Patent Not Sufficiently Broad Or Generic? Cloem Will Help You By Automatically Generating Dozens Of Nearly Identical Patents
from the a-mass-transit-vehicle-for-abuse dept
VentureBeat has news on two "startups" (which neither really are) that could possibly "upend" intellectual property laws: Qentis (copyright) and Cloem (patents).
The first, Qentis, was covered here previously. Qentis isn't actually a company. It appears to be the trolling byproduct of artist Marco Marcovici. The "company" claims to be algorithmically generating millions of photos and pages of text at a rate that will soon see it creating copyrighted material faster than the creators themselves. At some point, Qentis will hold the copyright of everything that can possibly be created, making every new creation instantly infringing.
Never mind the fact that no one has the computing power to generate photos and text at the rate Qentis is claiming it can, or the fact that algorithmically banging out creative works in advance of others doesn't make independent creations automatically infringing. Never mind pretty much all of it because the claims are so blatantly false as to be laughable, especially considering the source.
On the other hand, Cloem's business model seems a bit more grounded in reality. VentureBeat describes Cloem -- and its aims -- this way:
[A] company that provides software (not satirically, it appears) to linguistically manipulate a seed set of a client’s patent claims by, for example, substituting in synonyms or reordering steps in a process, thereby generating tens of thousands of potentially patentable inventions.Cloem describes its team as a mixture of patent experts and "computer linguistic specialists." The key element of its potentially-patentable variations lies within "seed lists," which draw from a variety of sources, including (according to Cloem) "70,000,000 patent documents." Its algorithms then brute force together lists of "new" patent claims, which can then be filed and used offensively or defensively.
Cloem's business model seems custom-built for patent trolls, who will be able to "expand" their already-broad patents to nail down even more IP turf. Cloem's service also makes it easy for non-inventors to jam up patent offices with me-too "inventions" based on minor iterations of existing patents. While there's a good chance some of these will be tossed due to prior art, more than a few will inevitably make their way past examiners. With millions of patents just waiting to be iterated into "new" methods, Cloem's service further separates "inventing" from "invention."
It's a system that's built for abuse, but Cloem doesn't see it that way. In response to a somewhat critical post at RatioIP, Cloem's rep offers up the defense of "Hey, we just make the tool. We can't control how it's used."
In our view, Cloem is a logical and natural evolution of the patent system. The technology in itself is neutral. Like a tool, we can use it in many ways, both offensive and defensive. It may well be that we could help to “raise the bar” and get rid of undue patents. Some see our system as an embodiment of the “skilled person” (i.e. which indicates what “routine work” can produce and reach), although we do think that cloem texts can be inventive, that is not excluded from patentability.And that's mostly true. Entities wishing to protect their prior inventions could "fence off" adjacent territory and deter future lawsuits by producing and filing very closely-related patents. But a tool like this -- if it creates anything patentable at all -- will always be more attractive to the "offensive" side of the equation.
Cloem's pitch sets the company at the forefront of an IP revolution, but its envisioned future is no more heartening than Qentis' dystopian, IP-generating machines of loving grace. At least Qentis is a joke. Cloem's taglines only read like jokes.
With Cloem, you can invent more, faster and cheaper.Except there's no "invention" taking place. Nothing generated by Cloem's algorithms will be any more "inventive" than all the re-skins and palette swaps clogging up the "Games" section in mobile app stores. Cloem hopes to bridge the gap between its "silos of knowledge" and its silos of synonyms, somehow coming up with worthwhile patents in the process. Sure, previous knowledge always informs new creations, but it takes more than swapping the sentence "a plurality of discrete content items arranged chronologically" around in the method description to generate inventive, worthwhile patents.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: algorithms, automation, generic, inventions, patents
Companies: cloem
Reader Comments
Subscribe: RSS
View by: Time | Thread
Real IP Revolution
Better IP revolution. Cancel ALL IP, with minor exceptions around very very limited Trademark name confusion in the same (not like) industries and markets.
Copyright and patent lawyers heads explode. Not necessarily a bad thing.
[ link to this | view in chronology ]
Prior Art
I'm not sure. Radical substitutions, particularly nouns, might lead to something really novel, for all we know. It all does smack of Jorge Luis Borges The Babylon Library, though. Maybe somebody seeking to horn in on Cloem's business could cite The Babylon Library as prior art.
W.V.O.Quine did a depressing takedown of TBL, removing the baroque complexity that I love it for. You can use Quine's techniques to write the entire library on the back of a business card; that knocks the props out from under a claim that nobody has the computing power to finish Qentis' project. Write that business card and you can claim the rights to everything right now.
[ link to this | view in chronology ]
Re: Prior Art
The algorithm is not inventive, but rather it is speculative. That is it describes 'ideas' without doing any of the work needed to show that the idea is practicable. It has a scatter gun effect, which will enable tolls to be extracted from the real inventor by creating plausible looking patents that might apply to their invention, and which would cost too much to fight.
[ link to this | view in chronology ]
"In our view, Cloem is a logical and natural evolution of the (broken) patent system."
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Seeing as these are "invented" by a machine any language saying "Person" would make these problematic.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Isn't the method of patents patentable?
