Zenimax vs. Facebook Part 4: The Airing of Grievances

By Shamus
on May 23, 2017
Filed under:

Like I said last week, Zenimax seems to be suingYes, the initial trial is over, but appeals are ongoing and I expect this will drag on for ages. Oculus over the “theft” of it’s technology. This is a tricky argument because they’re accusing Oculus of stealing technology that Carmack was apparently sharing freely. You can argue that Carmack shouldn’t be sharing information against his employer’s wishes, but that makes this into a fight between Zenimax and Carmack, and Zenimax doesn’t want that fight.

Disclaimer: I am not a lawyer. This is a complicated case and I am not an expert on the law, VR, or corporate contracts. I’m working with incomplete records of complex events where there was often more than two sides to every story. I’ve done what I could to be accurate, but series is intended as opinion commentary, not authoritative historical record.

I am not a lawyer, but it seems to me that going after Carmack would mean Zenimax admitting that it was their own fault that Oculus had all this technology. They needed to push blame on Luckey’s team, because if they try to blame Carmack then the blame just boomerangs back on them. Why did you continue to share secrets you didn’t want the other party to have? Why did you continue to employ someone who was giving out your secrets? This sounds like an internal company dispute. Whatever disagreements Zenimax has with Carmack, he doesn’t have billions of dollars for them to sue him for and he doesn’t own the Rift.

They really wanted in on this VR stuff, and the best way to do that was to humor Carmack and try to entice Luckey to show up at the bargaining table. Or it was, until the moment when Carmack left.

Punishing Carmack was a dead end. So was bargaining with Luckey. And once Carmack was gone, their only move was to take Oculus to court. Carmack left for good in November 2013, and the Facebook acquisition was announced in March of 2014, just three months later.

My own unsupported conjecture:

Going by the personalities of the people involved, I’m willing to bet that Luckey, Carmack, and the rest of the Oculus team were not eager to become an arm of mega-corporation Facebook. The documents I’ve read don’t detail exactly when Zenimax began legal action against Oculus, but it was definitely before the Facebook deal. It’s entirely possible that the Zenimax was hoping to weaken Oculus through prolonged legal action (or simply scare them with threat of same) until they had no choice but to accept some sort of buyout. Instead it drove them into the arms of an even larger company.

Zenimax needed to prove that Oculus had stolen from them. This is a bit tricky. Aside from the problem that Carmack seemed to be “giving away” technology by simply collaborating with LuckeyAs allowed by the NDA, there was the additional problem that you can’t own concepts or algorithms.

But you can own source code that implements those concepts or algorithms, and so their best bet was to allege that Oculus had stolen their source code. Zenimax had several complaints against Oculus, but there are four I want to examine specifically:

Complaint #1: Non-literal Copying.

Nice laptop. I`ve decided it non-literally belongs to me.

Nice laptop. I`ve decided it non-literally belongs to me.

Zenimax accused Oculus of Non-literal copying of source code owned by Zenimax. Carmack wrote a bunch of rendering code for the Oculus while he worked at Zenimax. Later, the Oculus Devkit began using code that operated similarly. Oculus maintains they wrote the code themselves, while Zenimax claims they appropriated the code written by Carmack.

Zenimax claimed this appropriation was “non-literal copying”, which is apparently a thing?

Bob Zenimax:
Officer, you gotta help me! Shamus Young stole my laptop.

That’s pretty serious. What did it look like?

Bob Zenimax:
(Holds up laptop.)
It’s this one.

I thought you said this Shamus guy stole it. Did you recover it from him?

Bob Zenimax:
Well, he non-literally stole it.

You mean he used it without asking?

Bob Zenimax:
God no. He’s never been near the thing. But sometimes he pretends like he owns it. He makes these typing motions in the air and says he’s using my computer.

That… that’s not stealing.

Bob Zenimax:
Like I said, he non-literally stole it.

I get that the thrust of the argument is that Oculus supposedly stole the code by looking at the original and then writing their own version of it, as opposed to performing a brute-force copy & paste. But still, “Non-literal copying”? Wouldn’t that just mean “Not actually copying”? As in, the opposite of copying? What sort of linguistic shenanigans is going on here?

Consider these scenarios:

  1. You take my code and use it. (Literal copying.)
  2. You look at my code, and write your own version of it, using different variable names and formatting. (Non-literal copying.)
  3. I give you a detailed description of the mathematical operations you need to perform to complete a task. You then take that and write your own original code. (Which is allowed, in this particular case, since no patents are involved.)

The thing is, it should be impossible to differentiate #2 and #3 by simply looking at the resulting code. Ask any coder who’s gotten lost in their own codebase and they’ll tell you: Sometimes code can look superficially very similar and yet accomplish very different things. At the same time, two pieces of code can look completely different and accomplish exactly the same thing. Consider this code snippet:

int add_two_numbers (int a, int b) {
  return (a+b);

And then let’s say this code shows up at another company:

int AddIntegers (int iValue1, int iValue2)
      int iResult;
      iResult = iValue1;
      iResult += iValue2; //WARNING: No overflow checking!
      return iResult;

These two bits of code accomplish the exact same thing. They look different. They’re formatted differently. They follow different coding conventions. But they accomplish the exact same thing: They both take a pair of numbers and return the sum. How can you tell if I wrote the second from scratch or if I used the first example as a guide? You can’t. If you want to claim that the second represents “non-literal copying”, then how could I possibly have written it to be MORE unlike the original? The function does exactly what it needs to do, and if it did anything else then the code wouldn’t work. By the logic of non-literal copying, I can never again write a function to add two numbers together, because it’s impossible to do it without it looking like I non-literally copied it from whoever paid me to write the first one.

I’m assuming it wouldn’t hold up in court on an example this small and elementary, but since complex programs are made up of simple operations, the problems remain even if you scale it up to something useful!

There are a lot of bullshit patents out there. But at least in those cases someone had to go to the time and trouble of applying for the patent! If you take non-literal copying to its logical extreme, it gives you patent-esque powers, applied to all of your code, without the idea ever needing to pass muster at the patent office. This is silly. In computer science there is often only one right way to interface with a library, simulate a process, or render a polygon. If we judge code by what it does rather than what it looks like, then nobody is safe from accusations of non-literal copying.

