Today's Date: December 20, 2014 Search | Home | Contact | Services | Patent Attorney | Patent Search | Provisional Patent Application | Patent Application | Software Patent | Confidentiality Agreements

Do Scientists Read Patents?


Written by Lisa Larrimore Ouellette
Fellow, Yale Law School Information Society Project
Written Description Blog | Twitter | SSRN
Posted: July 18, 2013 @ 7:45 am

Tell A Friend!

Do patents disclose information that is technically useful to researchers in the same way scientific publications are? The Supreme Court has described disclosure as one of the main purposes of the patent system in many cases. But most patent scholars argue that researchers don’t read patents because (1) patents don’t actually have useful technical information—they are just confusing legalese; and (2) researchers wouldn’t read patents anyway because they’re worried about being liable for willful infringement.

I decided to study this issue by conducting my own survey, since it turns out that we have relatively little information about how researchers actually use patents. Most previous surveys predate the ready availability of patents online, but even these surveys suggest that patents are more useful than many commentators suggest. For example, a 1994 survey of manufacturing firm R&D managers is frequently cited for propositions like that “scientists don’t in fact gain much of their knowledge from patents, turning instead to other sources.” But what it actually found is that about 50% of U.S. respondents said patents were “moderately” or “very” important as a source of information for a recent project, which is less than the 62% who said the same of scientific publications, but which is still a sizable number. (Interestingly, patents were by far the most important source of information in Japan!)

I focused my survey on researchers in nanotechnology, partly because that’s the field I did my Ph.D. work in, but also because it is an emerging field where most of the work is still being done by basic researchers—which means there’s a vibrant scientific literature in addition to the patent literature, and one might thus expect patents to be even less useful than in more applied fields. You can find all the details about my 211 respondents and the specific survey questions in my article, Do Patents Disclose Useful Information?, but I’ll highlight four key results:

  1. 64% of the respondents have read patents.
  2. Of respondents who were reading patents to look for scientific information (rather than for legal reasons), 60% found useful technical information. They said things like:
    • “Patents are a useful source of information on how others have approached particular technical problems and can also help you from going down a road that has already been traveled.”
    • “Patents offered some plausible views as to what we were seeing in our own research.”
    • “Protocols or recipes for preparing samples or performing experiments are described that are not found in other published literature.”
    • “The way a new technology is described is much more reliable and reproducible in a patent than in a scientific paper” because in papers “academic researchers purposely remove essential steps for reproducing data,” so “I trust patents more when I need to try other people’s technologies.”
  3. Even though patents were useful for many scientists who read them, only 38% of patent readers thought the patents they read were reproducible.
  4. Only 3 out of the 211 respondents might be characterized as avoiding patents because of willful infringement concerns.

Some of my former nanotech colleagues and I also looked at specific nanotechnology patents, and our findings complement the survey results: we found that there was useful technical information in the patent literature that was not duplicative of the scientific literature, but we also found that many patents did not contain all the details that we thought would be necessary to reproduce the claimed inventions.

I think these results show that the disclosure function of patents is working to at least some extent, at least in nanotech: patents can be a useful source of technical information, and willful infringement concerns are not preventing most nanotech researchers from looking at patents. But there are still problems with patent disclosure: the result that only 38% of patent readers think those patents are reproducible raises questions about whether disclosure ought to be improved just to meet current legal requirements.

Now, this does not mean that technical disclosures are a justification for the patent system’s costs—I think the reason we have a patent system is to incentivize the creation and commercialization of new inventions. To me, the interesting question is: given that we have a patent system, how much disclosure should we require? In other words, how much benefit are we getting from patent disclosures, and how does that compare to the costs to innovation incentives and the administrative costs of requiring that disclosure? And could we benefit more by requiring better disclosures without harming innovation incentives? I think these questions open up at least three interesting avenues for further research:

First, what kind of useful information do scientists find in patents, and how does the use of patent disclosures vary by field? Do researchers use patents primarily to build on the patented invention, or are there significant spillover benefits for users not contemplated by the patentee? I plan to expand my survey into additional fields, including more industry researchers, and to complement large-scale survey results with more in-depth interviews. I expect results to look very different for software developers than for pharmaceutical researchers, but it will be interesting to see how. Perhaps areas where disclosure isn’t working well can learn something from areas where it is.

