by Dennis Crouch
Although you might be the first-to-file a patent application covering a particular new innovation, certain market areas are so competitive that you should expect follow-on patents from competitors that take the original idea and push it in other similar-but-different directions. Those follow-on patents won’t block you from practicing your core invention, but they may well block you from implementing practical and important elements of a market-ready solution.
One potential partial solution to this problem is being offered by a new company out of Canes, France. The company, known as Cloem, has an interesting business model of creating tens-of-thousands of what they call “cloems” based upon a client’s submitted claim set. A cloem is a computer-generated claim (or statement of an invention) that is created by the various permutations possible as well as by considering alternative definitions for terms as well as “synonyms, hyponyms, hyperonyms, meronyms, holonyms, and antonyms.” The owner’s idea is that the cloems will be immediately published and then serve as prior art to prevent competitors from claiming rights to the whitespace surrounding the original patent. Customers might also want to cloam a competitor’s patent application to prevent them expanding that subject matter.
Cloem admits that a substantial number of its results are nonsense, but a substantial number are also on-the-money and, with 50,000 cloems per patent, they can afford to waste a few.
Questions remain: (1) whether these smart-but-random computer-generated cloems should count as prior art; (2) when one of the cloems represents a seemingly patentable invention, who is the inventor and is it patentable? Robert Plotkin wrote a book that considers some of this: The Genie in the Machine: How Computer-Automated Inventing is Revolutionizing Law and Business
[…] recently came across an interesting post on PATENTLYO about a new service from France called Cloem. I’ll get to what they do in a […]
What happens when they get a restriction requirement and 1,000 of their 2,000 claims (or whatever they actually file) are restricted out? They then have to pay a huge fee to file a divisional (you don’t get the money back for filing 2,000 claims even though only 1,000 will be reviewed).
Cloem(R) texts are not necessarily filed. They can be published, this is an option. They absolutely can remain “private”, i.e. unpublished. You are free to build on top and to further file whatever you find interesting. We provide * some * examples of use i.e. patenting tactics. There are many more.
Regarding creation, Cloem(R) is a system which is everything but random. The creation is NOT random.
Best regards,
As others wrote, patenting/publishing around is a very common practice.
It is interesting that now it can ba a commodity offered by a third party.
An additional thought: what if a Patent Office acquires the services of Cloem to “clear the ground” around any filed application? and reversely, to search and indexing patents (or is this already done somehow)?
One potential use for the Cloem (R) service would be to potentially expand the scope of the First-To-Publish (FTP) grace period under AIA 102(b)(1)(B) and 102(b)(2)(B).
The Final Rules narrowly interpret the sub(B) FTP grace periods as applicable only if the “subject matter disclosed” as the triggering publication is the “same” as the subject matter of the piece of intervening art. Given this narrow interpretation accorded under the Final Rules, having a publication like this that presents numerous “variations on a theme” of the claimed subject matter would likely increase the chances of the FTP grace period being applicable in seeking to exempt a piece of intervening art.
The safest course is to Always File First; but if you can’t do that, or if you want to clear out more time and space for possible continuation-in-part filings, allowing publication of the results of this service may prove to be an interesting option for US applicants.
To the extent the Cleom is mirrored/supported by the specification the answer would likely be yes. It may shift a question from 103 to 102.
Query what happens to the specification as a whole if some Cleoms are incompatible with, logically contradictory with, or exclusions from, other Cleoms and the teachings of the specification. These would likely need to be removed, possibly proofread (AI is not perfect), and likely would cost a great deal of money to deal with.
It also opens a Pandora’s box of risks regarding claim construction… they almost definitely would need to be read in detail.
112 Fail?
Not necessarily.
Worth repeating since it’s come up several times in the comments: poisoning the well so your competitor can not obtain patent claims on obvious or even non-obvious variants of your innovation is a tried and true practice. Computers have been used for many years as aids in this strategy. Specifically, the practice has been effective in the area of polymeric chemicals such as polynucleotides where the units comprised by the polymers (e.g., the nucleotides A, T, G, C, U) are discrete and limited in number and the polymers are not too terribly long (the computer time and file size for longer and more complicated polymers creates cost/benefit issues — does anyone know the size of the largest sequence file ever submitted to the USPTO, who filed it, and why?).
