New 'Company' Claims It Uses Algorithms To Create Content Faster Than Creators Can, Making All Future Creations 'Infringing'

from the [algorithmically-generated-trollface] dept

Over the weekend, TorrentFreak covered the discovery of the latest thing in copyright enforcement: algorithmically-generated content created solely for the purpose of extracting infringement settlements and licensing fees.

That’s the staggering notion being put forward by Qentis Corporation. The outfit, which claims a base in Russia, says that its business model is to use massive computing power to generate digital intellectual property on a never-seen-before scale and transfer the rights to its partners.

“Our clients are private high net-worth individuals (HNWI), investment funds and corporations that act as pure investors,” Qentis explains.

What Qentis is proposing is the bulk algorithmic creation of content – music, text, images etc – on such a large scale that in a few years its clients will own the rights to just about anything people might care to create and upload.

The creator of Qentis, Michael Marcovici, told TorrentFreak that his “company” had the potential power to generate content before actual creators can, resulting in a world where every new work is already infringing.

“Qentis aims to produce all possible combinations of text (and later on images and sound) and to copyright them,” Qentis’ Michael Marcovici told TorrentFreak.

“Concerning text we try this in chunks of 400 word articles in English, German and Spanish. That would mean that we will hold the copyright to any text produced from now on and that it becomes impossible for anyone to circumvent Qentis when writing a text.”

By 2020, supposedly every possible photograph will have been created and registered by Qentis. Text content generation is advancing at a faster rate.

Qentis — as a concept — is frightening. As an actual entity, it’s an ultra-dry satirical device. Marcovici’s website isn’t the future of anything. The computing power needed to accomplish this is beyond the means of anyone. Brute force creation results in tons on unusable “content,” something Marcovici readily admits.

“About the mathematics, this is mainly about working with n-grams, we don’t work iteratively with misses because that would produce as you mention a LOT of misses, probably only 1 out of few million would be readable,” the company’s Michael Marcovici told us.

Qentis is a piss-take on utilitarian content creation and over-broad content protection. It seeks to embody the worst aspects of automatically-generated content and copyright trolling. And it pretty much nails both, presenting a respectable corporate front that almost masks the insanity leaking in around the edges. A quasi-proof of concept page claims Qentis’ software “wrote” Lady Gaga’s “Applause” four years before she did. On its About page, it notes that it has already generated “97.42%” of all 400-words-or-less text in several languages before dropping this bombastic (and misspelling-laden) statement.

If you are planning to publish any text in these languages we must inform you that the chances are almost 100% that they are already part of the copyrighted inventory of the Qentis Corporation and that you are about to violate these and you will be held responsible for this.

Qentis does not issue permission to individuals to publish any of its texts or images, please do not try to inquire. Qentis grants writes for reproduction only two is selected group of publishers.

Another page claims this company will free online content providers from the hassle of creating content. Instead, all content roads will lead to Qentis, from which rights to its algorithmic creations will be distributed to a variety of middlemen (“high networth individuals”), who will then license the content.

Whoa if true, but you’d have to ignore the computing power needed to brute force content creation that covers almost every conceivable combination of words — especially given that the language keeps evolving and changing, adding massive new permutation and combinations. The claims Qentis/Marcovici make would be impossible in one language. Quentis claims to be doing this in several.

Then you’d have to ignore the fact that solely computer-generated content (i.e. content created without an actual creator) generally isn’t copyrightable. From the USPTO:

Similarly, the Office will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.

While it wouldn’t take much to skirt this in real life (the presence of an editor or someone who tweaks algorithms before generating content), in the Qentis world where millions of pieces of content are being “created” every year, it would be impossible.

Furthermore, even if Qentis could create all those works and even if they were found to be copyrightable, Qentis would still run into a different problem: under the law, if someone truly comes up with the identical works independently, there’s no infringement which would kind of break Qentis’ entire business model (were it real). Independent invention, while not allowed in patent lawsuits, is a defense against copyright infringement. As Judge Learned Hand once famously wrote: “if by some magic a man who had never known it were to compose anew Keats’s Ode on a Grecian Urn, he would be an “author,” and, if he copyrighted it, others might not copy that poem, though they might of course copy Keats’s.” In short, even if all the other impossible situations above were taken care of, others creating these works independently likely would not be infringing anyway.