Second, how can we measure the benefit of improved disclosure? In a new working paper on patent policy experimentation, I suggest that it might be possible to make some progress on this using small-scale experiments. For example, programers could be asked to implement fictitious software patents with varying types of disclosure (e.g., as part of an undergrad CS class, or in something like this experiment with a TopCoder competition), and their programs could be compared based on the time needed to produce a functioning program and the program’s quality based on metrics such as runtime.

And third, if we decide that it is worth improving patent disclosures, how do we do that? Examiners I’ve spoken with prefer to focus on prior art issues than on enablement, which makes intuitive sense to me: even having a Ph.D. in physics, there are only a tiny number of inventions for which I really would feel comfortable determining whether “undue experimentation” would be required for a PHOSITA to practice the claims. Is it possible to make it easier for examiners to accurately evaluate enablement? And are there ways to improve the value of patent disclosures to scientists beyond better enforcement of current legal requirements?

I don’t know the answers to these questions, but I’d like to figure them out; my hope is that further increasing scientists’ engagement with the patent literature will benefit both science and patent law. And I welcome suggestions from IPWatchdog readers about any of these research directions!


About the Author

Lisa Larrimore Ouellette is a Postdoctoral Associate in Law and Thomson Reuters Fellow at the Information Society Project at Yale Law School. She studies IP law both empirically and theoretically, and her latest scholarship is forthcoming in the California Law Review and the Texas Law Review. She also blogs about IP scholarship at Written Description and Tweets about new papers and other patent news as @PatentScholar. Before receiving her J.D. from Yale Law School, she earned her Ph.D. in physics from Cornell. She has also clerked on the Federal Circuit for the Honorable Timothy B. Dyk.


24 comments
Leave a comment »

  1. Very interesting. Thank you.

    It may be useful to distinguish between “discovery” (e.g., laws of nature, etc) and “invention”.

    The patent system by design (USC 101) excludes discovery.

    Each researcher may choose to work somewhere between all discovery and all invention; with some around the 50/50 overlap in the middle.

    A spillover from invention (patents) back into discovery, seems like a unintended bonus of the patent system?

  2. “(Interestingly, patents were by far the most important source of information in Japan!)”

    Funny you should mention that since there are glaring differences in how Japanese patents are typically drafted, and I can see how their ways of drafting an application make the eventual publication many orders of magnitude more useful to those of skill in an art. Take a look for yourself sometime at how differently they are drafted. There is some legalese in Japanese documents, but it takes a backseat to actually disclosing a very specific embodiment usually in crystal clear terms. On the whole I much prefer Japanese references for substance than I do US documents even though Japanese isn’t my language and I have to read them in broken English translations.

  3. “But there are still problems with patent disclosure: the result that only 38% of patent readers think those patents are reproducible raises questions about whether disclosure ought to be improved just to meet current legal requirements.”

    Something which I have championed for quite some time now. The current enablement requirement is simply too low in this day and age to make what we get from the old school quid pro quo (especially when it is combined with the newschool drafting methods) of enough value to justify the patent term. In the past things were usually much simpler to reproduce. But over the years I find my concerns somewhat tempered and I understand why the Federal Circuit put that standard in place. Even so, a more demanding requirement I still feel like would be for the overall benefit of the system and the American people that are supposed to benefit from the quid pro quo.

    “Now, this does not mean that technical disclosures are a justification for the patent system’s costs—I think the reason we have a patent system is to incentivize the creation and commercialization of new inventions.”