I’ll come back to non-literal copying when we get to the verdict. For now let’s just move on to…

Complaint #2: Literal Copying

I`d be able to make programming mistakes so much faster and more efficiently with this keyboard.

I`d be able to make programming mistakes so much faster and more efficiently with this keyboard.

On top of allegations of Non-literal copying, Zenimax was also alleging that Oculus literally copied their code. Like, the real sort of literal. Actually literally.

It seems as though this complaint would be mutually exclusive with the “non-literal” one. It’s like claiming that someone tricked me into handing over my wallet, and also that they pick-pocketed it from me. These two things can’t both be true. Apparently this sort of thing is allowed in a lawsuit? Apparently so. It sounds strange to a layperson like myself.

We can argue about who invented what or who should get credit for various innovations leading up to the original Doom 3 demo of the Rift, but one fact that neither party denies is that Carmack wrote the original rendering code to make it work. From here one of two things is possible:

  1. After he worked out the required process, Carmack simply describedOr perhaps even wrote a spec. the process required, and from there the Oculus team was able to write the code themselves. As I said above, this wouldn’t be copying – literal or otherwise.
  2. Carmack just handed over the source code of the project to Oculus.

What does the evidence say? It’s… complicated.

It’s true that Luckey signed an NDA saying he wouldn’t use the Zenimax technology to compete with them. But it’s also true that you can’t own, patent, or copyright, or control an algorithm, and a lot of the breakthroughs in VR (like FoV distortion and chromatic aberration) are arguably algorithms. But again, this doesn’t matter if Zenimax can prove that Oculus appropriated their source code – literally or otherwise.

At one point Oculus VR co-founder Brendan Iribe requested (in writing!) a license for the “source code shared by Carmack”. That seems pretty damning. Why would Iribe request a license for code they didn’t need? That suggests that at some point Oculus was using Carmack’s code. However, we’re missing a lot of context surrounding this conversation.

For example: Perhaps Carmack shared the code with Luckey early in the project. (Which would have been permitted by the NDA.) Then Brendan Iribe – upon hearing that the code exists but not having seen it himself – requested licensing the code so they don’t have to write their own version.

We don’t know what the answer to his inquiry was. Maybe Zenimax said “No”. Maybe Zenimax asked more than Oculus could payI’m sure the “price” would have been a controlling stake in Oculus, and not money. Zenimax REALLY wanted the VR tech and didn’t care about the stupid Oculus piggy bank and its pathetic little $2 million in Kickstarter money.. In any case, once the license deal fell through then Oculus would have written the code themselves.

All of this is just to point out that even though Iribe’s request sounds damning, there are a lot of possible explanations that don’t prove guilt. It all depends on the context and timing of the discussion. (Which has so far not been made public.)

The Hard Drive Problem

Protip: Grabbing onto a hard drive plate and moving it back and forth with your hand doesn`t sound anything like a record turntable.

Protip: Grabbing onto a hard drive plate and moving it back and forth with your hand doesn`t sound anything like a record turntable.

The other major piece of gossip floating around in the wild is the detail that – hoping to prove that Carmack stole source code from ZenimaxA funny idea in itself, since he’s the one that wrote it in the first place. But that’s how copyright works. – Zenimax demanded that Carmack hand over his hard drive. When they got it, they discovered that 90% of the data had been wiped.

Note that a data forensics expert said that it had been wiped. If Carmack had just dragged a bunch of crap into the recycle bin, then they would have been able to recover most of it. (Our data forensics technology is pretty good!) If the expert could only say that something had been wiped and not what had been wiped, it suggests that the data was very thoroughly and deliberately expunged. This probably wasn’t a casual cleanup job.

The expert went on to say that the wipe happened after Carmack got notice of the lawsuit.

Keep in mind this is presumably Carmack’s personal machine. That’s likely to have lots of private information, browser history, and other sensitive stuff on it. Imagine if someone was taking you to court. Would you want them to know your browser history? Would you want them to have access to all those website auto-logins you’ve got saved? Do you want them to know what programs you’ve got installed, what games you’ve played recently, and whoever might be on your contact lists? If I was being sued like this, my first thought would be, “Oh crap. They’ll probably demand I hand over this machine at some point. It’s full of company secrets and possibly even personal information not pertinent to the case. I should just clean it off so that stuff doesn’t wind up in the wrong hands.”

But you’re not supposed to do this!

When you’re being sued – or even when you THINK you’re about to be sued! – you’re not supposed to delete things. Even if it’s standard procedure at your company to shred documents after 30 days or purge emails after 90, you’re supposed to stop doing these things when you’re under threat of a lawsuit.

It’s possible that Carmack didn’t realize this. It’s also possible that this happened very close to when he was leaving Zenimax and would have been obligated (under normal circumstances) to purge the machine of company information.

It’s possible this was an innocent act and it’s possible this was willful destruction of evidence. Usually in cases like this the judge can handle this a couple of different ways:

  1. The judge can give the jury an adverse-inference instruction. This tells them, “You may (but are not required to) assume that the deleted evidence contains information that is unfavorable to the defendant.”
  2. The judge can give the jury a mandatory adverse-inference instruction. This tells them, “You must assume that the deleted evidence contains information that is unfavorable to the defendant.”

The judge went with #1, which suggests that Carmack’s action was viewed as a mistake, but not a deliberate and willful attempt to destroy evidence.

I’m not saying Carmack is definitely innocent of the charge of destroying evidence. I’m just saying that once you start looking into this stuff there’s a lot of nuance. Before I began reading up on this case it seemed like the Zenimax claims were preposterous. But now most of them seem basically reasonable. At the same time, the accusations over deleted hard drives sound a lot less scandalous.

In any case, I think the Zenimax case would be a lot stronger if they hadn’t pushed the “non-literal copying” stuff. Their silly claims and their dodgy expert make this look less like an honest inquiry and more like them being willing to throw everything at the wall and hope that something sticks.

Complaint #3: False Designation

The Oculus Kickstarter pitch, which shows a Zenimax employee showing off the Rift using a Zenimax game at a Zenimax trade show booth.

The Oculus Kickstarter pitch, which shows a Zenimax employee showing off the Rift using a Zenimax game at a Zenimax trade show booth.