Additional concerns (or features, depending on your perspective) may be raised when the computer is used to generate variants (linguistic or otherwise) which may be patentable in their own right because the state of the art is such that the description of the composition (e.g., a polynucleotide) or function (e.g., a computer-implemented information processing function) is “self-enabling”, i.e., the claim is arguably enabled upon its recitation and one skilled in the art could make at least one embodiment and use it based on the recitation of the claim alone (or perhaps with some minimal boilerplate describing the state of the art).
If that is of concern to you, then be prepared to double your concern because currently the Federal Circuit has interpreted 35 USC 103 to allow applicants to support an argument for the non-obviousness of a particular variant (typically a variant claimed only after a lawsuit target has been identified) by presenting evidence that was obtained years the application has been filed. As of this moment, that evidence can include evidence obtained from any source, including the target of the lawsuit. As far as I know, nobody’s taken this issue up to the Supremes yet. I don’t think they’d approve.
The evidence obtained later needs to apply to the filing date.
But you already knew that, right?
The evidence obtained later needs to apply to the filing date.
Is that a difficult requirement to meet? “Apply to the filing date”?
Most applications contain some boilerplate about broadly defined “improved” uses of different embodiments of the invention (“improved” relative to stuff that isn’t the invention). So the applicant provides later-generated (post filing) evidence of improvement over the prior art only when it needs to, e.g., in the litigation-worthy cases where one of these otherwise obvious or substantial utility-lacking compositions actually does function unexpectedly well and someone — not necessarily the applicant (who had no idea at the time of filing which of the obvious multitude of alternative species would actually exhibit the alleged improvement — hence it’s “unexpected” nature) — independently discovers the “unexpected” improvement or is otherwise benefitting from that discovery after the critical date of the application.
If the counter-argument is that, under the Federal Circuit’s current case law, that something quite specific is needed for later-filed evidence of unexpected results to “apply to the filing date” and be accepted, I’d like to know what you think that specific something is. Would the recitatal of a computer-generated fifteen or fifty or fifteen-hundred “ranges” of improvement that “may be achieved” using this or that specific test (again, this is the circ umstance where none of the specifically claimed species or only a tiny fraction of them were actually tested as of the filing data)? Would that sort of giant all-futures-encompassing collection of faux-detailed predictions in the original specification suffice for the later-obtained “unexpected results” to “apply to the filing date”? If so, why?
To me it seems like an excessive reward of quid for some very dubious quo. It’s giving patent rights away to people who say “Hey, if you go ahead and explore this ten billion member computer-generated list of obvious variants and you might discover something useful and surprising amongst all the worthless ones, in which case I’ll claim that specific thing, use the evidence I was unaware of to justify my patent rights to that specific thing, and then sue you on it.”
That just doesn’t seem like a great way to promote the making of those discoveries. It does seem like it would provide encouragement to a certain class of patent-minded desk-bound entrepeneurs to file endless numbers of applications describing reams of “innovations” that they haven’t reduced to practice and then lying in wait for someone else to come up that “unexpected” data that turns an otherwise unpatentable invention into solid gold.
You should check into the concept of inherency.
If I patent a basic invention, and then publish publicly 1000 variants any one of which might be useful, I’ve contributed to innovation by disclosing my invention and its variants and my end of the fundamental quid pro quo of the patent system is satisfied. If you want to pick it up and capitalize on it so be it, but pay me for 20 years, no?
This seems to be the whole point of the patent system.
I’d have to price compare against monkeys clattering on keyboards.
Well, they don’t work for peanuts.
Perhaps some math courses will assist your interpretation of the value of this.
I hope everyone sees that this is yet another reason that the foundation of prosecution should be TSM. In fact with each year that goes by, more and more is published and available to the PTO.
Are they able to easily access it? Do they access it? Do they apply it? (Or do they simply rely on U.S. patents and published applications?)
“Are they able to easily access it?”
They know how to google don’t they?
They just put their fingers together and brows.