But it’s all a joke… or at least, yet another art project from Marcovici. At his personal website, it’s listed along with other concepts like Bitcoin paper money, rats in a Skinner box entering trading orders, an underground package delivery system and advertising on paper money.

Marcovici’s publicity bio that looks suspiciously like a Wikipedia page notes that he also owns the Domain Developers Fund, conveniently located out of reach of US regulators in the Cayman Islands. The website seems to be dead, but his personal site gives some details as to its purpose. The language used is decidedly more flowery than informative, but it appears to be (if it actually exists) a domain squatting business. While this is listed alongside other Marcovici projects (like Qentis), this one at least appears to have some basis in reality. Marcovici’s email address (mike@qentis.com, according to the Qentis.com registry) is linked to at least 1,649 domain registries.

Interestingly, one of those is Fontsy.com (also listed on Marcovici’s website), a site that gives away “free” fonts (many of which can only be licensed by their creators), providing the following warning to those who partake of its services.

The fonts which available on this website are their authors’ property. If you want use any font from this website commercially, you should contact the author. Look at the redme-files for more informations. is there no readme-filme, open the font file. Under windows there are copyright informations.

So, Marcovici (or at least the administrators of this website) have a pretty slippery grasp on intellectual property rights, something Qentis.com definitely shares.

Qentis.com’s history as a domain dates all the way back to 2003, when it linked surfers to Marcovici’s ebay store. By 2006, it had gone dormant. From there it became a platform for pushing his book on his ebay experiences, only morphing to its current form sometime this year.

So, this is Marcovici’s stunted, but expansive, satire. A copyright-trolling automaton that will cleanse the world of creativity using brute force computing power and a team of outsourced rights enforcers. But behind all the copyright monopoly bluster, there are small hints at the message Marcovici is trying to send.

The same page where “Howard LaFarge” states that Qentis will become the “universal source of all web content,” thus “freeing” corporations from their dependence on “expensive” creators, this paragraph appears.

what is left now to creatives is not anymore the repetitive low quality text they currently produce mainly from machines for SEO but to engage in real creativity at the level where context becomes more important than words.

It’s even more explicit on a page detailing an interview with a “Russian TV station” that likely never happened.

The first way is simply to create something new, something really new, not just the remix of parts that are already there, really creativity is when people grow out of the usual stuff, we have been written books for thousands of years, produced images for thousands of years, it’s time to make use of new technologies and the combination of technologies to create content in new ways. such new ways of content, combinations of acting, sound, text, smell and and more can never be reproduced in an automated way. Yes Qentis makes it useless to continue to write average texts because they already exist writing text has become an activity like harvesting potatoes, or washing the car, jobs we want to eliminate so that we can grow over this and focus on more intellectual activities.

That’s the statement of intent. I don’t agree with all of it, especially since creativity is informed by predecessors and influences, but if Marcovici’s Qentis “project” is meant to mock SEO-friendly filler and bots that compile web detritus into ebooks, then I can get behind the concept. There’s nothing here that’s based in mathematical reality, but using hyperbolic bullshit to take an (admittedly blunted) swipe at “brute force” content generation (millions of web pages generated with all the care and creativity of “harvesting potatoes”) is a worthy windmill tilt. Unfortunately, Marcovici — with his domain-name squatting and casual use of the IP of others (at Fontsy, but likely elsewhere as well) — isn’t the best medium for the message.

Filed Under: , , ,
Companies: qentis

Rate this comment as insightful
Rate this comment as funny
You have rated this comment as insightful
You have rated this comment as funny
Flag this comment as abusive/trolling/spam
You have flagged this comment
The first word has already been claimed
The last word has already been claimed
Insightful Lightbulb icon Funny Laughing icon Abusive/trolling/spam Flag icon Insightful badge Lightbulb icon Funny badge Laughing icon Comments icon

Comments on “New 'Company' Claims It Uses Algorithms To Create Content Faster Than Creators Can, Making All Future Creations 'Infringing'”

Subscribe: RSS Leave a comment
88 Comments
Josh in CharlotteNC (profile) says:

Re: Cost Effective?