Doom 3 was featured prominently during the E3 Demo and the Kickstarter campaign. One of the Kickstarter reward tiers included a version of Doom 3 designed to work with the Rift. Zenimax claims this made it look like they were directly involved in all of this when they weren’t. All of this was done without any official blessing from Zenimax, who owns the Doom franchise.

If I did a big “Coca-Cola Giveaway!” and covered this whole blog in Coke branding, then some people might assume Coke was supporting or sponsoring my blog when in fact they had nothing to do with it. In this case I would be in danger of being accused of “False Designation”.

You can imagine how frustrating this was for Zenimax: They can’t get Luckey to cut any kind of deal with them, but then Luckey goes around using Zenimax IP as promotional material. The fact that this didn’t immediately trigger legal action from the persnickety Zenimax legal team makes me think Zenimax was still trying to entice Luckey to sit down at the bargaining table and sell them a stake in his company.

Complaint #4: Violation of the NDA

Contracts might be boring to read, but there`s nothing quite like a massive lawsuit to make them seem exciting!

Contracts might be boring to read, but there`s nothing quite like a massive lawsuit to make them seem exciting!

Like I said earlier in this series: Luckey wasn’t allowed to use “Zenimax secrets” to compete with Zenimax. But his entire company was based on a product that depended on those secrets (or some of them) to work. I think this one depends on two ideas:

  1. Are the “secrets” that Carmack shared really things that could reasonably be said to violate the NDA? Zenimax worded the thing very broadly, but that doesn’t necessarily mean it can be applied broadly. Just because Carmack tells me how to calculate surface normals while I’m under NDA, it doesn’t mean I’m automatically violating the NDA when I use that knowledge to do something else. Is this something that’s considered an algorithm? Is it something I already knew before Carmack told me? Is it something that could be considered obvious or common knowledge? Carmack blabbed a lot of VR techniques as part of the normal “Thinking out loud” stuff he does in public appearances. Are those facts and ideas now free of the NDA because they’ve been openly shared in public?
  2. Is social-media company Facebook really “competing” with videogame company Zenimax by developing and marketing the Rift?

I don’t have the knowledge or expertise to know the answer to any of these questions, and in any case it probably comes down to what the jury thinks. Like the rest of this case, it’s one of those things that seems obvious at first but then gets really complicated when you try to zoom in on the details.

Next time I’m going to wrap this up by looking at the verdict.

Enjoyed this post? Please share!


[1] Yes, the initial trial is over, but appeals are ongoing and I expect this will drag on for ages.

[2] As allowed by the NDA

[3] Or perhaps even wrote a spec.

[4] I’m sure the “price” would have been a controlling stake in Oculus, and not money. Zenimax REALLY wanted the VR tech and didn’t care about the stupid Oculus piggy bank and its pathetic little $2 million in Kickstarter money.

[5] A funny idea in itself, since he’s the one that wrote it in the first place. But that’s how copyright works.

2020201272 comments? This post wasn't even all that interesting.

From the Archives:

  1. Matt Downie says:

    Are literal and non-literal copying really mutually exclusive? They could have copied some code and non-literally copied other code. Maybe the guy in charge of disguising the stolen code forgot to modify one or two of the files.

    • methermeneus says:

      Or, you know, Carmack gave Luckey some source code and some algorithms, and ZeniMax is claiming the algorithms are nonliterally copied. No intent involved. Well, except that Carmack has been doing this for long enough that he really should know better. His reputation may make him essentially immune to being sued for sharing code, but he should know that his contract says he’s not supposed to and that sharing with a commercial entity could have consequences for that entity. I think his excitement for the tech may have made him short-sighted about that.

      Their silly claims and their dodgy expert make this look less like an honest inquiry and more like them being willing to throw everything at the wall and hope that something sticks.

      To be fair, this is actually a tried-and-true legal tactic. It’s encouraged because a truly aggrieved party has a better chance of getting a judge and/or jury to agree with them on some point if they make a lot of points. In practice, it often ends up looking kinda silly, like when a drunk driver hits a telephone pole and sues the town for the road surface not being perfect, the vehicle manufacturer for not providing sufficient injury protection, and Verizon for putting that pole in the way. Not kidding: my mom used to work for a lawyer who represented Verizon locally, and this exact thing happened on a monthly basis.

    • DivFord says:

      I think it was more a chronological distinction. They literally copied the code (i.e. were given it) before they realized they weren’t going to get a license for it, at which point they non-literally copied it, presumably because they were trying to avoid getting sued for literal copying.

      From the point of view of anyone who understands software development, they took reasonable steps to not break the law after Carmack put them in an awkward position. Zenimax’s case seems to hinge on the fact that the judge, jury, etc. probably don’t understand it.

  2. Syal says:

    The example of non-literal copying I can think of is (however you’d code it,) “return a+b, do not return 1000a”, or a similar bit of superfluous tack-on. Also I think you’ve said people can write comments in there for clarity? Comments in the code could be reworded in a way the code itself can’t.

    It’s like claiming that someone tricked me into handing over my wallet, and also that they pick-pocketed it from me. These two things can’t both be true.

    Do… do you only carry one wallet? You know what they say about putting all your eggs in one basket!

    Are those facts and ideas now free of the NDA because they’ve been openly shared in public?

    This one’s a yes; you can’t say public information is confidential. I would assume the defense would have to point to the public speech and say “we got it from here”, though.

    • Sabrdance (MatthewH) says:

      It sounds like it’s in the same category as George Harrison’s “Unconscious Plagiarism” of “He’s So Fine” in “My Sweet Lord.”

      • Echo Tango says:

        Am I the only person who can’t hear the allegedly obvious similarity / same-ness between those two songs? The tempo is definitely different, the lyrics are obviously different, the sound of the individual instruments seems different[1], and one is sung by one person whereas the other is sung by four people in harmony. The only thing left that I can think of, is the actual progression of notes being played in the melody, but because of all of those other differences, I can’t actually tell that they’re the same notes. i.e. The best I can guess, is that My Sweet Lord is a slowed-down version of the same notes, but I definitely cannot hear it.

        [1] To my not-very-musical ears, My Sweet Lord has a very acoustic, round, folksy sound to the guitar, and He’s So Fine has a very blues-y, snappy, twang-y sound to the guitar and other instruments.

    • King Marth says:

      Alternatively, you were a victim of the Cuftbertian tactic of being tricked into handing over your wallet which is then pickpocketed back into your pocket. Anyone can verify that this happened because your wallet is in your pocket!

