Google v. Oracle
April 5, 2021 7:38 AM Subscribe
The United States Supreme Court has decided in favor of Google [pdf] in the case of Google v. Oracle, essentially resolving a case begun 11 years ago. The 6-2 majority* avoided deciding whether or not the Java API was copyrightable. Rather, it held that, even if the API is copyrightable, Google's use of the API for Android was fair use. SCOTUSblog has more.
* Barrett did not participate, having joined the Court mid-way through the case.
For civil procedure nerds: the Court also held that fair use is ultimately a matter of law, not fact, and does not fall under the Seventh Amendment. That is, there is no constitutional right to have the question of fair use decided by a jury.
Full disclosure: I am the co-author of an amicus brief that was filed in support of Google.
* Barrett did not participate, having joined the Court mid-way through the case.
For civil procedure nerds: the Court also held that fair use is ultimately a matter of law, not fact, and does not fall under the Seventh Amendment. That is, there is no constitutional right to have the question of fair use decided by a jury.
Full disclosure: I am the co-author of an amicus brief that was filed in support of Google.
This is good news. The idea that an API can be copyrighted is completely crazy, and if the court had held in favor of Oracle, it would have turned the software industry on its head. It's horrifying that it took 11 years, hundreds of millions of dollars, and an appeal all the way to the Supremes to confirm this, but the decision itself is good news.
posted by Mayor West at 7:58 AM on April 5, 2021 [42 favorites]
posted by Mayor West at 7:58 AM on April 5, 2021 [42 favorites]
It’s a damning indictment of the American legal system that this trial was allowed to stumble along for 11 years.
Oracle never had a compelling argument, and the impact of a precedent-setting ruling in Oracle’s favor would have been an unmitigated catastrophe.
It’s difficult to calculate the sum total of the chilling effects that this lawsuit had over the past decade. Many companies undoubtedly paid unnecessary licensing fees as a hedge against a potential Oracle victory. Many others likely just chuckled nervously, given the extinction-level-event that such a victory would have inflicted on the software industry.
posted by schmod at 7:59 AM on April 5, 2021 [24 favorites]
Oracle never had a compelling argument, and the impact of a precedent-setting ruling in Oracle’s favor would have been an unmitigated catastrophe.
It’s difficult to calculate the sum total of the chilling effects that this lawsuit had over the past decade. Many companies undoubtedly paid unnecessary licensing fees as a hedge against a potential Oracle victory. Many others likely just chuckled nervously, given the extinction-level-event that such a victory would have inflicted on the software industry.
posted by schmod at 7:59 AM on April 5, 2021 [24 favorites]
Nice way to keep intellectual property litigation ambiguous. As in win win for legal billings.
posted by sammyo at 7:59 AM on April 5, 2021 [9 favorites]
posted by sammyo at 7:59 AM on April 5, 2021 [9 favorites]
Famed Copyright Attorney and Emotional Support Lawyer Mike Dunford (familiar to many from post-election litigation discussion) discusses the ruling as a first pass (Tweetthread).
posted by delfin at 7:59 AM on April 5, 2021 [4 favorites]
posted by delfin at 7:59 AM on April 5, 2021 [4 favorites]
I must confess, this feels a bit like a time capsule being opened. Where is Groklaw when you need them??
posted by wenestvedt at 8:00 AM on April 5, 2021 [14 favorites]
posted by wenestvedt at 8:00 AM on April 5, 2021 [14 favorites]
Oh thank god it came down correct. Copyrightable APIs would just be one more legal clusterfuck in the way of interoperability.
posted by Your Childhood Pet Rock at 8:08 AM on April 5, 2021 [12 favorites]
posted by Your Childhood Pet Rock at 8:08 AM on April 5, 2021 [12 favorites]
Yeah, it’s unfortunate that question wasn’t actually resolved. So it seems there’s still no legal precedent to guide or restrict essentially identical litigation in the future?
posted by darkstar at 8:10 AM on April 5, 2021 [1 favorite]
posted by darkstar at 8:10 AM on April 5, 2021 [1 favorite]
Right ruling. Wrong reason.