I don’t believe that the USPTO’s job is to search out invalidating prior art. Its your job as the applicant to put prior art in front of the USPTO to improve the value of your patent. Its my job as the later challenger to find the prior art you were to lazy to find. And economics takes care of the rest (if your patent is worthless then good for you, you have a lousy patent that wouldn’t stand up in court but won’t be challenged because no one cares; and on the otherhand if its worth something I have an incentive to go find that prior art).
But the point of Cloem is prevent other people from patenting the improvements to your own patent, by creating a public record of those improvements before someone has a chance to figure out which ones are valuable enough to patent. That way you don’t have to license someone else’s patent to practice your own patent in useful ways.
“(2) when one of the cloems represents a seemingly patentable invention, who is the inventor and is it patentable?”
The computer AI that generated it.
Something abstract cannot generate something new and non-obvious. We must first ask what the abstract concept is and then remove all elements of the claims but the general purpose computer and then ask whether anything than a generic solution to each element is provided. After performing this highly objective and precise analysis, we must move on to elaborate how the claims would preempt at least one more claim.
Tell that to any actual inventor…
I believe what people think count as “abstract” and do “generate something new and non-obvous”.
And if “Something abstract cannot generate something new” then no inventor can.
Also, there appears to be a corollary in that “Something abstract” is not supposed to be patented either…
That isn’t necessarily the purpose of Cloem.
Interesting in concept. Difficult to implement in practice. USPTO will probably have a call for Tums and Rolaids. Just publishing a blob of text without context like the specification in a patent application provides is just text manipulation. What’s next claims on Twitter?
The Cloem team should come out from the shadows and identify themselves. Who are these people? Dennis are you a partner?
Thanks Arleen. Its not me and I don’t know the identity of the ‘team.’
But you do know the identity of the poster who signed off as “the Cloem team,” as well as the identity of the contributor of the post to which these comments are amended to, right?
I think that Arleen is seeking sunlight and your answer has a bit of penumbra to it.
I do not.
Your answer is unclear.
Do you not think that Arleen is seeking sunlight?
That is not credible.
Do you not think that your answer has a bit of penumbra to it?
Also, that is not credible – given your history and all.
Or does the statement refer to a “purposefully blind, I do not know because I have not checked, although I could know if I did check” answer?
This, while credible, is exactly the point of the comment of the penumbra.
Not sure why this would have to be spelled out.
“Not sure why this would have to be spelled out.”
Not sure why you bother to do it then.
I think anon is just having some fun here. Cheers, everyone!
Dear Colleagues,
The thread of existing comments is fascinating. Thanks for your interest.
Many aspects of Cloem are discussed in the FAQ section (“About” section in the top right menu).
Some particular aspects:
– machine and human contributions are deeply intermingled during the creation process. Discriminating between human and machine generated contents is not quite possible a posteriori. Nothing prevents anyone from copying one or more cloems from the database. You can further fix flaws in cloems (the skilled person also can repair a technical teaching to some extent);
– read about the technology in the corresponding section (About>Technology);
– search features will be considerably developed in the near future;
– there are many other aspects to be discussed, including epistemology.
Given the present interest, we will have the occasion to further communicate.
PS: we are currently updating the database which is currently very small (only a couple of hundreds of base claims), we soon will scale up to considerable numbers. Please forgive the bugs if any, we launched just last Monday (after considerable R&D effort)
Best regards,
The Cloem team
You should write a program for DC judges to invalidate claims under Alice.
1) remove all technical language from the claim.
2) Cut and paste claim in “I find the following to be an abstract idea:”
3) Recite the claim with all the technical elements removed and then say: “Each element can be implemented with a conventional computer.”
4) Conclusion: “If we were to permit this claim, then all would be preempted.”
: o ) Thanks. We certainly will. One step at a time. We have so many steps beforehand: electronic third party submissions, automated clearances, etc.
Can’t give you the list here. Any further idea ?
Best regards,
The Cloem team.
Can you call your program “the Duell Deal?”
I do engineeering stuff not organic chemistry but it has always troubled me when chemists tell me they have to include countless pages of individual formulae, a little diagram of each one of the hundreds of thousands of molecules within the ambit of their generic formula (with R1, R2, R3 etc) of Claim 1.