  3. Tizzy says:

    Related somewhat: a Supreme Court verdict yesterday disallows picking your favorite court to enforce your troll patents. Looks like certain Texas courts will get a lot less busy soon.

    • Echo Tango says:

      That seems pretty reasonable to me; Patents are (broadly speaking) something that needs to be enforced across an entire country, and between countries, because the businesses who are protected by them do business in those same large areas.

  4. Bubble181 says:

    1 Non-literal coding is more or less akin to plagiarism. Yes, you can use another work as a source. No, you can’t just re-word some sentences and invert some paragraphs and claim it’s an original work without sourcing it. This can be far more serious than literal copying: it’s not because the phrase “Harry took a big gulp from his drink” appears in a HP novel and also in a Harry Styles biography, that one’s cribbing from the other. However, my wholly original work “Parry Hotter and the Stone of the Sorcerer” might raise some eyebrows, despite having literally 0 exactly-copied phrases from HP. Sure, making slightly-different code to accomplish the same thing isn’t necessarily a problem…But how did you know/realize that’s what you needed to do?

    2. Literal copying and non-literal copying are mutually exclusive *if* they pertain to the same thing. I’m perfectly capable (but don’t worry, I won’t) of stealing your wallet *and* writing a great big Mass Effect write-up with all of your ideas and thoughts.

    3. This is a hard one. Did they clearly state Zenimax wasn’t sponsoring? There’s more and more of a push towards big companies to check where and how they advertise – without meaning to turn this political : some big companies have withdrawn commercials from pages that might incite hatred or advocate violence, since their ads appearing next to the content might imply they support that content. If I’m doing a big Coke rally in support of….handicapped puppies, I doubt Coke will mind if they get associated with it.If I did a big Coke rally in support of seel clubbing, while I don’t know Coke’s point of view on that, I darn well better put up in big letters “NOT OFFICIALLY ENDORSED BY THE COCA COLA COMPANY”, and even so they might still ask me to take it all down.

    4. This is one of those things… Inventors and such get caught out in this on occasion. Anything you think up or develop while employed by a company is quite likely to be property of that company, as long as it’s in any way related to your job, even if invented off company time and with your own means. Lawyering sucks. I’m more or less inclined to believe Carmack probably *did* break the NDA, even if it was only about his own thoughts. Which gets pretty weird, admittedly.

    • Steve C says:

      Plagiarism is not illegal. It’s highly frowned upon. It will likely get you kicked out of academic circles and fired (with cause) from a job. It is still not illegal. It is not criminal and nor is it protected under tort law. Non-literal coding is just nonsense.

      • Viktor says:

        Why doesn’t it fall under copyright, though? Is there a significant difference between song lyrics, poetry, and a speech?

        • Steve C says:

          It can, and does fall under copyright. But if it’s not copyright, (and not fraud) then the fact it was plagiarism doesn’t matter to a court. Maps are the classic case. Maps are just data and cannot be copyrighted. The *expression* of the information on map is protected. This is why fake towns etc are put into maps. Someone publishing a map can use other maps to build their map (aka plagiarize it). Soon as they put in someone else’s fake town then they’ve copied the expression of that map and it’s copyright infringement.

          With code, it comes down to the expression of that code. You cannot copyright data. You can copyright the expression of how that data is arrived at. You can copyright how a function is written. How a program is structured etc. What it actually does in terms of functions- that you cannot copyright.

          This whole non-literal copying of code cannot win. It might though. If it does, it creates a new precedent that expands copyright.

          • sheenyglass says:

            Don’t look now, but it’s been established as part of copyright law for quite a while – https://en.wikipedia.org/wiki/Substantial_similarity

            I think the non-literal vs. literal copying divide is just a different way to articulate the distinction between copying exactly and making something so similar as to be effectively comparable. If I were to write novels about Larry Porter, the magical boy, in his seven years at a magician academy suspiciously similar to the educational institution that must not be named, it wouldn’t be an exact copy, but JK Rowling may have a lawsuit. Software is weird in that it is a kind of hybrid between technical invention (patent subject matter) and creative expression (copyright subject matter), but similar principles still apply for those aspects of code which can be subject to copyright.

            But, yeah, plagiarism is an academic concept regarding proper attribution/citation of other’s work, not a legal concept. You could infringe copyright without plagiarizing as long as you make clear that what you are copying was not written by you. Every time someone refers to copyright infringement as plagiarism, a kitten fails the bar exam.

            • Steve C says:

              That’s what I was referring to when I was talking about “expression”. What you are talking about is a test that occurs *after* copying has been established as fact to determine if it is materially significant. Code is different. It’s math. 2+2=4 but more complex. If you had a math proof that was 100 pages long then it should look substantially similar to someone else math proof. It’s the same terms. Same technically notation. Same end result. That doesn’t mean it was copied.

              It’s like code that displays a window on a screen. It’s logically going to be substantially similar to different code that displays a window on a screen. It’s math doing a function. Like I said, this whole non-literal copying of code cannot win. If it does, it creates a new precedent that expands copyright.

              • Steve C says:

                BTW @Shamus: What “Non-literal copying” is referring to in the case is explained in sheenyglass’s link.

                And you’re right. It is crazy talk when applied to source code, data or any other work that is trying to accomplish a task. It does make sense when referring to any other art that exists for artistic purposely only (novels, paintings, photographs, music etc). Which also includes the output of a video game.

      • Kylroy says:

        It’s not a criminal matter, but the moment you start making money off the plagiarized work I’m pretty sure it can become a civil matter.

        • Alex Broadhead says:

          You don’t even necessarily have to make any money to get sued. The author of the original work can sue simply to protect their copyright. For reference, see John Oswald’s ‘Plunderphonics’ CD, and the resulting lawsuits.

        • Echo Tango says:

          You don’t need to be making money to be violating copyright, however. A good example would be for a YouTube video by Brady Haran. He makes some amount of money[1] off of the ads that play when his videos are played. If I upload my own copy of his video on ransomware computer viruses, I’m causing harm to the creator, because the ad-revenue would go to me , because people aren’t watching his video on YouTube. That by itself is causing harm, without me even editing the video to have my name instead of his, or a link to donate me money, etc.

          [1] I can’t actually remember from any podcasts he’s done, if he actually gets a lot of money from this. For the sake of argument, let’s assume he does.