posted by CheeseDigestsAll at 8:19 AM on April 5, 2021 [4 favorites]
posted by CheeseDigestsAll at 8:19 AM on April 5, 2021 [4 favorites]
It’s technically not precedent-setting, but “The Dalvik implementation of Java 6 is an example of an API reimplementation that qualifies as fair-use” is absolutely going to used as a benchmark.
posted by schmod at 8:25 AM on April 5, 2021 [10 favorites]
posted by schmod at 8:25 AM on April 5, 2021 [10 favorites]
[making a mental note to check in with this topic in a month so the discussion has a chance to catch up and include open source software]
posted by snerson at 8:28 AM on April 5, 2021
posted by snerson at 8:28 AM on April 5, 2021
Why, snerson? You get a copyright licence in well-enacted open source or free/libre software projects, which is mostly BSD, MIT, Apache or one of the GPL's. The specific one which crosses with this case is Library/Lesser GPL and both v2 and v3 aim to allow a combined work of any other licence with the programming interfaces (here, API's) of the freely-licensed code.
Compatible reimplementation (here was JDKv6) say in Wine, ReactOS or Darling (Darwin for Linux, running macOS apps) would be at issue if we'd had a different outcome.
posted by k3ninho at 9:00 AM on April 5, 2021 [3 favorites]
Compatible reimplementation (here was JDKv6) say in Wine, ReactOS or Darling (Darwin for Linux, running macOS apps) would be at issue if we'd had a different outcome.
posted by k3ninho at 9:00 AM on April 5, 2021 [3 favorites]
It’s a damning indictment of the American legal system that this trial was allowed to stumble along for 11 years.
This. At the same time, remember when SCO more or less claimed to own linux and was going to sue the world, and did sue Novell/IBM? While Novell was settled, the IBM suit is still going on. And, the company that bought SCO group is still involved in the SCO/IBM suit, and they apparently just started a new suit suing IBM and Red Hat: https://arstechnica.com/gadgets/2021/04/xinuos-finishes-picking-up-scos-mantle-by-suing-red-hat-and-ibm/.
posted by nobeagle at 9:12 AM on April 5, 2021 [6 favorites]
This. At the same time, remember when SCO more or less claimed to own linux and was going to sue the world, and did sue Novell/IBM? While Novell was settled, the IBM suit is still going on. And, the company that bought SCO group is still involved in the SCO/IBM suit, and they apparently just started a new suit suing IBM and Red Hat: https://arstechnica.com/gadgets/2021/04/xinuos-finishes-picking-up-scos-mantle-by-suing-red-hat-and-ibm/.
posted by nobeagle at 9:12 AM on April 5, 2021 [6 favorites]
Yeah, it’s unfortunate that question wasn’t actually resolved. So it seems there’s still no legal precedent to guide or restrict essentially identical litigation in the future?
I don't think that's right; I think the fair-use parts are pretty far-reaching. From the decision:
I don't think that's right; I think the fair-use parts are pretty far-reaching. From the decision:
The upshot, in our view, is that fair use can play an important role in determining the lawful scope of a computer program copyright, such as the copyright at issue here. It can help to distinguish among technologies. It can distinguish between expressive and functional features of computer code where those features are mixed. It can focus on the legitimate need to provide incentives to produce copyrighted material while examining the extent to which yet-further protection creates unrelated or illegitimate harms in other markets or to the development of other products. In a word, it can carry out its basic purpose of providing a context based check that can help to keep a copyright monopoly within its lawful bounds.An IP lawyer I follow and trust says:
This is a pretty far reaching holding. While fair use is usually pretty hard to argue, the language here suggests that using an API is almost always going to be considered a fair use.posted by dorothy hawk at 9:16 AM on April 5, 2021 [6 favorites]
So thankful the decision came down this way. Given the history of this case and the back and forth rulings it really felt like a roll of the dice. I do not know a single working software engineer who thinks this case should have been decided any other way but for freedom to reimplement APIs. Shame on Oracle and its legal and expert team for even trying to argue otherwise.
