All Prior Art
April 29, 2016 3:00 PM Subscribe
All Prior Art is a project attempting to algorithmically create and publicly publish all possible new prior art, thereby making the published concepts not patent-able. The concept is to democratize ideas, provide an impetus for change in the patent system, and to preempt patent trolls. The system works by pulling text from the entire database of US issued and published (un-approved) patents and creating prior art from the patent language. While most inventions generated will be nonsensical, the cost to computationally create and publish millions of ideas is nearly zero – which allows for a higher probability of possible valid prior art.
An example: 1461187629-8442a1e2-f8ca-4427-8aa8-a370ab09ec4d
An example: 1461187629-8442a1e2-f8ca-4427-8aa8-a370ab09ec4d
A pressure vessel closure for an access opening that has a sealing surface surrounding the access opening and one of either a diaphragm or insert that spans the access opening with a peripheral flange that rests on the sealing surface. Also provided are the seeds of soybean variety XBP37005, cells from soybean variety XBP37005, plants of soybean XBP37005, and plant parts of soybean variety XBP37005. The control device performs a compensation operation on the manual focus with the focus control after completion of the manual focus. The depth information and the primary image are cooperatively used to identify whether a primary pixel images a foreground subject or a background subject.From the About page FAQs:
This is stupid / it won’t hold up / the patent office won’t use it / etc..Sister site All The Claims
While it will be great if this turns out to be a viable tool to fight patent trolls, as long as it is sparking discussion and thinking, it is performing its purpose. It’s in a way fighting an unintelligent and single-minded problem with an equally silly and brute-force method, which I find humorous. If it does turn out to not hold up in court, maybe a similar idea will. This is running off an old server in my studio, imagine if there was a patent troll with the resources of Amazon or Google putting effort towards this idea – coupling much more hardware along with better algorithms and things like deep learning actually publishing algorithmically generated patents.
This is stupid. It won’t hold up because the patent office won’t use it.
posted by The Tensor at 3:16 PM on April 29, 2016 [6 favorites]
posted by The Tensor at 3:16 PM on April 29, 2016 [6 favorites]
Love this. It also shows, ironically, that patents are valuable precisely because figuring out which possibilities in the space of potential inventions is commercially attractive is actually hard. It takes drug companies 20,000 trials to find one compound that might be worth testing, and they aren't starting from a Markov Chain generator.
posted by blahblahblah at 3:16 PM on April 29, 2016 [6 favorites]
posted by blahblahblah at 3:16 PM on April 29, 2016 [6 favorites]
This is awesome and stupid. It is art.
posted by Nelson at 3:23 PM on April 29, 2016 [18 favorites]
posted by Nelson at 3:23 PM on April 29, 2016 [18 favorites]
This is clearly infringing on the prior-art-o-matic from 2002.
posted by langtonsant at 3:24 PM on April 29, 2016 [8 favorites]
posted by langtonsant at 3:24 PM on April 29, 2016 [8 favorites]
/Patents this but on the Internet.
posted by Artw at 3:42 PM on April 29, 2016 [4 favorites]
posted by Artw at 3:42 PM on April 29, 2016 [4 favorites]
[searches for 'Hyperdrive', gets zero results]
C'mon typewriter monkeys! The clock is ticking!
posted by CynicalKnight at 4:01 PM on April 29, 2016 [3 favorites]
C'mon typewriter monkeys! The clock is ticking!
posted by CynicalKnight at 4:01 PM on April 29, 2016 [3 favorites]
Hello MetaFilter,
I created All Prior Art along with All The Claims
Saw traffic to the site from here, thanks for checking it out!
I'm happy to answer any questions.
posted by artBoffin at 4:06 PM on April 29, 2016 [40 favorites]
I created All Prior Art along with All The Claims
Saw traffic to the site from here, thanks for checking it out!
I'm happy to answer any questions.
posted by artBoffin at 4:06 PM on April 29, 2016 [40 favorites]
Welcome artBoffin! To be honest, my first reaction was, "That's ridiculous." Then I thought, well, prior art doesn't have to be enabled, it just has to be publicly available. My guess is that the USPTO might reject this as actual prior art, simply because, as the EPO's guidelines state:
posted by Existential Dread at 4:26 PM on April 29, 2016 [1 favorite]
It is enough that someone, somewhere, sometime previously has described or shown or made something that contains a use of technology that is very similar to your invention. [emphasis mine]In this case, the prior art would be randomly generated, and no person would be involved in the creation. I do think it's a pretty cool idea, as a way of highlighting the sheer volume of crap patents out there mucking up the waters.. Another serious issue is the sharp rise in patent litigation by non-practicing entities, as described in this Science article.