Assuming you used computers (cheaper and more efficient in every respect than monkeys), you’re going to very quickly run out of particles in the entire observable universe to store your creations. In fact, you wouldn’t even make a dent before you ran out of storage space.

A quick Google gives me a few sources that say there’s over 1 million words in the English language.

Therefore, there are more than x possible 400 word combinations. 1 million to the 400th power.

x = 1,000,000^400
or
x = 1 * 10^400,000,000

There are 10^80 particles in the observable universe (which is already unimaginably huge). Well, 10^80 if you could convert everything to hydrogen atoms. So, if you could do that, and then use each atom to store a single 400 word creation, you would have gotten 0.0000002% of the way through the possibilities.

Yes, I know the claim is nonsense, or a joke. But maths is fun.

mcinsand (profile) says:

maybe we're seeing things differently

To me, this looks like a potential copyright melt-down. Where we are now, with MPAA, RIAA, and Disney, those souls working to generate original content are brave, to say the least. What if they create something that is similar to a work that they never saw or heard. Decades ago, the probability was minimal. Now, with the Internet, the number of creators with an outlet is rising exponentially, and automation will only increase the effect. In a scenario that I fear, the little guys will be at the mercy of the big guys with deep legal pockets. However, if Qentis’ investors gain enough momentum, the big guys over here might start pushing their paid-for legislators to roll back some of the ridiculously over-reaching copyright extensions; backpeddling could become their best survival choice.

DogBreath says:

Unforeseen consequences

Or as some might say, “When keeping it real wrong, goes into outer space.”

By 2020, supposedly every possible photograph will have been created and registered by Qentis.

So the Qentis Company, by it’s own admission, will have created all the child porn that had or ever will exist.

Bye bye Qentis Company and co-conspirators. Off to prison you go. Prepare your 7th planet.

Anonymous Coward says:

“By 2020, supposedly every possible photograph will have been created and registered by Qentis. Text content generation is advancing at a faster rate…Furthermore, even if Qentis could create all those works and even if they were found to be copyrightable, Qentis would still run into a different problem: under the law, if someone truly comes up with the identical works independently, there’s no infringement which would kind of break Qentis’ entire business model (were it real).”

Well, were a company to do this, and try to claim copyright infringement even lacking any actual copying then they’d run into a big problem, because their program, by their own argument, would have just **infringed** on every possible copyright. And they would know, mathematically, that would be true, so *willful infringement* at $150,000 per instance…

antidirt (profile) says:

Re: Re:

Well, were a company to do this, and try to claim copyright infringement even lacking any actual copying then they’d run into a big problem, because their program, by their own argument, would have just infringed on every possible copyright. And they would know, mathematically, that would be true, so willful infringement at $150,000 per instance…

Ha! I was thinking the same thing. By their own logic, they’re infringing lots of works that are already under copyright.

Anonymous Coward says:

All joking aside, the number of possible variations of, say, a 256 color image 32 pixels wide by 32 pixels high would be:
1090748135619415929462984244733782862448264161996232692431832786189721331849119295216264234525201987223957291796157025273109870820177184063610979765077554799078906298842192989538609825228048205159696851613591638196771886542609324560121290553901886301017900252535799917200010079600026535836800905297805880952350501630195475653911005312364560014847426035293551245843928918752768696279344088055617515694349945406677825140814900616105920256438504578013326493565836047242407382442812245131517757519164899226365743722432277368075027627883045206501792761700945699168497257879683851737049996900961120515655050115561271491492515342105748966629547032786321505730828430221664970324396138635251626409516168005427623435996308921691446181187406395310665404885739434832877428167407495370993511868756359970390117021823616749458620969857006263612082706715408157066575137281027022310927564910276759160520878304632411049364568754920967322982459184763427383790272448438018526977764941072715611580434690827459339991961414242741410599117426060556483763756314527611362658628383368621157993638020878537675545336789915694234433955666315070087213535470255670312004130725495834508357439653828936077080978550578912967907352780054935621561090795845172954115972927479877527738560008204118558930004777748727761853813510493840581861598652211605960308356405941821189714037868726219481498727603653616298856174822413033485438785324024751419417183012281078209729303537372804574372095228703622776363945290869806258422355148507571039619387449629866808188769662815778153079393179093143648340761738581819563002994422790754955061288818308430079648693232179158765918035565216157115402992120276155607873107937477466841528362987708699450152031231862594203085693838944657061346236704234026821102958954951197087076546186622796294536451620756509351018906023773821539532776208676978589731966330308893304665169436185078350641568336944530051437491311298834367265238595404904273455928723949525227184617404367854754610474377019768025576605881038077270707717942221977090385438585844095492116099852538903974655703943973086090930596963360767529964938414598185705963754561497355827813623833288906309004288017321424808663962671333528009232758350873059614118723781422101460198615747386855096896089189180441339558524822867541113212638793675567650340362970031930023397828465318547238244232028015189689660418822976000815437610652254270163595650875433851147123214227266605403581781469090806576468950587661997186505665475715792896.

