years. And we'd have even less issues if software patents were only valid for 0 years (i.e. not exist).
- tracker1 8y ago
  I would rather they front load the cost for the application fee, reduce the actual license fee. And then hire reviewers with a mindset to reject if it doesn't clearly describe how to recreate a process properly, or isn't unique or non-obvious. I think another extension to the standard for patent-ability should be does this patent make something less effective (as is often the case with extension patents).
jordigh 8y ago
There often isn't even any "investment" in patents. Richard Feynmann would describe how his superiors would just walk around the labs and say, so! who has some ideas we could patent? And Feynmann's fellow scientists would say, well, you could grab this nuclear reactor and stick it into a submarine to power it, but that's pretty dumb and obvious. "Nonsense!" replied the superiors. "Patent that right away!" I don't know how true it is that a lot of patents come just from employees sitting around navel-gazing in their free time and aggressive management telling them to write down those ideas so they can file more patents, but it seems believable that it would happen at least sometimes.
- snarf21 8y ago
  Yeah, that seems like the IBM (and others approach). The problem with 0 software patents is that any small player no longer has a window to be a first mover. Google and FB have thousands of engineers sitting around and if they could just duplicate the new idea without consequence there would be no reason to do their own research and no relief for the little guy.

rayiner 8y ago

Suggesting that they’re “impossible to execute” correctly is a bit odd, as all US technology arose in the context of the patent system (the first Patent Act predates the Constitution).

derf_ 8y ago
You haven't given any evidence that it arose because of that system or in spite of it. There are plenty of examples going back to Singer sewing machines [1] and the Wright Brothers [2] that demonstrate that the surest way to delay the widespread deployment of any transformative technology by many years is to patent it. [1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1354849 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1354849 [2] https://en.wikipedia.org/wiki/Wright_brothers_patent_war https://en.wikipedia.org/wiki/Wright_brothers_patent_war
- rayiner 8y ago
  You can also find dozens of examples where investors pour billions of dollars into developing technology because they know they can patent it and won’t immediately be undercut by free riders. I’m typing this on a machine that has CPU technology licensed from ARM, GPU technology licensed from Imagination Technologies, and cellular technology licensed from dozens of companies. Those companies did, in fact, invest in all of those technologies in reliance on patent protection for the resulting developments. One can imagine alternative ways to monetize that R&D. Unfortunately, the main alternative historically was proprietary products and vertical integration. But one can imagine there is something better.
  - derf_ 8y ago
    The claim wasn't that the system is "unworkable" in the sense that it's causing world hunger or something, but that the system doesn't "work" in the sense that, in aggregate, it's not "promoting progress of the useful arts." People will certainly use patent protection where it is available to them, but that doesn't mean they wouldn't have invested in those technologies solely on the basis of first mover advantage [0], and the lack of any observed effect of strengthening the patent system and increasing the scope of what is patentable on R&D spending tends to back that up. [0] ARM would need a different business model, but it's unclear to me that other business models (i.e., Intel's) wouldn't have been just as successful, and wouldn't come with the social deadweight loss of patent trolls highlighted in the article.

renox 8y ago

> I always thought that their purpose was to eliminate trade secrets: people are going to invent stuff whether you want to or not, but without patents, they'll take the secrets to the grave. Except that if you look at the history of inventions you'll notice that quite frequently several people invented the same thing independently at almost the same time (the phone for example). So the risk of losing a 'trade secrets' isn't that high..

CalChris 8y ago
Two people can hold the same (independently discovered) trade secret. If I slave away and re-discover the Coke recipe, I can keep it and use it as a trade secret. But if I intern at Coke and steal it, I can go to jail for trade secret theft. A patent incents me to slave away and then requires me to teach the patent in exchange for a limited monopoly. Trade secret law only punishes me for stealing but incents me to keep secrets. The trouble with trade secrets is that I have to be able to exploit my invention and still keep it a secret. I probably can't do that with the phone. So I wouldn't slave away working on that problem if the best protection I could get is a trade secret.

matz1 8y ago

Trade secret is becoming less and less issue these days. Nowdays information spread and leak like crazy.