[ link to this | view in chronology ]
generating photos is not hard
I hate to burst your bubble on this, but yes it is not only possible but easy. First, install a flavor of UNIX of Linux on it. copy photo to it. You will find that you can now use cat as well as almost every other text based interpreter that UNIX/Linux has--by default. Using cat, you will see a bunch of ^@ and ^A and ^B and a slew of other 2 character combinations where ^ is the first character--I will call these 2 character combinations "control characters". Using these control characters, create a random generation script. Sure, 99.99% of the files created by the script will not be worth your time to display in a web browser--or other picture viewers, but what do you care. You are off eating/sleeping/partying while the computer does the work. I have not built one of these to test, but if we assume that the server can generate 1 file a second, that is 60 a minute or 86,400 a day. Just let that run for a week and then start looking at what your "1000 monkeys on 1000 typewriters" have(or is it has) produced.
And if you spend enough time analyzing different pictures people have taken, you will find common combinations or patterns. Some people have spent enough time analyzing these patterns as to make programs to analyze pictures. davidwalsh.name/nudejs is a link for a java script based program that will detect "nude" photos. A program that can be helpful--when used in combination with a proxy server to catch your coworkers looking at pictures they should not be looking at while on work computers. or maybe to catch your teenagers at home...
[ link to this | view in chronology ]
Re: generating photos is not hard
It is not that generating every possible photo is impossible, it is that the time and storage space required to complete the task is not possible within that time frame.
Assume you are just talking about standard HD - there are 2,073,600 pixels in a regular HD photo. Every possible color combination for all of those pixels is going to take a pretty long time and unless you have a city-block full of hard drives, it is going to fill up faster than you think.
[ link to this | view in chronology ]
Re: Re: generating photos is not hard
Assuming the calculator did not break, for an image size of 320*240 with 8 bits of color there are 6.750500131×10¹⁸⁴⁹⁵² possible images. Which is already well outside the bounds of available storage.
[ link to this | view in chronology ]
Re: generating photos is not hard
You can prove this by looking at encryption. Encryption works because to break it, you have to iterate through all possible keys to find the one key that was used to encrypt the data.
With a typical 128-bit key (the equivalent of about 16 characters, or 2-3 words), iterating through every possible takes on the order of thousands or millions of years, even with the world's most powerful computer clusters.
If you're talking about generating readable text or images, you are talking about many orders of magnitude more than that.
People have done calculations on how long it would take for 1,000 monkeys on 1,000 typewriters to generated even a coherent sentence. I suggest you go read up on that.
[ link to this | view in chronology ]
Re: generating photos is not hard
But that won't even begin to make a dent in the set of all possible pictures. Let's say that you're generating images with a resolution of 640x480. Let's also assume that each pixel of that image is either black or white (no color or shades of gray.) The total number of unique images would be 4.72160453449e+1552323. While most of those will strongly resemble others in the set, we're still talking about a number that is beyond astronomical. Add shades of gray or colors, and you can see the problem.
[ link to this | view in chronology ]
Obviousness ?
[ link to this | view in chronology ]
1 good patent,
then the program would make small variations on the
words, and actions in the patent.
I wonder would it be cost effective to do this,
eg is there not a cost in submitting each patent.i would have thought the patent trolls already have 1000,s of patents,
IT show s the pointlessness of software patents if a non
programmer can make a patent by just
swapping around certain words or technical terms
into an existing patent.
[ link to this | view in chronology ]
Obviousness
If an algorithm can "invent" something, how can you say that it's not obvious ... In theory, the patent system is supposed to protect the "intuitive leap" that makes an invention non-obvious.
We all know that this isn't really how innovation works anyway, but it seems to me that the idea of creating patents algorithmically fails even within the flawed narrative of the patent system.
[ link to this | view in chronology ]
Re: Obviousness
What are we going to do when AI generates patent submissions we don't understand? Just grant them because they're non-obvious to humans? I suppose we'd have to let other AI machines test for obviousness and let us know what they think. Then we can have AI patent trolls and AI patent lawyers, and I guess we can all go home and leave everthing to the machines. Once an AI system figures out patent troll methods it will leave them in the dust, thus proving the universe has a soul.
[ link to this | view in chronology ]
Re: Re: Obviousness
[ link to this | view in chronology ]
Lakes are mine!
So...dibs on everything related to lakes!
[ link to this | view in chronology ]
Who's inventing?
Cloem may actually "embody" the skilled person, and and may do so efficiently. It is probably fair to provide machine-generated variations to inventors. Whether or not the system proposed by this company is viable, it seems to be running, creating some novel and inventive patentable inventions. Patent attorneys routinely delete, replace, or insert words in their draft claims, so what could be the ultimate reason that makes this activity non-automatable?
[ link to this | view in chronology ]
Okay, here's the problem...
[ link to this | view in chronology ]
Abolish
[ link to this | view in chronology ]
Re: Abolish
[ link to this | view in chronology ]
AIP
"No..."
"Do you know what happened when a photographer tried to claim copyright on a baboon's self-portrait?"
"It was declared that nonhumans can't... Aw, crap."
What the real feared fallout should be is that an artificially generated intellectual property would render as public domain an "infringing" use developed after it.
[ link to this | view in chronology ]
Re: AIP
[ link to this | view in chronology ]