Whether or not the countless pages are included should make no difference to what the case discloses to the skilled person, say I. Oh but it does make a difference, say the chemists. You see, if it turns out that, surprisingly, the one on page 189 cures a disease, and a competitor files on this, his patentable novelty will turn on whether or not that little diagram is there on page 189.
So here is an easy and lucrative job for Cloem, no? Malcolm?
A broad genus represented by a chemical formula in the prior art might not be sufficiently specific to anticipate a claim to a species. So the novelty of the claim to the species depends on the breadth of the prior art’s disclosure.
If one were to produce a collection of random permutations of atoms and bonds by computer, it would likely end up with a large number of impossible chemical structures. This would render the collection useless as prior art, as a practitioner could simply argue that one of ordinary skill would have had little expectation of success to make the compounds described in the collection.
RH – your point about dismissing the set because of flaws in the data is an interesting one. In theory though, as time goes on it will become more reasonable that one could design a computer program for generating large volumes of chemical structures which will be useful. If that is the case, it is possible that such a computer-generated chemical data set might be used against later patent claims. Your underlying question then remains (and certainly applies to what Cloem is doing) – should computer-generated lists be considered prior art?
Whether or not this more useful volume of structures would be prior art is an interesting issue. If the list were published and the compounds in the list were (generally) enabled, I can’t think of a good reason that it wouldn’t be prior art.
However, even if it were prior art, it would probably only be useful to anticipate claims drawn to the compounds themselves. The big list of compounds would probably have very little to say about how or why the compounds would be used, so it likely wouldn’t be useful prior art for an obviousness rejection against claims drawn to compositions, methods of using the compounds, etc.
Point taken regarding use-based claims – a computer-generated list would likely not serve as prior art to stop later applications of the compounds listed.
But, as a general proposition, there is more value in a composition claim than in a process claim (granted, many assumptions are built into that statement).
Focusing only on compositions then, let’s refine the hypothetical. Assume that at some point in the future a computer program can be developed which spits out long lists of compounds having particular properties with a reasonably high degree of success. Assume one of those lists is a list of compounds which would be suitable as an adhesive, and contains compound A. Later, an applicant files an application directed to the composition of an adhesive having the structure of A. In such a case, it would seem that such an application should be rejected under a straight-forward reading of 102.
My question though – should it be rejected from a policy standpoint? Assume at some point there is an “open source” chemistry movement that works to publish computer-generated structures solely for the purpose of destroying novelty for others. These people have never practiced the inventions – their only goal is to publish for the purpose of destroying novelty.
These questions hit at the heart of what Cloem is trying to pull off here. It will be interesting to see the reaction (if any) in the courts and congress. I posit that the quantity of such computer-assisted prior art generation will only increase as time goes on. While such prior art may technically qualify as prior art under the statute, something about it makes me uneasy and feels roughly equivalent to what the Supreme Court keeps complaining about with bad patents being allowed simply because of cleaver drafting. In other words, seems like form over substance.
“let’s refine the hypothetical. Assume that at some point in the future a computer program can be developed which spits out long lists of compounds having particular properties with a reasonably high degree of success.”
Why limit the hypo to a computer list?
Digital printing is here today.
If Cloem somehow (rightly) expected that this class of compounds would be useful as adhesives, then it might indeed be useful as prior art against claims to adhesives containing these compounds. If Cloem were able to develop a reputation of correctly predicting the utility of chemical compounds, then I don’t see what one of ordinary skill wouldn’t be able to use their publications as references.
For me, the really interesting question is how Cloem could possibly foresee the potential uses of compounds. In this case, how would Cloem predict that these compounds would be useful for adhesives (but not for other uses that they would not actually work for, as this would raise issues of enablement/expectation of success) before any prior publications? Quite a few chemists make a living finding new uses for old compounds (or combinations of them), so it seems unlikely to me that Cloem would have the omniscience to forestall an entire field!
I think that would call for a good chemical simulation system.