DannyB2 8y ago
And can be disassembled, decompiled and studied.
rhino369 8y ago
In some ways they are more powerful than ever since you can hide behind an API.
dnautics 8y ago
It's actually better in many cases. For processes, in order to enforce the patent, you have to get a warrant (how can you prove from the product that the process was used?). A judge will generally refuse to grant a warrant to search a competitors premises because that would become an open door for anyone to disrupt their competitor's work.

wtdo 8y ago

Not that this would be an efficient use of anyone's time but... Assuming the purpose of patents is 1) to eliminate trade secrets, 2) that patents must be novel, and 3) that patents must be non-obvious to someone skilled in the art, then perhaps a solution would be: take a handful of people skilled in the art, tell half of them to implement the patent and tell the other half only the end goal and see if any of them come up with the same process. The first group would help to see if the patent is written well enough. If they can't implement it then your patent is automatically rejected. If anyone from the second group comes up with the same process, then that's an automatic patent rejection. Even if they don't, it's still helpful information for the patent examiner. For example, if they can solve the problem, just not in the same way as the patent application proposes, perhaps the patent examiner will see that the patent isn't necessary. Go ahead and have your trade secret, the public can obviously solve that problem with or without that knowledge.

jpadkins 8y ago
the costs of your system are probably way more than the value created for society via patents.
- AmIFirstToThink 8y ago
  The cost of creating artificial exclusivity, enforced by the state against natural human learning/discovering capabilities, is even higher.
  - pnw_hazor 8y ago
    Patents cover stuff that is already invented and put into public use. How does that go against natural learning/discovery? Patents may stop someone from on purpose or accidentally copying someone else invention, but that isn't stopping discovery.
    - AnthonyMouse 8y ago
      People spend resources trying to discover things because they want to use them. If it turns out that when you go to use your invention, a patent troll jumps out from under the bridge to shake you down, people will be less inclined to spend resources trying to invent new things to use and instead just stick to the status quo.

philipkglass 8y ago

I mostly read patents about chemistry, materials, and electronics. I find a lot of older patents to be straightforward and informative. Occasionally I will run across a 21st century patent that is as readable, but they're a lot rarer now. If I had to estimate when the transition happened, I'd guess around the 1980s, maybe late 1970s. Weirdly, this is also about the time that academic publications appeared to decline in information-per-page and reproducibility-per-procedure. The societal disclosure-value of the patent seems to be dropping off over time, both because reverse engineering has improved and because the body of a typical patent is becoming less informative about how to construct a useful embodiment of the patented invention.

marcosdumay 8y ago
At the 70's - 80's our hyperoptimization cycle started. Before that society was focusing on expensive growth, since there was a lot of "empty space" created by the wars and recent science progress.
gowld 8y ago
That's because the PTO is granting garbage patents, even in the case of legitimately patentable inventions. It a main problem with patents today (along with the lifetime of a patent which was originally chosen during a much slower time for humanity's intellectual advancement).
- philipkglass 8y ago
  You inspired me to look for US patent grants by year, which brought me to this table: https://www.uspto.gov/web/offices/ac/ido/oeip/taf/us_stat.htm https://www.uspto.gov/web/offices/ac/ido/oeip/taf/us_stat.ht... In 1963 the USPTO granted patents to 54% of applications. In 2015 the rate was actually slightly lower (52%). But as "The Case Against Patents" notes, the total volume of applications has soared. I kind of expected to see some sort of upward inflection around 1980 (Bayh-Dole Act) but didn't spot one by eye. The growth rate just in this century is astonishing; there were "only" 288,811 applications in 1999. Total applications per year were below 500,000 as recently as 2009. They're now well over 600,000/year.