Alex Macgillivray, formerly a lawyer for Google, puts this case in context
Ironically I think basing Android on Java and open source was a total mistake by Google. At the time it seemed like the "right" thing; use a standard language, use open source because of course you would! But both decisions have ended up being real millstones hung around Google's neck. In retrospect they would have been better off with a free, source-published system where they were the only primary vendor but it was not fully open source. And Java, well, that choice hasn't aged well at all.
posted by Nelson at 9:18 AM on April 5, 2021 [7 favorites]
Alex Macgillivray, formerly a lawyer for Google, puts this case in context
With the Oracle decision today, Google wins the third of a set of big copyright cases about decisions made >10yrs ago by a strong team of lawyers with whom I was privileged to work:Think what you want about Google as a business, but I think all three principles benefit end users. I'm glad Google pursued all three.
1) book search is fair use
2) the DMCA protects YouTube
3) Android's use Java APIs is fair use
Ironically I think basing Android on Java and open source was a total mistake by Google. At the time it seemed like the "right" thing; use a standard language, use open source because of course you would! But both decisions have ended up being real millstones hung around Google's neck. In retrospect they would have been better off with a free, source-published system where they were the only primary vendor but it was not fully open source. And Java, well, that choice hasn't aged well at all.
posted by Nelson at 9:18 AM on April 5, 2021 [7 favorites]
Google didn't choose Java for Android - Andy Rubin and the group at Danger did, before they were acquired by Google. And Danger had just done a pretty kickass phone using Java, so it seemed like the logical step. What else were they supposed to use? Symbian? (BARF)
As part of the original OHA meetings, Google didn't give a shit what Android ran on, as long as it got advertising to the eyeballs of the % of users that didn't have a PC but had a mobile phone back in 2007.
posted by JoeZydeco at 9:22 AM on April 5, 2021 [6 favorites]
As part of the original OHA meetings, Google didn't give a shit what Android ran on, as long as it got advertising to the eyeballs of the % of users that didn't have a PC but had a mobile phone back in 2007.
posted by JoeZydeco at 9:22 AM on April 5, 2021 [6 favorites]
Ironically I think basing Android on Java and open source was a total mistake by Google
While I agree with that sentiment on technical grounds, Android was originally developed by a company that Google acquired in 2005. Google had no hand in the original selection of the platform.
Back in the early 2000s, I’d have imagined that Java’s combination of OSS-friendliness and commercial backing actually made it seem like a fairly safe choice.
posted by schmod at 9:23 AM on April 5, 2021 [3 favorites]
While I agree with that sentiment on technical grounds, Android was originally developed by a company that Google acquired in 2005. Google had no hand in the original selection of the platform.
Back in the early 2000s, I’d have imagined that Java’s combination of OSS-friendliness and commercial backing actually made it seem like a fairly safe choice.
posted by schmod at 9:23 AM on April 5, 2021 [3 favorites]
And Java, well, that choice hasn't aged well at all.
it's not for nothing that Google has been pushing Android devs to move to Kotlin since 2017.
posted by i used to be someone else at 9:24 AM on April 5, 2021 [3 favorites]
it's not for nothing that Google has been pushing Android devs to move to Kotlin since 2017.
posted by i used to be someone else at 9:24 AM on April 5, 2021 [3 favorites]
Well, because I don't have any background on this at all, and I'm hoping somebody on the Internet will do a postmortem I find legible due to the high profile nature of this ruling. Your comment has a lot of great TLAs and namedrops, but I don't understand the specifics and I don't have the time or energy to unwind all of it. And I work adjacent to these concepts, even if I'm not coding myself.
I just think it's likely that somebody who has a better view of the field will be paid to write a snerson-level overview of the case, and hopefully do a drive-by the interplay between fair use, copyright, and open source. I'm sure there are mefites here who would be willing to write it all out in a comment, but in my experience, my learning about these concepts is directly linked to, well, how much the piece has been sat on, edited for clarity, and re-evaluated for baseline cluelessness of reader.