It would take a lot of effort, but it should be able to filter a class of applications for proposed chemicals. Alternatively, it could be fed to an automated lab…
…filter…
…automated…
Sounds like you want more “math” there…
Lots of math. Especially for the chemical simulation. Some I have monitored ran for days just to get a few seconds of simulated reality.
The automated lab performs physical transformations. The measures of the results can be analyzed by math.
It has been called “synthetic evolution”. Another term is “computer invention”, though you won’t like it.
Would the test be “could be useful” or would the test be that a use be disclosed?
In the mid-1990s I was tasked with an exercise in reviewing and analyzing for my company the last 24 months of the “Research Disclosure” publication. For any unfamiliar, it was a ready way of getting invention disclosures published for prior art purposes for the (then) reasonable price of $80 per page of disclosure. I’ve not followed it, so I have no idea if this publication still exists today.
Anyway, IBM (and only IBM) had their own, separate section in each monthly publication, which generally was about 50% of each magazine publication. It was evident that many of the disclosures published by IBM were of the “iterations-on-a-theme” type, whereby they were attempting to publish as many as possible of the variations of their actually filed apps as possible. Perhaps (likely) they only did this with respect to those inventions they viewed as very important.
This (Cloem) to me just seems a logical outgrowth of IBMs publication rationale.
Noting the good discussion at 10 and 10.1, I’ll add that while an invention to be patentable may necessarily be the creation of a human, I don’t think there’s any similar limitation on the source of prior art.
So far as I can tell, it simply need be published.
Apologies: after now skimming the commentariat, I see the main point of my comment is largely redundant of themes well covered already by several others.
“… I don’t think there’s any similar limitation on the source of prior art.”
Yet, by definition / accepted understanding, doesn’t “art” … in the patent context … have to be created by a human?
Ergo, such non-human, computer-generated claims would in fact not be / qualify as prior art?
“Yet, by definition / accepted understanding, doesn’t “art” … in the patent context … have to be created by a human?”
Hi Steve. I don’t think so… Let me flail about a bit here.
What is the basis for our policies around nonpatentability of discoveries (or even independent creation) of naturally existing things?
Is is because their existence in nature is novelty destroying? The person attempting to patent it can’t get a patent because s/he was not the first inventor. Or is it because these naturally occurring things are already in the public domain, and therefore it’s not proper subject matter for patenting? Either way, the person does not get their patent.
Wouldn’t either rationale work to qualify computer generated publications as viable prior art?
The first inventor of a real transporter is not going to be anticipated by Star Trek.
Of course not, since (in any episode I’ve seen) it is presented as only an idea, i.e. the ordinarily skilled man is not enabled by that fantasy.
That said, I struggle to place the relevance of your comment in this sub-thread?
The relevance has to do with enablement and the theory here that the wiring is more akin to monkeys pounding on keys as opposed to a coherent (and enabled) writing by the hand of man.
Another Star Trek item can be brought into the discussion (and is actually closer to reality): the replicator. Of course such a thing does resemble my point often offered for discussion of my big box of electrons, protons and neutrons which would make obvious all items composed of such.
“coherent” is relative.
Even monkeys pounding on keys will eventually produce a “coherent” writing.
The only difficulty is filtering out the “incoherent”.
I wasn’t referring the law.
I was referring to the difficulty of separating the wheat from the chaff.
jesse, – you really should care about the terrain about you.
“The relevance has to do with enablement ”
And yes, I replied directly to that point. My other point was that it appears Ned intended to reply to some different thread (i.e., one in which the question of enablement was under discussion), given that the discussion Ned did reply directly to related to Steve’s question of whether or not prior art must be created by a human to be capable of being counted legally as prior art.
Mellow,
We know this:
Consider a known compound having a known use. One cannot re-patent that known compound when one discovers a new use.
Consider a product of nature. One cannot patent it because one discovers a use, even if the product of nature is not known. The product of nature is deemed “old” even if not known because it physically exists.
Now consider a compound described in a publication, but having no described way of making it or using it.
This is the problem presented by here by Cloem. This is the Star Trek problem as well.
If we allow patents on new compounds without limitation to the disclosed uses, provided we describe how to make it and to use it, why cannot we allow patents on described compounds that are not known to exist physically when an inventor first discovers how to make the compound and to use it?