Togashi (profile) says:

Re: Re:

Now let’s have some fun with a 1920×1080 photograph in 24-bit color. That’s actually a fraction of the resolution modern cameras can record, but let’s go with it. At that resolution and color depth, the number of possible images (with 14981180 digits) is so large that it would be half again as long as the Guinness World Record holder for longest novel, at a mere 9.6 million characters spanning 3031 pages.

Shrubery says:

Re: Re: Re:

“Now let’s have some fun with a 1920×1080 photograph in 24-bit color. That’s actually a fraction of the resolution modern cameras can record, but let’s go with it. At that resolution and color depth, the number of possible images (with 14981180 digits) is so large that it would be half again as long as the Guinness World Record holder for longest novel, at a mere 9.6 million characters spanning 3031 pages.”

I got bored and worked it out as:

6.057878255625903915397895493134545504683543 × 10^14923080

Which I believe is slightly less than the claimed losses this fiscal year by the copyright industry.

Uriel-238 (profile) says:

Re: Re:

You could narrow it greatly by constraints, say images that have shapes of a significant size. I bet a neural network could cull out images that are “of something” in contrast to those that aren’t.

Still, the number would be astronomically immense (e.g. more than the photons in the universe), just not as astronomically immense.

Anonymous Coward says:

If information stored on a computer can infringe copyrights, then copyright applies to numbers, since the sum total of information on a computer is essentially a very large number. In that sense, a program that iterated through every possible number would take on the role of the famous “infinite monkeys typing on infinite typewriters”, eventually reproducing every possible work of every type of medium that can be stored as data. Moore’s law means that even if it can’t be done now, it will be doable eventually.

Of course, as you said, it wouldn’t be a valid legal claim. However, if a valid legal claim were needed to make use of the law, Veoh wouldn’t have gone bankrupt. Legal trolls only need a half-baked excuse to make a claim. An actual troll, able to employ the strategy referred to (or lampooned) by Qentis, would be able to come up with a half-baked excuse for absolutely any media whatsoever created after they started operating. In that sense, a program that purportedly generates every possible media file could be considered a “perpetual lawsuit machine”. (“Perpetual legal motion machine”?)

beltorak (profile) says:

Re: Sorry Quentis, it's not Copyrightable

which raises an interesting question; aren’t they creating a vast body of effectively public domain works? If you create something, that happened to be exactly what some big corp created before you, which also happened to be exactly something that this algorithm spit out before that, by definition it’s not copyrightable, so you are safe, right? all we need is an (incredibly) efficient indexing algorithm!

/s

Anonymous Coward says:

It costs $300 to register each copyright with the US Copyright Office, does it not? I seem to remember something like this from the porn trolling lawsuits?

I’m pretty sure you have to pay $300 every time you want to register a copyright, to get the legal benefits.

Is this company paying the US Copyright Office $300 every single time they create a new work? I guarantee you they aren’t.

DannyB (profile) says:

Re: That's not their business plan.

I think they might have found a way to cut down on the space of outputs to create.

Obviously, every possible 256 x 256 pixel image is not interesting. But many of them are.

They might have a machine learning algorithm trained to recognize noise and non-noise. Non-noise images get further sub variations created.

From the non-noise, you could use trained recognizers to classify images. Images containing certain objects (say people, animals, plants, etc) could have more variations produced.