posted by Existential Dread at 4:26 PM on April 29, 2016 [1 favorite]
As a bonus, this also shows how nearly nonsensical actual patents are and how pathologically ambiguous the language is in patents.
posted by selenized at 4:27 PM on April 29, 2016 [2 favorites]
posted by selenized at 4:27 PM on April 29, 2016 [2 favorites]
@Existential Dread
The question of "enabling" was brought up by some lawyers but dismissed by others. The reason I created the second "All The Claims" site is that it uses the language in the claims part in the patent, which might be enabling enough (but gibberish to read for a normal person). Although, I would argue even if not seen as enabling or prior art, it could be seen as demonstrating obviousness which also could invalidate a patent.
The "someone" part you highlighted might not be defined in the actual law for prior art. We can come up with counter-examples of things made by machines which should be protect-able or patentable.
-Imagine a machine that tries mixing different substances together to cure a disease in a petri dish. If the machine finds the solution though automation and that substance is then published (or patented), could one not argue that the machine itself actually discovered the innovation and a person just set it up and programmed it?
I don't know the answer, but it is an intriguing question as more and more things we thought only humans could create and discover are done by machines. Add recent advances in, such as deep learning, and the question becomes quite timely...
posted by artBoffin at 4:39 PM on April 29, 2016 [8 favorites]
The question of "enabling" was brought up by some lawyers but dismissed by others. The reason I created the second "All The Claims" site is that it uses the language in the claims part in the patent, which might be enabling enough (but gibberish to read for a normal person). Although, I would argue even if not seen as enabling or prior art, it could be seen as demonstrating obviousness which also could invalidate a patent.
The "someone" part you highlighted might not be defined in the actual law for prior art. We can come up with counter-examples of things made by machines which should be protect-able or patentable.
-Imagine a machine that tries mixing different substances together to cure a disease in a petri dish. If the machine finds the solution though automation and that substance is then published (or patented), could one not argue that the machine itself actually discovered the innovation and a person just set it up and programmed it?
I don't know the answer, but it is an intriguing question as more and more things we thought only humans could create and discover are done by machines. Add recent advances in, such as deep learning, and the question becomes quite timely...
posted by artBoffin at 4:39 PM on April 29, 2016 [8 favorites]
As a side note, I would recommend a trip to artboffin's website. His Engineering Psychology gallery show looks quite interesting.
posted by njohnson23 at 4:48 PM on April 29, 2016 [1 favorite]
posted by njohnson23 at 4:48 PM on April 29, 2016 [1 favorite]
Very interesting!
I suspect the effect of this, if it were actually asserted to invalidate patents or reject applications, would likely be the opposite of what artBoffin wants - patent law would likely be reformed to contract the definition of prior art and limit what could be used as a reference, in ways that would lead to a strengthening of the patent regime. And if such reforms meant fewer references could qualify to be used to invalidate patents, then patent trolls would have an easier time obtaining patents, and it would be harder to invalidate their patents, and we'd all be in more of a mess.
Which is to say I think you're opening up a pandora's box! Godspeed.
posted by naju at 5:02 PM on April 29, 2016 [3 favorites]
I suspect the effect of this, if it were actually asserted to invalidate patents or reject applications, would likely be the opposite of what artBoffin wants - patent law would likely be reformed to contract the definition of prior art and limit what could be used as a reference, in ways that would lead to a strengthening of the patent regime. And if such reforms meant fewer references could qualify to be used to invalidate patents, then patent trolls would have an easier time obtaining patents, and it would be harder to invalidate their patents, and we'd all be in more of a mess.
Which is to say I think you're opening up a pandora's box! Godspeed.
posted by naju at 5:02 PM on April 29, 2016 [3 favorites]
/Patents this but on the Internet.
Patents this but in the cloud. Sues ArtW.
posted by ChurchHatesTucker at 5:10 PM on April 29, 2016 [3 favorites]
Patents this but in the cloud. Sues ArtW.
posted by ChurchHatesTucker at 5:10 PM on April 29, 2016 [3 favorites]
It's a pair of underpants that has been featured in Star Trek and can be controlled by willpower alone!
posted by kenko at 5:27 PM on April 29, 2016 [1 favorite]
posted by kenko at 5:27 PM on April 29, 2016 [1 favorite]
It's just monkeys with typewriters.
posted by Bovine Love at 5:57 PM on April 29, 2016 [2 favorites]
posted by Bovine Love at 5:57 PM on April 29, 2016 [2 favorites]
It's a pair of underpants that has been featured in Star Trek and can be controlled by willpower alone!