    Whereas I believe pretty much the opposite. If anything the later is nothing but a tagged on after thought. The reason I come to this conclusion is that the technical disclosure is what we get in the quid pro quo. If we truely want patents to incentivize the creation and commercialization of new inventions we should relax the obviousness requirement to an absurdly low level, and also require commercialization in order to procure the patent and/or to maintain it for more than a few years. In other words, the quid pro quo should mirror exactly what we want to get out of the system. Just as I believe it currently does. We get the technical disclosure, period, and we may or may not get commercialization and creation of new inventions.

  4. Even if the patent disclosure itself is not a reference that people go to, the patentee, by allowing publication of the information in the patent, has voluntarily forfeited his/her right to keep the information a secret. Since it is no longer a secret, other disclosures by the patentee may be more informative than they would be if the information was being protected as a trade secret.

  5. “Even if the patent disclosure itself is not a reference that people go to, the patentee, by allowing publication of the information in the patent, has voluntarily forfeited his/her right to keep the information a secret. Since it is no longer a secret, other disclosures by the patentee may be more informative than they would be if the information was being protected as a trade secret.”

    This is the correct answer. It doesn’t matter if scientists read patents, the filing of an application is what allows the information to get out, be disseminated, etc.

    My cousin works at a lab of a major east coast university. He doesn’t read patents. He does read journals and other papers relevant to his field. He said to me once, “I heard that Dr. X got a patent on [insert esoteric subject matter here]. Everybody’s known about that for years, so how could he get a patent on it?”

    I laughed and explained to him that probably what happened was the day after Dr. X filed his application, he published a paper in the Journal of Stuff Interesting to PhD’s in EE Working in the Field of Applied Physics, which my cousin does read. Of course, the patent takes 3+ years to issue, but by that time, “everybody knows about it.”

  6. Excellent point from John Darling. By the time the issued US patent emerges, the contribution to the art has long since been assimilated by those at the cutting edge.

    But what about A publications, 18 months after the priority date. You know, what they read everywhere else in the world. WO publications and the like?

    There is some respectable academic paper, research Report, out there somewhere, as I recall, that shows that A publications regularly beat all other publications, as the earliest to disclose innovations to the general public.

  7. While I appreciate John Darling’s as well as Cowboy’s post, and I appreciate that such disclosures are routine, especially in some arts, the fact of the matter is that “allowing” the information to get out shouldn’t be sufficient justification. If this “beefed up secondary disclosure in a journal” or whatever else publication is what we want to ultimately happen then we should go ahead and put that in the quid pro quo since that is what we’re assuring we get at a minimum. These somewhat esoteric and abstract concepts of the mere possibility of enhanced tacked on disclosure seems to me a far cry from a sufficient baseline justification for the whole system and its costs. Especially today. Funny also that the founders never mentioned this as their justification. They were trying to give the useful arts liberal encouragement if I remember correctly. Not a tiny shot in the arm. I likewise doubt if you told congress that the major or sole justification for the costs of the patent system (totalling at least several billion dollars a year) was the mere possibility that some patentees would beef up their other disclosures that the members would fall all over themselves to support it.

  8. Brother Smith, I am not sure it makes much sense to try to get into the heads of the founders to interpret current patent law. When the Founders were drafting the Constitution and setting forth the quid pro quo, publishing something was difficult, expensive and rare, unlike today. Keeping secrets was easy. And inventions in their day were mostly mechanical devices that are often quite fully disclosed by drawings.

    And bear in mind, that patent disclosures are sufficient if PHOSITA can make and use what they purport to describe. A PHOSITA in microbiology may very well not know how to make and use a microchip invention from the patent on the microchip, but that does not matter, as long as the microchip PHOSITA can. The original posting was that “only 38% of patent readers think those patents are reproducible.” Well, if you are reading a patent for which you are not PHOSITA, you may very well not think the thing is reproducible.

  9. “When the Founders were drafting the Constitution and setting forth the quid pro quo, publishing something was difficult, expensive and rare, unlike today.”