Which is not to say that mefites can't do that, but... it's not your day job you know? + I'm sure if we were at a water cooler discussing it, that would be fine, but in my experience, I don't do well with this sort of learning/digesting in a thread setting.
posted by snerson at 9:27 AM on April 5, 2021
I just think it's likely that somebody who has a better view of the field will be paid to write a snerson-level overview of the case, and hopefully do a drive-by the interplay between fair use, copyright, and open source. I'm sure there are mefites here who would be willing to write it all out in a comment, but in my experience, my learning about these concepts is directly linked to, well, how much the piece has been sat on, edited for clarity, and re-evaluated for baseline cluelessness of reader.
Which is not to say that mefites can't do that, but... it's not your day job you know? + I'm sure if we were at a water cooler discussing it, that would be fine, but in my experience, I don't do well with this sort of learning/digesting in a thread setting.
posted by snerson at 9:27 AM on April 5, 2021
Remember kids, while we all complain about Facebook/Microsoft/Google, Oracle is the original evil monster software company based on predatory business practices. Besides all of Oracle's software sins, Larry Ellison is one of the motherfuckers who planted the hydroxychloroquine miracle cure idea in Trump's head.
posted by benzenedream at 9:27 AM on April 5, 2021 [30 favorites]
posted by benzenedream at 9:27 AM on April 5, 2021 [30 favorites]
Well, I guess one of them had to win.
No matter your view on how terrible Google is, Oracle is 10x worse.
posted by sideshow at 9:29 AM on April 5, 2021 [28 favorites]
No matter your view on how terrible Google is, Oracle is 10x worse.
posted by sideshow at 9:29 AM on April 5, 2021 [28 favorites]
Eh. They are both toxic in their own ways. If the two companies involved had traded places, by happenstance, I suspect Google's legal department would have worked just as hard at defending its business interests. Generally, having rights determined or guided by monopolists doesn't seem to bode well for anyone who isn't an equity holder. Whatever benefit that comes out of this seems like a lucky side effect.
posted by They sucked his brains out! at 9:32 AM on April 5, 2021 [4 favorites]
posted by They sucked his brains out! at 9:32 AM on April 5, 2021 [4 favorites]
If the two companies involved had traded places, by happenstance, I suspect Google's legal department would have worked just as hard at defending its business interests.
That statement is objectively false, because Firefox abandoned its own extension API due to unfixable security issues, and instead copied Chrome's extension API. Google has never objected in the slightest. Oracle is really a much, much worse company than Google.
posted by gsteff at 9:44 AM on April 5, 2021 [19 favorites]
That statement is objectively false, because Firefox abandoned its own extension API due to unfixable security issues, and instead copied Chrome's extension API. Google has never objected in the slightest. Oracle is really a much, much worse company than Google.
posted by gsteff at 9:44 AM on April 5, 2021 [19 favorites]
It’s a damning indictment of the American legal system that this trial was allowed to stumble along for 11 years.
The two trials only took a couple of months between them, even with the first trial involving separate copyright and patent phases. Everything else has been wrangling about the legal issues. This always takes forever because parties are typically given months to prepare for each stage of the appellate process. The two trips to the Supreme Court really exploded the litigation timeline because the Supreme Court is incredibly inefficient and only decides about 1 case per week, on average. The pandemic also slowed down the second Supreme Court round. I find the Court's pathetic (and declining!) output to be a strong argument for significantly expanding the Court and not requiring every Justice to hear every case, but I digress.
The combination of novel legal issues with wide-ranging consequences, parties with essentially unlimited resources, and high stakes even by the standards of such large companies meant that it was almost invariably going to go on for a long time with numerous appeals. Oracle in particular had every incentive to drag it out because Android's success meant that the potential damages award kept getting bigger and bigger. Also there was no significant counter-claim against Oracle to induce settlement.