          • Echo Tango says:

            Ugh. I mangled my own argument by being half-asleep when writing it. Assume that the copying-person in the example is not making money off of their copy of the original work. It still can cause money to not go to the original author. Even if they don’t cause a loss of money, they could also do non-monetary harm to the original auther.

      • Bubble181 says:

        Depends on where. It is very much illegal as a form of fraud over here. I should know, one of my former professors went to jail over it.

      • Parkhorse says:

        and fired (with cause) from a job.

        That is heavily dependent on what your job is. Journalism or speech writing? Yeah. But in my job (Procurement, in the defense sector), plagiarism is a way of life. Do we have a chunk of language that was previously approved by government auditors, which fits a given situation? Yes? You better believe we’re using it again. Writing anything original, instead of copying existing things that worked and modifying them to fit the current situation, is discouraged. It increases risk.

        • Alex Broadhead says:

          Is some individual copyrighting the ‘chunks of language’? Seems unlikely. Is you employer or the government copyrighting them? If so, you probably have permission, which makes it ‘not plagiarism’. Or possibly this is a ‘fair use’ exception – I don’t know.

          • Steve C says:

            Works created by the US government are explicitly public domain on creation and not protected by copyright. It applies to work for hire for the US government too.

            That said, I have seen claims of copyright (like on websites) by various levels of government. I’m unaware of any situation where the US government pressed their claim and won. I’m not even aware of any time they’ve pressed their claim and lost. (If anyone knows of a case, I’d be interested in reading it.)

            • Decius says:

              Many government websites liscense third-party software to operate.

              Selling SaaS to the government does not make your product public domain.

              • Echo Tango says:

                The SaaS product / software running the website wouldn’t be public domain, but the content provided by the government would.

              • Steve C says:

                I’m not referring to things being sold to the government. I’m referring to someone including at the bottom of a webpage something like “©2017- City of Faketown, USA. All right’s reserved.” The US government is not legally allowed to hold back copyrights for themselves like that.

                I don’t think this is politics but one site I know had language a few years ago to this effect. Today I just checked and it now says: “Pursuant to federal law, government-produced materials appearing on this site are not copyright protected. The United States Government may receive and hold copyrights transferred to it by assignment, bequest, or otherwise.” Which is now correct. I’m guessing it was corrected by someone more versed in the law than the IT guy who wrote the page. (I’m not going to say which page, but it is a well known important official government page that you can’t have mistakes like that on.)

          • Decius says:

            Yes. Someone copyrights every novel work they created.

            Normally that’s trivial, since copyright can’t be enforced on lots of things.

            But no action beyond creating a thing is neccessary to copyright it-only to register the copyright.

    • Victor says:

      Regarding item 4:

      I was going to post something similar. I have been to interviews where, as part of the application process, you are informed ahead of time that, should you be hired, any new invention that is in any way related to your work will be considered property of the company, and you automatically waive any rights to the discovery in favor of the company. This is so you, as an applicant, understand what you are getting into ahead of time. From what I understand, this practice is quite common, and I would not at all be surprised if it applied to the Carmack / Zenimax situation.

      • Echo Tango says:

        It’s pretty common; The general phrasing for my own employment is that it belongs to the company if any one or more of these are true:
        – it directly relates to the business the company is doing[1]
        – company hardware or software licences were used
        – it was performed on company time

        I’ve read about software companies that have more far-reaching contracts, where all software you produce while employed by them belongs to that company. Where I work however, you can make your own software and even sell it, as long as you’re not violating any of the above rules. i.e. Use your own code-editors and software, on your own laptop, away from the office, during hours / days when you’re not supposed to be at work, for business in another field. :)

        [1] You also can’t compete with the company, but that has less to do with ownership of things produced, and more to do with using inside knowledge to gain an unfair advantage.

  5. Da Mage says:

    The fact that this didn’t immediately trigger legal action from the persnickety Zenimax legal team makes me think Zenimax was still trying to entice Luckey to sit down at the bargaining table and sell them a stake in his company.

    Or they could have been just waiting to let Lucky dig himself an even deeper hole. Get him to be using Carmack/ZeniMax stuff enough, that when it comes to bargaining they could hold that over them. Lawyers aren’t above playing the long game over stuff like that.

  6. “By the logic of non-literal copying, I can never again write a function to add two numbers together, because it’s impossible to do it without it looking like I non-literally copied it from whoever paid me to write the first one”

    Others here have called this plagiarism, I think it’s closer to the birthday paradox.

    Eventually someone will code the same code, eventually someone will compose the same sequence of note, the same sentence of text.

    • Bubble181 says:

      And, as with plagiarism, it’s usually not a problem….IF you did so without knowledge of the original.

    • Snappy Steven says:

      This *would* be a huge logical issue for copyright law…if not for the fact that such technically neccessary solutions *aren’t* protected by it in the first place.

      Simply put, if there is only one (or only a select few) ways of accomplishing something in an economic fashion, you can’t claim the exclusive right to profit from everyone doing so. Not even if you’re the inventor and this is a revolutionary idea.

      Of course, in practice, it would be virtually impossible to argue that a new kind of idea is the only viable solution to any given problem. This kind of exception is limited to the building blocks of what people usually think of as new inventions or the like.

  7. Daemian Lucifer says:

    Non-literal Copying.

    I get the gist of it,but it still seems like a bass ackwards thing to me.The problem isnt what you wrote,but rather that you had unauthorized access to company secrets.It really shouldnt matter if you used said access to simply read and then try to recreate the stuff,if you actually copied it,or if you didnt do a thing with it.

    Apparently this sort of thing is allowed in a lawsuit? Apparently so. It sounds strange to a layperson like myself.

    It makes sense for a courtroom though.For example,if you get into a bar fight with someone,you can get accused for a plethora of things that are all various degrees of the same thing:Assault,assault with a deadly weapon,attempted homicide,etc.If I understand this correctly,you get accused of all of those,and during the trial its determined which of those accusations,if any,actually applies to what you did.

    • Matt Downie says:

      Possibly pedantic, but I don’t see that having “unauthorized access to company secrets,” could be considered a crime in itself. If you did something illegal to gain access to those secrets, that’s a crime. If you did something naughty with those secrets (e.g., made an exact copy of the project they were working on and released it first), that could be a crime. But if John Carmack emailed all his company’s secrets to me, that shouldn’t automatically make me a criminal.