    Agreed, and that was a more substantial reason to even have a patent system in the first place. Note how important the quid pro quo we preserve today was back then, and how unimportant it is in the modern day relative to then.

    “And bear in mind, that patent disclosures are sufficient if PHOSITA can make and use what they purport to describe.”

    I’m well aware of the current standard, I simply think a slightly higher standard is more appropriate for today. It does rather seem to me that if we’re going to go through all the motions we may as well give a full and rather complete make and use. The half disclosures of yesteryear simply strike me as not worth the bargain. John proposes that perhaps we’ll get a better disclosure in a journal later for some inventions, which is true, but it surely isn’t true for all of them. But then again I’m not congress, and perhaps they really do prefer nearly, but not quite, useless disclosures. And perhaps this is because there are so few engineers in congress relative to the number of lawyers.

    “The original posting was that “only 38% of patent readers think those patents are reproducible.” Well, if you are reading a patent for which you are not PHOSITA, you may very well not think the thing is reproducible.”

    I’m not entirely sure, but I would presume that the people reading the patents that were surveyed stayed more or less within their own field. Also note him and his buddies in nanotech verifying the findings. It isn’t surprising in the least. Read some nanotech patents, there is a ton left out that you’d really need to make a genuinely useful embodiment, just as most lawyers will tell you is “proper” under our current scheme. Sure, one of skill could eventually figure it out, but I’m simply arguing for a higher standard so that if we’re giving out a patent, the person copying the thing after expiration doesn’t have to spend a million dollars to re-invent the invention that was supposedly enabled already when we could have just got a full manner of making and using the thing in the first place from the patentee for no higher price than fiddling with the statute a bit.

    But hey, it’s just my opinion. I piped up because of the nature of the original post.

  10. In the Chemistry space, patents have overtaken non-patents as the primary (>50%) source of new Chemicals deserving of a registry number from Chemical Abstracts (American Chemical Society). This was mentioned as a side note in a talk I attended by one of the CAS talking heads and I apologize for not providing a link to the presentation…

  11. John Smith.

    I cannot identify exactly what you mean by patent costs.

    Particularly, “the costs of the patent system (totalling at least several billion dollars a year)

    Can you explain?

  12. I wanted to echo the point made by John Smith about the value of Japanese patent disclosures. My field was steel making from my R&D days. I remember one Japanese patent on galvanizing that was especially useful since it provided detailed results from their experimentation. I was very impressed.

  13. ” I likewise doubt if you told congress that the major or sole justification for the costs of the patent system (totalling at least several billion dollars a year) was the mere possibility that some patentees would beef up their other disclosures that the members would fall all over themselves to support it.”

    John Smith,

    It sounds like you are under the impression that the US patent system (e.g. USPTO) is supported by tax dollars. Not true. The entire cost of the US patent system is born by applicants. And at least by my estimate,the cost to applicants is a about 10-20 billion per year. (USPTO about a billion, drafting maybe another 5 billion, legal fees for enforcement about the same)

    Which is sad, really. US GNP is about 15 trillion per year. That means that collectively patents only add about 0.1% value to our total economy. Not a big number.

  14. MN,

    The concept of ‘cost’ is exactly what I was trying to understand from John Smith.

    It does not appear that the concept is being understood (or at least) articulated with the meaning intended.

  15. A Patent System Financial Model:

    Patent System promotes investment into Research & Invention when:

    [$ for use of inventions (patents)] / [discount factors] > [$ spent on research-invention] + [$ for patent system overhead].

    where:

    [discount factors] = discounts for time-value-of-money, risk, etc (e,g., net present value).

    [$ patent system-overhead] = [$ USPTO] + [$ write-obtain patent] + [$ enforcement overhead-inefficiencies].

    Note: [$ enforcement overhead-inefficiencies] does not include [$ paid for use of inventions (patents)].