The ne plus ultra of incredibly drawn-out intellectual property litigation is probably Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co Ltd., which started in 1993 and didn't finally, truly end until 25 years later after something like 77 district, appellate, and Supreme Court opinions. The main opinions are called Festo I through Festo XIII, like a horror franchise that just didn't know when to quit.
posted by jedicus at 9:51 AM on April 5, 2021 [17 favorites]
The two trials only took a couple of months between them, even with the first trial involving separate copyright and patent phases. Everything else has been wrangling about the legal issues. This always takes forever because parties are typically given months to prepare for each stage of the appellate process. The two trips to the Supreme Court really exploded the litigation timeline because the Supreme Court is incredibly inefficient and only decides about 1 case per week, on average. The pandemic also slowed down the second Supreme Court round. I find the Court's pathetic (and declining!) output to be a strong argument for significantly expanding the Court and not requiring every Justice to hear every case, but I digress.
The combination of novel legal issues with wide-ranging consequences, parties with essentially unlimited resources, and high stakes even by the standards of such large companies meant that it was almost invariably going to go on for a long time with numerous appeals. Oracle in particular had every incentive to drag it out because Android's success meant that the potential damages award kept getting bigger and bigger. Also there was no significant counter-claim against Oracle to induce settlement.
The ne plus ultra of incredibly drawn-out intellectual property litigation is probably Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co Ltd., which started in 1993 and didn't finally, truly end until 25 years later after something like 77 district, appellate, and Supreme Court opinions. The main opinions are called Festo I through Festo XIII, like a horror franchise that just didn't know when to quit.
posted by jedicus at 9:51 AM on April 5, 2021 [17 favorites]
instead copied Chrome's extension API
Respectfully, I'm not sure that's quite the same, as Firefox does not negatively impact Google's business model. If anything, Mozilla is good for its bottom line by further solidifying Chrome as the standard for web browsing and by making its search engine the default in Firefox. There seems little obvious motivation to go after an open source non-profit foundation. It would be bad PR, but mainly it would hurt profitability and market share.
posted by They sucked his brains out! at 9:54 AM on April 5, 2021 [1 favorite]
Respectfully, I'm not sure that's quite the same, as Firefox does not negatively impact Google's business model. If anything, Mozilla is good for its bottom line by further solidifying Chrome as the standard for web browsing and by making its search engine the default in Firefox. There seems little obvious motivation to go after an open source non-profit foundation. It would be bad PR, but mainly it would hurt profitability and market share.
posted by They sucked his brains out! at 9:54 AM on April 5, 2021 [1 favorite]
Putting the good/evil argument aside, copyright maximalism aligns very poorly with most of Google’s businesses. It’s not a bear they’d want to poke.
posted by schmod at 10:03 AM on April 5, 2021 [3 favorites]
posted by schmod at 10:03 AM on April 5, 2021 [3 favorites]
Google could have shifted Android off of Java when it acquired Android. (Source: I was working there and knew a lot of the folks involved). It would have had a switching cost, for sure, but not a huge one. I think the main reason they didn't is Google at the time really bought into this ideal of an open phone platform and Java fit the bill reasonably well. I think they genuinely expected Android to be meaningfully open source, too. Also at the time Sun still existed and owned Java. They were a much more welcoming and reasonable business partner than Oracle, so Google may have been blind to the legal risk.
I still maintain the Java legacy is a major cost now. I probably would have made the same Java+open source choice in 2005 though, too.
While I'm here: your regular reminder that Google gave Andy Rubin $90M and did their best to hide his history of sexual harassment and misconduct when they pushed him out of the company in 2014. Part of a pattern of senior Google executives and sexual misconduct scandals.
posted by Nelson at 10:10 AM on April 5, 2021 [3 favorites]
I still maintain the Java legacy is a major cost now. I probably would have made the same Java+open source choice in 2005 though, too.
While I'm here: your regular reminder that Google gave Andy Rubin $90M and did their best to hide his history of sexual harassment and misconduct when they pushed him out of the company in 2014. Part of a pattern of senior Google executives and sexual misconduct scandals.
posted by Nelson at 10:10 AM on April 5, 2021 [3 favorites]
jedicus noted: "Full disclosure: I am the co-author of an amicus brief that was filed in support of Google."