As another example, take a set of types of things (cars, dogs, etc) that might appear in a photograph, and create ways of generating simple representations of those. On a larger scale combine various combinations of types of elements into composite images.

For sounds, generate random chord sequences according to some basic rules. Over any given set of chords, generate every possible melody as any note from first chord, followed by any note from second chord, etc. Use rules to discard obvious junk. Use machine learning to identify things that are appealing based on the training set.

All they need is to ‘create’ something similar enough to an artist in order to shake them down.

John Fenderson (profile) says:

Re: Re:

I don’t think there is provision in the law for this. I also don’t think it’s needed.

Copyright covers the expression of an idea, not the idea itself. It would be monumentally unlikely for two authors to independently come up with the same idea and express it in the nearly identical way that would be needed to be a copyright violation.

DannyB (profile) says:

A better use of Qentis technology

Dear Qentis:

I propose another important use of your valuable technology.

Generate every possible patent. A patent application doesn’t have to make sense. It just needs to be in a particular form with obscure terminology, impenetrable language and very simple vector drawings.

Please don’t generate patent applications faster than the USPTO can accept them. Just make sure that the USPTO is unable to process any other patent applications.

As a side effort you could form a betting pool on how many of your generated patent applications will be granted.

Good luck!

That One Guy (profile) says:

Copyright status irrelevant

Similarly, the Office will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.

Sounds like a slam-dunk right? Eh, not so much. Remember that in order for a DMCA claim to be legal it requires that the one filing it ‘swear under penalty of perjury’, something that as far as I know computers can’t do, and yet computer generated DMCA claims are still treated as valid.

The current legal system is set up in such a way that even if you are sure that you’re not infringing, it’s still safer to fold, rather than be forced to spend ridiculous amounts defending yourself, so while this ‘plan’ may have other problems, the legal status of the ‘copyrights’ isn’t really one of them.

Anonymous Coward says:

I think whats happening with software patents in america ,is similar
to this 1000,s of software patents that are vague and broad and pointless are granted ,
A Patent troll can buy them and use them to extract money from companys ,using the treat of high legal fees.
Most patents are granted ,
companys have a choice ,pay say 50k or go to court and pay legal fees,of a million dollars plus,
to fight and invalidate a patent .
Even microsoft has paid money on patents which were later
shown to be invalid .

DannyB (profile) says:

The importance of publishing!

Qentis needs to make all of their ‘creative’ ‘output’ available via the web. It’s not that anyone will read / view / hear it. It’s just that anybody can read / view / hear it.

It needs to be plausible that any big content, big copyright maximalist people could have possibly seen / heard / read Qentis’s ‘valuable’ work and then copied it.

Anonymous Coward says:

You wanna talk about the entertainment industry freaking out over copyright this is one fine way to set up a bogus method. No it won’t float and if it did, the first time something got close to hitting them in the pocket they would have their lobbying wagon in full swing to see the law changed.

Damn shame it isn’t a real threat. It would end a lot of the BS we see now dealing with copyright and the flagrant misuse of the court system. I would love to see this sort of scheme break copyright totally.

Andrew (profile) says:

On the other hand...

What if this were approached with the aim of producing a large number of (say) 400 word fragments and releasing them into the public domain?

Wikipedia reckons a vocabulary size of 5000 covers 95% of word use. While the combinations of that set over 400 word articles is still prohibitive, the number of grammatically feasible sentences will be much, much lower. If a suitable set of subject-specific CFGs were created (cf. SCIGen), it may be feasible to come up with a non-trivial subset of all possible articles in a particular field.

If this were possible (and I really don’t know how large the set would be) and the articles were all published online, what would the legal situation be?

Specifically, if someone (human) who subsequently (and independently) wrote an identical 400 word article, could we rely on the USCO’s comments in the monkey selfie case to negate any copyright interest the author may have (at least in the autogenerated copy)?

Get off my cyber-lawn! (profile) says:

Have you created this picture yet?

…………………./´¯/)
………………..,/¯../
………………./…./
…………./´¯/’…’/´¯¯`·¸
………./’/…/…./……./¨¯
……..(‘(…´…´…. ¯~/’…’)
……………………..’…../
……….”…………. _.·´
……………………..(
…………………………

That Anonymous Coward (profile) says:

I am willing to bet they have lots of investors.
One need only look at the returns currently available in the US for a 5 minute clip of people fucking.