    A few Model observations:

    The greater the [$ for use of inventions (patents)]; the more incentive to invest in Research & Invention.

    Patent System is efficient when: [$ patent system-overhead] << [$ spent on research-invention].

    Simply reducing [$ USPTO] and/or [$ write-obtain patent] may be counter-productive; by increasing [$ enforcement overhead/inefficiencies] by a greater amount.

    Note: This is only one of the ways-models of how the patent-system promotes Research & Invention.
    In addition, there are also many other ways-models of how it promotes commercialization.

  16. Inventor0875,

    You are taking a classic (and classicly mistaken) singular view of ‘promote.’

  17. I have to echo PS DIP. In the chemistry world, most especially the small molecule world, patents are very useful especially for filling in the blank spaces in the database. The first patent application for a new drug may disclose hundreds of analogs acroass a broad range of chemical space, each of which is a distinct chemical compound not previously found in the literature. In view especially of certain foreign patent jurisdictions in which a patent cannot be granted on a chemical compound which has not in fact been synthesized, each of these new analogs will typically have at least some analytical data connected. The papers which follow will typically show fewer than twenty compounds in a much narrower chemical space, and are less useful unless you are just trying to design around the patent. The patent disclosure is far more likely to open up chemical space to only peripherally related projects or targets than the papers directed to the most active compounds against a single target.

    Are the patent syntheses reproducible? Not always, and they are very rarely optimal reactions, but it certainly improves the useful hit rate of database searches for new chemical syntheses. Unfortunately, chemical patents are usually massive documents in the 100-300 page range, and are not always the most carefully edited manuscripts. After you have reported NMR and MS data on fifty or so compounds, you start to miss a few numbers. However, once the synthesis is published, it is much easier to get helpful information in informal discussions at ACS meetings about how the syntheses were actually carried out and other useful hints, especially if your use for the compounds is directed to a completely different target than the inventor’s use.

  18. “The current enablement requirement is simply too low in this day and age to make what we get from the old school quid pro quo (especially when it is combined with the newschool drafting methods) of enough value to justify the patent term.”

    I disagree and would echo AC’s comments.

    “If this ‘beefed up secondary disclosure in a journal’ or whatever else publication is what we want to ultimately happen then we should go ahead and put that in the quid pro quo since that is what we’re assuring we get at a minimum.”

    I’m not sure that every journal publication following a patent application filing is a beefed up disclosure. But the way that I’ve always viewed it is that a patent application is basically a snapshot of the technology at a particular point and time in its development. Presumably the inventor has consulted with counsel as to when the application should be filed, and has considered the relevant questions regarding conception, reduction to practice, enabling disclosure, sufficient possession of any invention that may be claimed, etc. But just because the application is filed doesn’t mean that the development of the technology stops. And if further developments lead to more patentable subject matter, the inventor is entitled to file another application to cover that subject matter and continue the process, including subsequent publication if that’s desired.

    MD,

    I’d agree that A pubilications are a fairly good indicator of trends in whatever technology you may be involved with. Many of them are brought to my attention. But the inventors I work with have additional sources of information about the trends in the technology beyond A publications. We have to pay attention to all of them.

  19. I quote: “4.Only 3 out of the 211 respondents might be characterized as avoiding patents because of willful infringement concerns.”

    That leaves me sceptical.

    I ask: IF a respondent were indeed avoiding any scrutiny of A publications because of concerns about willful infringement, THEN would such respondents feel free to admit it to the researcher?

    My guess is that any respondents that know about willful infringement also know that they are not to admit it.

    Going back decades, in the UK, the malpractice insurance contract I once had to peruse had in it a clause that the insurance is null and void if you let on to the client that you have it in place. What insurance do the respondents have in place, against willful infringement?