You and, like, at least 3 lawyers I know! Wow I knew there were several amicus/amica/amici curiae briefs filed in this case but now I see there were SCORES of them! BTW, now I am curious, does anyone know which Supreme Court case has had the MOST amicus briefs filed? Is it still Obergefell v. Hodges or has that record been surmounted?
posted by brainwane at 10:19 AM on April 5, 2021 [3 favorites]
You and, like, at least 3 lawyers I know! Wow I knew there were several amicus/amica/amici curiae briefs filed in this case but now I see there were SCORES of them! BTW, now I am curious, does anyone know which Supreme Court case has had the MOST amicus briefs filed? Is it still Obergefell v. Hodges or has that record been surmounted?
posted by brainwane at 10:19 AM on April 5, 2021 [3 favorites]
Thomas and Alito dissented so it must be good news.
posted by dances_with_sneetches at 10:22 AM on April 5, 2021 [19 favorites]
posted by dances_with_sneetches at 10:22 AM on April 5, 2021 [19 favorites]
does anyone know which Supreme Court case has had the MOST amicus briefs filed? Is it still Obergefell v. Hodges or has that record been surmounted?
Still Obergefell with 148, although Windsor & Perry had 156 between them. NFIB v. Sebelius (the main ACA case) was close with 136.
posted by jedicus at 10:31 AM on April 5, 2021 [5 favorites]
Still Obergefell with 148, although Windsor & Perry had 156 between them. NFIB v. Sebelius (the main ACA case) was close with 136.
posted by jedicus at 10:31 AM on April 5, 2021 [5 favorites]
Google could have shifted Android off of Java when it acquired Android.
To what, Nelson? Aquaman?
It was 2007. We didn't have Go. We didn't have Rust. Do we just throw devs into C++ deep end? C#? Mono still didn't even support .NET 2.0 little alone 3.5. Not to mention there was still the debate of which architecture would win in mobile space: x86, ARM, or MIPS. LLVM had no proper C frontend instead still being grafted onto GCC like Frankenstein's monster.
Java was a bad choice but it's not like there was a good one or even a viable alternative. Apple only solved the problem because it had its own statically compiled language that had its own cadre of developers already assembled.
posted by Your Childhood Pet Rock at 10:33 AM on April 5, 2021 [13 favorites]
To what, Nelson? Aquaman?
It was 2007. We didn't have Go. We didn't have Rust. Do we just throw devs into C++ deep end? C#? Mono still didn't even support .NET 2.0 little alone 3.5. Not to mention there was still the debate of which architecture would win in mobile space: x86, ARM, or MIPS. LLVM had no proper C frontend instead still being grafted onto GCC like Frankenstein's monster.
Java was a bad choice but it's not like there was a good one or even a viable alternative. Apple only solved the problem because it had its own statically compiled language that had its own cadre of developers already assembled.
posted by Your Childhood Pet Rock at 10:33 AM on April 5, 2021 [13 favorites]
I will say that using Linux as the HAL for Android was a pretty nimble move. Everyone was soooo tired of writing ground-up HALs and OSALs by the mid-2000s.
posted by JoeZydeco at 10:44 AM on April 5, 2021 [2 favorites]
posted by JoeZydeco at 10:44 AM on April 5, 2021 [2 favorites]
Google could have shifted Android off of Java when it acquired Android.
Moving Android "off of Java" would just mean writing an entirely new mobile OS, so why even acquire them?
C++ would have been the most viable option at that time (however imagine the wonderful world we would live in if they had invested in OCaml. Alas)
posted by dis_integration at 11:19 AM on April 5, 2021
Moving Android "off of Java" would just mean writing an entirely new mobile OS, so why even acquire them?
C++ would have been the most viable option at that time (however imagine the wonderful world we would live in if they had invested in OCaml. Alas)
posted by dis_integration at 11:19 AM on April 5, 2021
Oracle didn’t create Java, they bought Sun; and Sun had created Java something that was open and community driven. Oracle decided to change the terms and put up a padlock and try to charge rent to the the community that built Java with Sun.
posted by interogative mood at 11:21 AM on April 5, 2021 [11 favorites]
posted by interogative mood at 11:21 AM on April 5, 2021 [11 favorites]
> No matter your view on how terrible Google is, Oracle is 10x worse.