Now that the Prenda financials are public record, it really is hard to pretend there isn’t money in doing it. This seems little different than the statement made by Hans in the depo that caused me chills, I paraphrase, We take these worthless copyrights and put value into them. It is like what caused the housing bubble, using crap to get cash… the trick is to get out before the bubble bursts.

As we see more of these ‘bright’ ideas enter the market, there is a chance that finally there will have to be change to copyright… lets just hope that the people talk louder than the bribes.

any moose cow word says:

A quantum computer with a sufficient number of qubits (quantum bits) could represent all works simultaneously. Once my future-self finishes the machine, I’ll have him ship it back through time to my present-self on next Thursday, when I’ll be serving Qentis a cease and desist order to halt the use of their copyright trolling machine and to destroy all infringing works.

Uriel-238 (profile) says:

Could such an idea be used in reverse?

By taking all these paragraphs and giving them a creative commons free-use, no-attribution-necessary license, you could systematically eliminate copyright since that means any given work is a quote of something that has been freed to the public.

Ergo, you wouldn’t be quoting Disney but a segment of Quentis extensive library of open-access material.

Chak-Thoom!

Uriel-238 (profile) says:

Re: Is it possible to challenge a copyright...

On the basis that it could have been designed by a machine? I’m sure that some products of human innovation (e.g. lace trim reminiscent of a sin wave) were demonstrated to be less so once we developed the technology to compute it by algorithm. How simple does an algorithm have to be before we figure that it’s obvious? How sophisticated does an algorithm have to be before it is regarded as human quality?

Coyoty (profile) says:

If a monkey, as a nonhuman, cannot legally own a copyright on a photo, then infinite monkeys in the form of a nonhuman supercomputer cannot legally own the works it generates. Likewise, the owner of the camera can’t own the copyright on the photo because he had no creative input, so the owners of the monkey machine can’t own copyrights on its intellectual-property-aping word search puzzles.

Uriel-238 (profile) says:

Re: Brian Eno might disagree.

Last I checked, Brian Eno noted that if he used a songwriting algorithm to write a song, and all he did was set a few parameters and the computer happily churned it out for him, then he can still claom copyright on the piece.

Even if he just kept hitting “randomize” until something that sounded pleasant came out, Mr. Eno could still copywrite the song.

If you disagree, you’ll have to be ready to determine how much input is required by the human element before automated content creation is copywritable.

Not that I necessarily agree with Brian. Rather, copyright is too powerful a legal device to apply to wprks that are completely human, let alone autpmated processes, but in a wprld where there were reasonable copywrite terms, such a distinction might have to be made.

GEMont (profile) says:

Might be the best thing to happen to copyright.

My hopes are that this idea of computerized, algorithmic mass auto-manufacturing something to gain copyright control over all possible iterations of that thing, and thus being able to sue anyone who ever does anything similar in the future, will force the courts and judges of the world to finally realize that copyright law is being massively abused by non-creators and needs to immediately be returned to its original state for its originally stated purpose.

Before this incredible corporations-and-lawyers-get-rich scheme started and copyright was only for the protection of creators for a very limited time, there were very few legal proceedings – compared to today – dealing with copyright infringement.

Now it is as much a daily occurrence as weather and almost NEVER has anything to do with protecting creators or their works.

Yes I realize that lawyers will need to buy smaller yachts and that politicians world wide will receive smaller graft checks thereafter, but hey, I think we can all live with that.

Add Your Comment

Your email address will not be published. Required fields are marked *

Have a Techdirt Account? Sign in now. Want one? Register here

Comment Options:

Make this the or (get credits or sign in to see balance) what's this?

What's this?

Techdirt community members with Techdirt Credits can spotlight a comment as either the "First Word" or "Last Word" on a particular comment thread. Credits can be purchased at the Techdirt Insider Shop »

Follow Techdirt

Techdirt Daily Newsletter

Techdirt Deals
Techdirt Insider Discord
The latest chatter on the Techdirt Insider Discord channel...
Loading...