  8. Alex Broadhead says:

    I realize it’s largely irrelevant to the points that Shamus is making here, but:

    you can’t own concepts or algorithms

    while perhaps literally true (at least for algorithms), has in no way prevented the patenting of all sorts of concepts and algorithms. (Arguably, most patents are precisely for concepts, and require no working example.)

    For example, MPEG audio (and video) codecs require all sorts of licensing (or did until last month) – even when implemented entirely in software! As someone who once worked with such software, I’ve looked at some of the patents, and, while many are ridiculous (Windowing of signals? Really?) some are not, and further, all of them are bundled together and fiercely defended by very deep-pocketed corporations. Another example: FM synthesis of audio signals was successfully patented and defended by Yamaha for many years, despite the existence of prior art, and the fact that it is essentially a DSP algorithm.

    To pull this back into context, it sounds like several of the innovations that Carmack made could have been patented. I would guess that Zenimax might have tried to do so, had Carmack been keeping them more in the loop, but it sounds like they were giving him a long leash, and probably not aware that what he was doing had any real potential until pretty late in the project?

    • Echo Tango says:

      The patentability of algorithms varies from country to country. In the US, you apparently can patent algorithms. (second, non-checkmarked answer)

      That being said, I don’t know if this discussion can continue. The site has a no-politics policy, but this blog post is specifically talking about a lawsuit and patents. Would @Shamus care to elaborate on this? It seems like this blog post is soliciting comments that violate policy.

      • Mousazz says:

        Good question. At what point does is a topic inherently controversial enough to become ineligible for rules of cautious discussion?

        If Shamus made a deep, thorough analysis of the two main parties of the U.S.A., I’m pretty sure It’d be silly and hypocritical of him to try to enforce his No-Politics rule (at least without disabling the comments entirely, which would open up an entirely different can of worms).

        Are law, court, lawsuits and patents part of the topics too dangerous to be discussed? Because if so, I’ve been surprised just how civil the comment section was (as civil as usual, at least). This is all while Shamus is spouting ardent ridicule of concepts like “Non-literal copying”, which could very easily ignite into a flame war.

        • Viktor says:

          Is there anything that isn’t political, though? The existence of trans people is considered political. Scientific facts are political. Cancelling a TV show is political. Politics touches everything, how is any discussion supposed to avoid it?

          • JakeyKakey says:

            The fact everything can be political is why the rule exists. Some people really cannot help themselves.

            As for avoiding it, it’s a bit like that quote about pornography – I can’t define it, but I know when I see it.

            99% of the time you and everybody else blatantly knows what you’re doing, why you’re doing it and that you should really knock it off.

      • Echo Tango says:

        Also, I can’t seem to find a link to a page on this website, for the no-politics rule. The closest I can find is this blog post when I search with Google. Perhaps there should be a link on the main part of the website somewhere? I can see a link to the forums and the about-the-author page, but not this rule. Or have a link on this box where you add comments, that brings you to the page explaining the rules of the website? Right now, it’s only got the text, “Thanks for joining the discussion. Be nice, don’t post angry, and enjoy yourself. This is supposed to be fun.”.

  9. Alex Broadhead says:

    On copyright, this has always been a source of confusion for me. In the context of working on MPEG audio (see above), I was provided with the (copyrighted) MPEG committee (example) code, and told by my employers to produce something free of legal entanglements (patent and copyright). I protested that I was not a one-man R&D group, and that there was no way in hell that I was going to be able to do original research that would circumvent the (very basic) patents.

    On the copyright side, the CTO claimed that ‘changing 10% of the lines’ was sufficient to remove copyright, which seemed unlikely to me, and implies that I could just run a prettifier on the sources and maybe do some find & replace and be in the clear. Of course, by the time I was done with the code, every line had been touched, whole files had been removed and others added, sections had been optimized, bugs had been fixed, names had been changed to protect the innocent, etc. But it was still (I would argue), a non-literal copy of the original, as it still, necessarily, traversed the same algorithms and en- or decoded an MPEG audio file. Even if I had simply read the spec and started from scratch, one could still argue that I’d made a copy. Do any of them get around copyright? I have no idea.

    (To be clear, I didn’t actually _need_ to get around copyright in this case, as the licensing that you had to go through to use MPEG patents presumably also allowed you free use of their example code. I’ve worked with other sources, though, where the question was not moot.)

    • Richard says:

      Depending on jurisdiction, this may have resulted in a new work that was not protected by the original copyright. (Lawyers make a lot of money from drawing the line between “copy”, “derivative” and “inspired by”.)

      It would still be covered by the patent because the patent is on the algorithm, and you can’t change that and have it still work!

      • Echo Tango says:

        Hey, we should pause this comment thread, at least temporarily until @Shamus says something. See my comment up one; We might be violating the no-politics policy.

        • Shamus says:

          So far we’re good.

          I know the line can get blurry. I THINK we’re safe here because most of this is descriptive (you’re supposed to cross in the crosswalk) rather than proscriptive (crosswalks were invented by the extremist middle and should be abolished!) so I don’t think this will lead to flamewar.

          My main reason for prohibiting politics is that it starts arguments that:

          1) Make people very angry.
          2) Are debates we’ve all read dozens of times before.
          3) Aren’t going to inform / enlighten anyone.

          I don’t think those are a risk when talking about this stuff. If it turns out I’m wrong, then I’ll close the thread.

          Thanks for being thoughtful.

    • Decius says:

      What you made is probably classified as a derivative work.

      It is not legal to claim sole ownership of a derivative work without permission from the copyright owner of the original.

  10. Paul Spooner says:

    Yikes! What happened to Jon Carmack’s face in that screenshot? Looks like the artists ran his photorealistic physically accurate ray-traced visage through a filter or two. It’s always the purists who suffer.

  11. Mousazz says:

    From the link explaining “non-literal copying”:

    For example, with a work of literature, perhaps in the form of an historical novel, the literal element comprises the words, sentences and paragraphs as expressed in print, while the non-literal elements, in which case they may have their own copyright independent of the copyright subsisting in the finished novel.

    I’m having difficulty parsing this sentence. While the non-literal elements what? What do they do/comprise? Did the writer forget to write a part of the sentence?