I actually think this is the law.
I thought that was exactly what drug companies do. Find a new use, patent it again…
…or companies like Monsanto, who had a patent on a certain crop, had that patent expire, and then extend the life of the effect of the patent by patenting the chemical’s converse action in bioengineered foodstuffs.
Do you mean like that?
Thats one.
I was thinking of the “reformulated” pills using the same active ingredient but using different fillers… and extending the patent on the active ingredient.
“Now consider a compound described in a publication, but having no described way of making it or using it.”
Right. You apply the same standard as always – if this description is not enabling to the person skilled in the art, then it’s not enabling.
Mellow,
I may be off, but I think that the point here of even bringing up enablement is that the computer program that spits out permutations is ONLY spitting out permutations, and is not ALSO spitting out anything even remotely related to the enabling of those permutations.
One without the other, is nothing more than science fiction – regardless of the base used to start the computer program permutations.
In the “art” sense, there is no human involvement, thus the end product does not meet the patent sense of the word “art.”
No human involvement, no enablement, not prior art.
This is a bit cheeky, but something created without human involvement is known as a product of nature and may also prevent patentability.
In the particular case of Cloem, there is a significant amount of human involvement in developing the process of transforming a single claim into a set of 10,000 cloems. And at this point in their process it appears that there is human involvement with each case (making design choices regarding areas of technology and other factors) and so we might have crossed a threshold of ‘human involvement’ if it is required.
But, I don’t think that human involvement is required in any way for the creation of prior art other than the art being ‘publicly available.’ But it is only prior art for what it discloses.
anon – you are probably not “off” in many cases and perhaps in entire fields of art, that permutations tossed off by a computer will fail enablement. But in other fields, tossed off iterations on a theme are going to be enabled to the skilled person, once merely identified.
As for whether the word “art” itself requires a human, I’d say yes in the classical sense of the word. But 35 USC doesn’t apply any definition to the word “art” itself, and it says that “art” is that which is published.
Mellow, I disagree – as has been pointed out, mere publication is not enough.
If it were, science fiction would defeat patents.
Do you have a case where such is so?
“This is a bit cheeky, but something created without human involvement is known as a product of nature and may also prevent patentability.”
The only thing cheeky here is the deliberate conflation between 101 and 102 – especially since you were kind enough to print out 102 where prior art is defined.
What would be cheeky would be my questioning why a teacher of patent law seems intent on conflating different parts of the statute.
You just keep saying you disagree…. I disagree with you. I think we’re at loggerheads because we’re in very different art areas. I can easily imagine scattershot iterations-on-a-theme based on one chemical being enabling to the skilled person upon first glance at their structure.
As for “If it were, science fiction would defeat patents. Do you have a case where such is so?”
– I don’t have any concrete cases where science fiction “defeated” a patent. I recall Samsung made arguments in one of the Apple suits that Kubrick’s Space Odyssey was enabling PA as against some of the asserted Apple claims, but didn’t follow it and don’t know the outcome. Then there’s the unverified story that Heinlein’s written description of a waterbed was used against Hall’s first attempt at patent, forcing him to file a CIP and drop that parent case (the CIP did issue). But the abandoned file history is not online (not free) and I’m not going to spend money to see if the story is true or not. (My guess is, most likely internet legend).
The movie Borat was used in a rejection of a garment utility patent application, check the file history for US20090216171.
I was for a long time a reader of “hard” science fiction, where the authors felt they had to seriously connect dots as to their science. So I don’t have any trouble imagining that there could be sufficient written description to enable use of such matter in, e.g., and obviousness rejection.
Mellow, speaking of hard science fiction, may I recommend the Martian, by Andy Weir. It has been lauded by the Wall Street Journal as the best pure science fiction novel to be written in a long time.
As well, the movie version, starring Matt Damon, directed by Ridley Scott, is scheduled for release on November 25, 2015, just three years after first publication.
It is an amazing sci-fy novel. Great!
Ned, many thanks for the novel tip.
(Can’t imagine how long I’ve waited for an opening permitting venue-appropriate use of that phrase)