jandrewrogers 8y ago

The decline of practical enforceability in some patent areas has created adverse incentives with respect to how you use the patent system, leading to much of the mess you see today. As a practical consequence, less R&D is being patented since disclosure is not adequately protected, and patents are instead used strategically to increase the cost of reverse engineering a trade secret. This leads to the large number of patents designed with enough specificity to deny competitors access to technology without enough critical implementation detail to replicate the technology from said patent. Even trade secrets often only buy you a small amount of time these days before a competitor reverse engineers your technology, so increasingly elaborate (and often futile) measures are being taken to extract an advantage from R&D that justifies the investment in the first place before your competitors use it against you. Especially in computer science, as much as many people complain about algorithm patents, an enormous amount of state-of-the-art research is being solely done as trade secrets now and will never be patented or published. I suspect many people that strongly favor banning computer science patents would not be happy with the reality that this drives advanced computer science research to never be published at all. There is no longer a viable path for recovering investment in computer science research that involves publishing it, which should be cause for concern. (People that want to monitor the advancement of AI tech should find this reality particularly distressing, as it severely hampers their efforts.)

philipkglass 8y ago
"Even trade secrets often only buy you a small amount of time these days before a competitor reverse engineers your technology" -- this tends to undermine the idea that good algorithms would remain secret without software patents. Someone else suggested that you can hide some kinds of algorithms if you only provide software-as-a-service. That way there's never even a binary available for reverse engineering. The same applies for software that is used just internally, like if an aircraft manufacturer has their own geometric modeling kernel that aircraft buyers never get near. But even with NDAs for employees and no external distribution of software, good ideas leak as people change jobs and find inspiration from systems they built/used before. Sometimes the main thing you need to know is that a particular line of research was fruitful; exactly reproducing the original implementation is less important. I tend to think that it would usually take less than the 20 years of patent exclusivity for a good, intended-to-remain-secret algorithm to diffuse out of a for-profit company, or to be independently invented again by somebody who will publish regardless of patentability. Pondering a world without software patents, is it worth it to learn about a good idea 10 years "late" (assuming secrecy is attempted), in exchange for being able to freely use that idea 10 years "early"? I say with complete confidence: maybe.

thayne 8y ago

But the reason to eliminate trade secrets is to increase innovation, because other innovators can build on the released information to make new innovations. But in practice, at least in the us, that isn't what happens. Patents are often difficult difficult to reproduce They are in effect so long it isn't worth creating derivative works unless you have significant negotiating power (usually as a corporation), because then licensing your patent requires also licensing the original patent. And in some ways it is worse than trade secrets, because even if you reverse engineer a product, you can't make your own version without paying royalties. I suppose one benefit of patents is to reduce incentives for corporate espianage.

jnmd 8y ago

Glad to see your edit. They also point out the theoretical backing for patents not having this functionality: if a trade secret is expected to stay a secret longer than a patent lasts, companies don't file patents (why would they?) They themselves admit some holes here, but cite a general consensus that patents fail to act this way, both empirically and theoretically, in most cases.

bdowling 8y ago

Patents and trade secrets are tools for two different types of IP. If you've got a thing you can keep hidden, then you keep it as a trade secret. The quintessential example is a secret recipe like the formula for Coca-Cola or Google's PageRank. If you've got a thing you can't keep hidden, for example because it gets shipped to a customer and can be reverse-engineered, then you can apply for a patent on it. Regarding a patent revealing trade secrets, an applicant really has no incentive to do so. A patent only needs to enable a person of ordinary skill in the art to practice the invention. That has been interpreted to mean without undue experimentation, meaning some experimentation may be necessary. [0]. And there is a requirement that a patent set out the "best mode" contemplated by the inventor, but a failure to do so doesn't result in the invalidation of the patent. [1]. [0] https://www.uspto.gov/web/offices/pac/mpep/s2164.html https://www.uspto.gov/web/offices/pac/mpep/s2164.html [1] https://www.uspto.gov/web/offices/pac/mpep/s2165.html https://www.uspto.gov/web/offices/pac/mpep/s2165.html