  20. Sure Anon, the costs of the patent system are at least a. all the fees associated with the filing and prosecuting of patents both in the office and through lawyers and b. all the money exchanged through licensing and litigations/settlements to use patented technologies along with c. all the costs of litigations and other legal incidentals. Along with all the indirect costs of using the courts, forgone research or business expansion into fields rife with patents, taking bright engineers away from engineering to focus on patent law related things, taking accused infringer’s time away from running their business to deal with the patent matter, on down the line of similar incidental things.

    For the purpose of my comment however, I simply used the baseline costs of operating the PTO as a starting place, their budget is around 2 billion dollars a year which is funded by fees. Though some others estimate the costs at several hundred billion dollars a year.

  21. “It sounds like you are under the impression that the US patent system (e.g. USPTO) is supported by tax dollars. Not true”

    I’m not Mark.

    “The entire cost of the US patent system is born by applicants. And at least by my estimate,the cost to applicants is a about 10-20 billion per year.”

    That depends on how you define “costs” and my definition is a bit broader than yours. But I like your estimate for those limited costs.

    “That means that collectively patents only add about 0.1% value to our total economy.”

    Again, that depends on how you define value. I would take a more broad approach. Though I wouldn’t take the approach that the office used in a recent paper on the value of IP in total. Frankly I think the patent system is of enormous “value” though most of that value would be measured in disclosure, not dollars, and we could get a lot more of this “value” by tinkering with the sections of 112.

  22. John,

    b) is not a cost – you have that in the wrong column.
    a) and c) are simply the cost of obtaining the benefits – also not something you can treat as a pure negative number as you are attempting. You neglect the gains of IP.

    Your simple use of the baseline cost is simple error.

    Your tendency to define value seems far too subjective, and far too negative. Sorry, but you have lost credibility with me.

  23. “b) is not a cost – you have that in the wrong column”

    You may feel free to define it that way if you wish, as for myself I will be a realist and use standard definitions. Those monies are all values of money that are used up to produce a license/settlement/payment of a court imposed debt, i.e. a cost.

    “are simply the cost of obtaining the benefits ”

    Simply … cost(s). My point exactly.

    “You neglect the gains of IP.”

    I’m not “neglecting it”. The monetary “gains” of “IP”, or “patents” specifically which is at issue at the moment, are the “loses” or “costs” of another person. It is fine to count them as costs though you might also count them as gains in another context. Really it is just money that has been used up by one party to produce the “benefit” of the patent system for the patentee, that is traditionally known as a “cost” regardless of the fact the money went to someone else and the other person benefited. I’m sure that your indoctrination has helped you to overlook this plain economic definition but it is never the less quite a plain definition.

    But if you prefer, we can count that as both a gain and a cost. Doesn’t really matter, the cost is what we’re focusing on right now. If you prefer, we could look at the “gains” later.

    “Your simple use of the baseline cost is simple error.”

    It was my comment, I get to use baseline cost if I feel like it. You’re not king of what I choose to say. Nor is it an “error” to take the view I am, we can look at the flip side of the coin later on, just because I haven’t yet doesn’t mean it is an “error”.

    “Your tendency to define value seems far too subjective”

    I’m willing to expand my definition of value to encompass all that could reasonably be called “value”, including those things you feel are “benefits”, but just because I will throw something in with “value” doesn’t mean it will magically disappear from the column of “costs”. Those numerical values can go in both columns simultaneously.

  24. as for myself I will be a realist and use standard definitions.

    That’s the point, John, you are not using standard definitions. Your bias is showing, and quite frankly, there is little point in my explaining to you how the real world works because it is evident that you do not want to know how the real world works.

    As evidenced by: “The monetary “gains” of “IP”, or “patents” specifically which is at issue at the moment, are the “loses” or “costs” of another person” you wish to reduce business to a “someone must lose” game. That is not how the real world works. Being the king of what you say is not my goal. That you feel compelled to be such a king, and make the mistakes you make (while insisting that it is I that is not in the real world), paints clearly the picture of just who is inhabiting a kingdom of their own mind.

    Good luck with that.