  12. Sven says:

    The whole “you stole code from us because it looks similar” argument is, unfortunately, taken quite seriously.

    I work for Microsoft, on the Windows Subsystem for Linux. In short, our job is to emulate the Linux kernel on top of NT, with some added sauce to make things nicer for Windows users.

    Despite the fact that the Linux kernel is open source, it’s a problem for us because it uses the GPL license. If someone could claim that Microsoft included GPL code in the Windows kernel, it would be a big legal problem for us. As a result, we can’t look at the Linux kernel, ever. We can look at the documentation. We can look at user-mode source code (it’s amazing how much you can infer about the kernel’s behavior by looking at glibc). We can write tests to see what the actual behavior is. But we can never look at the actual Linux kernel source.

    All that, just so that if someone thinks our code is similar (which, let’s face it, it’s likely to be in some cases, disregarding coding conventions and such), we can truthfully claim that’s a coincidence and not because we “non-literally copied” the GPL Linux kernel code.

    • Alex Broadhead says:

      Yep. I nearly set off a gigantic shitstorm when I brought in part of my ancient Master’s thesis that included MPEG committee code to ask one of the in house experts (at Microsoft) about. He actually worked on the committee code itself under a previous employer, but both that employer and Microsoft are _very_ big on building ‘Chinese walls’ around anyone who develops critical IP and _any_ exposure to outside source code.

    • Echo Tango says:

      I presume that all of the user-made source code which you look at is non-GPL?

    • Matt Downie says:

      Seems like it would be kind of hard to prove whether you looked at it or not…

      • Viktor says:

        That’s why it’s company policy. MS took ordered their employees not to look at the code and took reasonable steps to ensure they didn’t. If an employee looked at the code anyways, that’s in violation of policy, they will be fired, and during any ensuing lawsuits MS can say “we didn’t use GPL, an employee did so without our knowledge, sue them not us.” It’s not a perfect defense, but MS has really good lawyers so it should be good enough.

    • Philadelphus says:

      So your job is basically…writing inverse WINE? That’s fascinating. I had no idea such a project existed.

  13. 4th Dimension says:

    You can’t own an algorithm?!? That sounds strange to me. I would expect if some company developed an algorithm that does some job 50% faster that this specific way of doing it would become their intellectual property.

    I guess if somebody independently also then stumbled on the same or similar algorithm, that would be fine, but I would still expect for the lawyers to be involved if somebody figured out both softwares use the same thing.

    I remember watching a video about IBM PC. The reason PC took off as we should know was that it was composed of basically off the shelf parts with known specifications which made it possible for third party manufacturers to produce hardware for the system (you could also buy the OS from Microsoft). The only part of the PC architecture that wasn’t known or accessible to somebody that wanted to make a PC knockoff was BIOS, the piece of software tying the whole thing together which was written by IBM and simply copying IBM bios into other chips would be a violation of various laws.

    How people got around it was a two stage process. One person would inspect the code contained on the BIOS chip, and using this knowledge he would write a SPECIFICATION of what BIOS needs to do, what functions it needs and such.
    At this point this person was then for all intents and purposes BIOS related considered CONTAMINATED.
    So another person or a team was then given the SPECIFICATION and using that they would write their own BIOS for their chips.

    So obviously having seen the code and what it does and then making your own implementation was considered a major NO-NO back then.

    • Richard says:

      Mathematical theories are not patentable in any jurisdiction, for the same general reasons that quantum field theory isn’t.

      – They are discovered, not invented.

      – They can only be proven* or disproven by large-scale peer review.
      It is only science if everyone can attempt to disprove your work.
      Otherwise someone could state that the sky was pink and deny anyone permission to test it.

      Algorithms are pure mathematics, and thus cannot be patented.
      – It’s not a coincidence that Turing was a mathematician.

      Unfortunately some jurisdictions have a thoroughly broken and dysfunctional patent system that is no longer fit for purpose. This is very bad for everyone.

      However, any given implementation is of course subject to copyright and cannot be copied without the copyright holder’s permission.

      In many cases, the implementation is highly constrained by the hardware and software interfaces – a BIOS must do what a BIOS does – and so there aren’t many ways to write that.
      “Clean Room” implementations are a legal framework created to defend against claims of copyright infringement – “Team B did not copy any of the implementation of your BIOS, they never saw it”.

      * Mathematics is the only scientific field where theories can actually be “proven”. In all others “proven” means “many different experiments have attempted to disprove it and have failed to find any problems”.

      • Decius says:

        Can you describe the difference between a mathematical algorithm and a program that plays a video of The Phantom Menace when run on specified hardware?

        • Echo Tango says:

          An algorithm would be something like the steps needed to be followed, to solve a particular problem. For example, how to find a way through a 2D maze, or how to find the shortest route to send data through a computer network. It would be a description of how to solve the problem in general, with general[1] guarantees about time / space requirements.

          A computer program written by me, which implements that algorithm, would be copyrighted to me. So, if somebody copied a substantial portion of code that I had written, or key pieces that made my implementation 50ms faster than the other software, they’d be violating copyright. If they made their own program from scratch, from their own expert knowledge and the known algorithm, then that program is copyrighted to them.

          [1] Big-O notation is fairly loose for timing and memory guarantees, although there could be more strict notation given, like Theta notation. (top of this section)

          • 4th Dimension says:

            Sure, but how do you differentiate between somebody copying your code, somebody rewriting your code and somebody stumbling on his own into the same trick you used?

            Because the last seems to me would be more or less indistinguishable from the second case.

            • Richard says:

              That’s where legal arguments can become very expensive – and is a significant part of the Zenimax/Oculus case.

              For a non-code example, consider writing a story in which a plucky heroine harnesses the power of Heart to defeat the forces of evil who wished to end civilisation as we know it!

              My version of that story would be similar, but also different to yours.

              However, many of my individual paragraphs will be very similar to yours. Some might even be word-for-word identical.
              Lots of plucky heroines only differ in name, there’s only so many forces of evil, and there’s a lot of very common tropes that would appear in such a story!

              Code is a little similar.

              Except different, as there are many algorithms where there’s only one possible implementation, and even more where there are exactly three optimal implementations on a given hardware platform – the fastest, the smallest code size and the smallest RAM consumption – and so everyone migrates towards one of those three.