Been done.
posted by Halloween Jack at 6:09 PM on April 29, 2016 [4 favorites]
Been done.
posted by Halloween Jack at 6:09 PM on April 29, 2016 [4 favorites]
I wonder if it created prior art for the Google patent for a vision-correcting electronic device that has to be injected directly into your eye yet.
posted by clawsoon at 6:15 PM on April 29, 2016 [2 favorites]
posted by clawsoon at 6:15 PM on April 29, 2016 [2 favorites]
There's some case law in the US regarding the way that publications need to be categorized in order to be considered publicly available (basically, student thesis in a library that sorted the theses by Author was not considered publicly available, but in a library that sorted by subject, it was). MPEP 2128.01. One challenge to All Prior Art, as it currently works is the lack of any meaningful tagging/categorization. This might be mitigated by the availability of the search function though.
But really, the overall theme is the need to show that persons interested in the invention disclosed in the publication would have been able to access that publication. The sheer amount of junk that All Prior Art produces might be enough for a smart patent prosecutor to make the case that publications from All Prior Art weren't really publicly available. The only way that you can find a "prior art" publication on All Prior Art is to have a patent application in front of you, so that you can search All Prior Art for that exact thing. Normal "persons of ordinary skill in the art" wouldn't go looking in that database for solutions to their problems.
The reason I created the second "All The Claims" site is that it uses the language in the claims part in the patent, which might be enabling enough (but gibberish to read for a normal person).
That's not how claims work -- claims in and of themselves are usually not enabling, and need the disclosure from the specification to provide the necessary details to actually allow a skilled artisan to make or use the claimed invention.
That being said -- I love this.
posted by sparklemotion at 6:17 PM on April 29, 2016 [4 favorites]
But really, the overall theme is the need to show that persons interested in the invention disclosed in the publication would have been able to access that publication. The sheer amount of junk that All Prior Art produces might be enough for a smart patent prosecutor to make the case that publications from All Prior Art weren't really publicly available. The only way that you can find a "prior art" publication on All Prior Art is to have a patent application in front of you, so that you can search All Prior Art for that exact thing. Normal "persons of ordinary skill in the art" wouldn't go looking in that database for solutions to their problems.
The reason I created the second "All The Claims" site is that it uses the language in the claims part in the patent, which might be enabling enough (but gibberish to read for a normal person).
That's not how claims work -- claims in and of themselves are usually not enabling, and need the disclosure from the specification to provide the necessary details to actually allow a skilled artisan to make or use the claimed invention.
That being said -- I love this.
posted by sparklemotion at 6:17 PM on April 29, 2016 [4 favorites]
claims in and of themselves are usually not enabling, and need the disclosure from the specification to provide the necessary details to actually allow a skilled artisan to make or use the claimed invention.
Yeah, and they should also not be "obvious" to skilled practioners in the field. The USPTO has utterly failed there (they weren't allowed to use google until recently) so brute forcing actually seems sensible.
posted by ChurchHatesTucker at 6:30 PM on April 29, 2016 [1 favorite]
Yeah, and they should also not be "obvious" to skilled practioners in the field. The USPTO has utterly failed there (they weren't allowed to use google until recently) so brute forcing actually seems sensible.
posted by ChurchHatesTucker at 6:30 PM on April 29, 2016 [1 favorite]
This is an issue that is near and dear to my heart, as I am a patent lawyer. (I know there are several of us on Metafilter; I don't claim to be smarter than any others who pop up if they have contrary opinions.) While I have no criticism of this as an art project or even just a stunt, I am abidingly confident that it will be of no use to actually invalidate any patent or prevent any "bad" application from issuing.
I will admit, though, that one question which is actually tantalizingly unclear is whether the various individual pages constitute "printed publications" under the statute. I was originally confident the answer is "no," based on the cataloging/indexing rationale mentioned above. But alas, that MPEP section (2128) is woefully out of date on the issue; the more current law is SRI International and In re Lister (the latter described in this [PDF] law review article). Long story short, it is a very fact-specific analysis and it seems at least possible that a court could find these pages to be prior art.
But there are at least a couple reasons I'm still pretty confident it would actually be useless as a source of prior art. First, a practical issue. Let's say I am a person trying to invalidate a bad patent. How, specifically, do you suggest I use this site to do that? Take terms from the patent and search this site for them? The site just takes all the words in all the patents and jumbles them up; why shouldn't I just go run my search against the actual corpus of patents, where those words came from? At least those results would be intelligible, whereas the AllPriorArt results are overwhelmingly likely to be nonsense. I mean, it may be Markov chained to make proper sentences, but it's still nonsense on a substantive level.
Second, there are the procedural and persuasive hurdles to overcome. In most proceedings, the burden is on the challenger to demonstrate the patent is invalid, typically by "clear and convincing evidence." To meet this burden you have to tell a story. You can't just throw a pile of words in front of the court or PTO and assert "therefore it is invalid"; you need to meticulously analyze the problematic claims line-by-line and explain how the claimed concepts are present in the art. And not just in a hodgepodge sense, with some words on some pages and some words on others and everything technically in there somewhere, but rather all woven together so that the substance of the target invention is fairly disclosed. Even where you don't technically bear that burden, such as if you just mail prior art to the PTO while the 'bad' patent is still being examined, you still have to persuade the examiner or the tribunal that you're right, which puts you on the hook for basically the same analysis.
posted by Joey Buttafoucault at 7:43 PM on April 29, 2016 [10 favorites]
I will admit, though, that one question which is actually tantalizingly unclear is whether the various individual pages constitute "printed publications" under the statute. I was originally confident the answer is "no," based on the cataloging/indexing rationale mentioned above. But alas, that MPEP section (2128) is woefully out of date on the issue; the more current law is SRI International and In re Lister (the latter described in this [PDF] law review article). Long story short, it is a very fact-specific analysis and it seems at least possible that a court could find these pages to be prior art.
But there are at least a couple reasons I'm still pretty confident it would actually be useless as a source of prior art. First, a practical issue. Let's say I am a person trying to invalidate a bad patent. How, specifically, do you suggest I use this site to do that? Take terms from the patent and search this site for them? The site just takes all the words in all the patents and jumbles them up; why shouldn't I just go run my search against the actual corpus of patents, where those words came from? At least those results would be intelligible, whereas the AllPriorArt results are overwhelmingly likely to be nonsense. I mean, it may be Markov chained to make proper sentences, but it's still nonsense on a substantive level.
Second, there are the procedural and persuasive hurdles to overcome. In most proceedings, the burden is on the challenger to demonstrate the patent is invalid, typically by "clear and convincing evidence." To meet this burden you have to tell a story. You can't just throw a pile of words in front of the court or PTO and assert "therefore it is invalid"; you need to meticulously analyze the problematic claims line-by-line and explain how the claimed concepts are present in the art. And not just in a hodgepodge sense, with some words on some pages and some words on others and everything technically in there somewhere, but rather all woven together so that the substance of the target invention is fairly disclosed. Even where you don't technically bear that burden, such as if you just mail prior art to the PTO while the 'bad' patent is still being examined, you still have to persuade the examiner or the tribunal that you're right, which puts you on the hook for basically the same analysis.
posted by Joey Buttafoucault at 7:43 PM on April 29, 2016 [10 favorites]
artBoffin: Based on the fact that a random GUID doesn't result in a page (eg), it looks like you're storing the generated text on the server?
If you use the GUID as a key to deterministically generate the text, you don't have to store the generated texts - you just recreate them when they're requested.
Does "indexed by Google" equate to "published"?
posted by Leon at 7:56 PM on April 29, 2016 [1 favorite]
If you use the GUID as a key to deterministically generate the text, you don't have to store the generated texts - you just recreate them when they're requested.
Does "indexed by Google" equate to "published"?
posted by Leon at 7:56 PM on April 29, 2016 [1 favorite]
The Laboratory of Babel
posted by thelonius at 9:10 PM on April 29, 2016 [7 favorites]
posted by thelonius at 9:10 PM on April 29, 2016 [7 favorites]
Here I was getting scared I was going to have to make the Borges reference. Thank you, thelonius!
posted by traveler_ at 9:29 PM on April 29, 2016 [2 favorites]
posted by traveler_ at 9:29 PM on April 29, 2016 [2 favorites]
Does this mean the framiskate latch really comes undone from the widget arm? I always wonder about this in posts that mention flanges.
posted by Oyéah at 9:33 PM on April 29, 2016
posted by Oyéah at 9:33 PM on April 29, 2016
Everything about this, including this thread, is way cool.
posted by Potomac Avenue at 12:40 AM on April 30, 2016 [1 favorite]
posted by Potomac Avenue at 12:40 AM on April 30, 2016 [1 favorite]
Did I ever mention that web comic I ran which was iterating systematically over every possible image, and therefore would eventually contain every other web comic?
posted by qntm at 4:57 AM on April 30, 2016 [1 favorite]
posted by qntm at 4:57 AM on April 30, 2016 [1 favorite]
« Older Nolite te bastardes carborundorum | The third year, we built a ship in the desert Newer »
This thread has been archived and is closed to new comments
posted by ignignokt at 3:10 PM on April 29, 2016 [2 favorites]