              On top of that, the compiler will convert lots of “different” code into the same machine instructions as it optimises, so identical binary fragments don’t mean identical source code…

        • default_ex says:

          I’ll give a bit more technical example for you.

          In computer science we have a structure known as a quad tree. This structure is used to divide a square space (typically an image) into smaller squares that are easier to handle. You get the “tree” part when you start dividing those smaller squares into even smaller squares yet, the resulting structure when plotted on paper resembles a tree branching out as you go from largest to smallest squares.

          From this description you might be imagining how you could implement this with the knowledge you personally posses. Perhaps your imagining the naive approach, divide the large square into 4 equally sized smaller squares, rinse and repeat. I imagined a method using Hilbert curves and a significant amount of binary mathematics. Your goal might have been just to make the stuff more manageable while mine was to make it fit into a smaller space while also allowing it to be easily handled by known rules the CPU cache obeys.

          Both of our implementations do effectively the same thing, they implement a relatively straightforward algorithm. The way we would go about doing it however is incredibly different. The naive approach relies solely on arithmetic. My approach has very little arithmetic with a lot of what’s normally gleaned by the arithmetic being implied by the structure formed in memory/cache. The point being our code will look very different and the resulting compiled machine code would also look very different.

          Note the idea of using hilbert curves and binary mathematics to accelerate quad trees is not in and of itself copyright or patent. Nor is the idea of using arithmetic and simple grid cell layouts that result from it. It’s the specific code each of us writes that is protected, that is after all the product and not the idea.

      • Boobah says:

        * Mathematics is the only scientific field where theories can actually be “proven”. In all others “proven” means “many different experiments have attempted to disprove it and have failed to find any problems”.

        That difference is because mathematics isn’t science.

        In math, you take some rigidly-defined assumptions (like ‘the shortest distance between two points is a straight line’) and see where logic takes you. You can make different assumptions (like ‘the shortest distance between two points is an arc’) and get an equally valid answer, although mapping such a math onto the real world can be problematic. Or simply lead to a much simpler way to define and answer a problem; the first assumption holds true on a flat surface, while the second is true on the surface of a spheroid.

        Science, on the other hand, is a series of best-guesses. You make a guess, try to come up with a way to prove the guess wrong, and if you can’t you assume that that’s what is actually happening. Even if it’s not literally the truth, if the world acts like it is you’re good. And if the world doesn’t act like that in some rare cases, congratulations! You now have a clue as to what is really going on.

        And even a disproven theory can be useful; we still learn Newton’s laws of motion despite the fact that they’re wrong (or at least incomplete); as long as we stay out of any absurdly powerful gravity wells, don’t move at relativistic speeds, and only ask questions about macroscopic objects they’re perfectly fine.

        • Daemian Lucifer says:

          That difference is because mathematics isn’t science.

          Weeeelllll….Ok,if we disregard that it actually is called a science and go into the deeper philosophy of it,like you have done,there is applied mathematics and theoretical mathematics.Applied maths is involved in various things,like physics,cryptography,engineering,etc,etc.It actually does have things that are models of real world,and things bound by actual laws of reality.So this can be considered science even by the definitions you are using.

          Theoretical maths goes into concepts like infinity,paradoxes,etc,which can be applied to the real world from time to time,but are primarily more interested in just the theory of the concepts and are bound merely by the initial axioms and underlying logic.That can be considered not science.

  14. Aaron Nowack says:

    Disclaimer: I am not lawyer, or an expert of any sort, I just remember writing a paper on this years ago in college. I haven’t read the Zenimax complaint, but I believe that the “non-literal copying” thing is a legal term of art in the copyright world, and not really anything silly.

    Under traditional copyright tests used for novels and the like, lots of obvious code copying wouldn’t be found to be copying. If I just rename all the method names and variables, we might find that only 1% of lines are found in both of the two programs’ source files! But, obviously, copying occurred here, just slightly obfuscated. And we can obfuscate more by rearranging code, making changes like swapping ‘++i’ for ‘i = i + 1’, and so on.

    More realistically, and relevant to this case, you can have a piece of copied software that is structured and designed near-identically to the original because the second team studied and referenced the original before writing their own. This may not even be intentional! (This is why you’ll hear terms like “clean room” when talking about reverse engineering and the like – you reduce potential liability by ensuring that your team can’t have copied from the original because they didn’t have access to it.)

    Now obviously, checking for this sort of copying in software is… non-trivial. It can’t just be reduced to a simple mechanical test, but is always very specific to the code in question, what it is doing, and the history of the code bases in question (who had access to what when, etc.)

    To make a very long story short, there was a series of court cases hashing out just how copyright law applied to software, and one set of those cases established the Abstraction-Filtration-Comparison test which I believe is still generally in use today (in the US, anyway) for determining this sort of thing. The “filtration” step is the one that is supposed to avoid the absurdities you describe, by removing from consideration parts of the code/design that are too common, simple, obvious, or necessary to be indications of copying.

    • Echo Tango says:

      The Abstraction-Filtration-Comparison test actually sounds very reasonable, although it sounds very expensive to perform. :)

      • Richard says:

        Very interesting, thanks!

        I wonder if anyone has attempted to point an AI framework at it?

        • Aaron Nowack says:

          My research into this stuff was actually part of exploring the possibility for a senior project on plagiarism detection software for computer programming assignments. At the time (~10 years ago), if I recall correctly the state of the art was something like “we can detect blatant cheating, some of the time, if the students aren’t too tricky and the assignment isn’t too simple or too complex.”

          I imagine throwing something like Google DeepMind at the problem could give better results nowadays, though.

Leave a Reply

Comments are moderated and may not be posted immediately. Required fields are marked *


Thanks for joining the discussion. Be nice, don't post angry, and enjoy yourself. This is supposed to be fun.

You can enclose spoilers in <strike> tags like so:
<strike>Darth Vader is Luke's father!</strike>

You can make things italics like this:
Can you imagine having Darth Vader as your <i>father</i>?

You can make things bold like this:
I'm <b>very</b> glad Darth Vader isn't my father.

You can make links like this:
I'm reading about <a href="http://en.wikipedia.org/wiki/Darth_Vader">Darth Vader</a> on Wikipedia!

You can quote someone like this:
Darth Vader said <blockquote>Luke, I am your father.</blockquote>