Would you rather have frequent large doses of something mildly poisonous but highly addictive, or small doses of something extremely toxic?
posted by at by at 11:34 AM on April 5, 2021
Would you rather have frequent large doses of something mildly poisonous but highly addictive, or small doses of something extremely toxic?
posted by at by at 11:34 AM on April 5, 2021
With the Oracle decision today, Google wins the third of a set of big copyright cases about decisions made >10yrs ago by a strong team of lawyers with whom I was privileged to work:
1) book search is fair use
2) the DMCA protects YouTube
3) Android's use Java APIs is fair use
10 years ago? Ah yes, back when "Don't Be Evil" wasn't yet a grim punchline.
posted by Mr.Encyclopedia at 11:53 AM on April 5, 2021 [3 favorites]
1) book search is fair use
2) the DMCA protects YouTube
3) Android's use Java APIs is fair use
10 years ago? Ah yes, back when "Don't Be Evil" wasn't yet a grim punchline.
posted by Mr.Encyclopedia at 11:53 AM on April 5, 2021 [3 favorites]
Would you rather have frequent large doses of something mildly poisonous but highly addictive, or small doses of something extremely toxic?
Or continuous huge doses of something fatally toxic, which is what Oracle is in your analogy.
posted by sideshow at 12:01 PM on April 5, 2021 [5 favorites]
Or continuous huge doses of something fatally toxic, which is what Oracle is in your analogy.
posted by sideshow at 12:01 PM on April 5, 2021 [5 favorites]
If y’all want to dive into the details about Google and Oracle’s respective misdeeds, let’s go ahead and do that.
If all you have is a witty quip about how Google and/or Oracle are evil, I think that point has already been made at this point.
posted by schmod at 12:35 PM on April 5, 2021 [17 favorites]
If all you have is a witty quip about how Google and/or Oracle are evil, I think that point has already been made at this point.
posted by schmod at 12:35 PM on April 5, 2021 [17 favorites]
Reading through Mike Dunford's twitter thread analysis, linked upthread, I'm completely mindblown that
a) there are two completely different schools of thought on what copyright is even *for*, and
b) this decision punted on whether copyright should have existed.
Copyright has become such a tower of babel not even the Supremes can agree on what it means!
posted by Popular Ethics at 12:51 PM on April 5, 2021 [4 favorites]
a) there are two completely different schools of thought on what copyright is even *for*, and
b) this decision punted on whether copyright should have existed.
Copyright has become such a tower of babel not even the Supremes can agree on what it means!
posted by Popular Ethics at 12:51 PM on April 5, 2021 [4 favorites]
>I do not know a single working software engineer who thinks this case should have been decided any other way but for freedom to reimplement APIs.
There can be actual bona-fide IP in APIs, Sun did do a lot of heavy lifting to get the details of the Java platform hashed out, which Microsoft and later Google just lifted wholesale.
That doesn't seem ~entirely~ right; like musical artists who cover other artists, there should be mechanical royalties involved here.
posted by Heywood Mogroot III at 12:53 PM on April 5, 2021 [2 favorites]
There can be actual bona-fide IP in APIs, Sun did do a lot of heavy lifting to get the details of the Java platform hashed out, which Microsoft and later Google just lifted wholesale.
That doesn't seem ~entirely~ right; like musical artists who cover other artists, there should be mechanical royalties involved here.
posted by Heywood Mogroot III at 12:53 PM on April 5, 2021 [2 favorites]
What interests me is that the decision skipped on whether APIs are copyrightable. It was deciding on whether this instance of copying was fair use or not.
I think a broad argument that APIs are not freely copyable could be that APIs are formally abstract programs in their own right. I personally think almost all copyright and patenting of software is effectively harmful to humanity, but from a consistency argument I can kind of see if accepting the premise of copyrighting code that it is hard to make exceptions for code just because it is abstract code that is nonexecutable. Like maybe their counterargument will be something like, you can't use copyrighted blueprints of a building, etc.
posted by polymodus at 12:54 PM on April 5, 2021 [4 favorites]
I think a broad argument that APIs are not freely copyable could be that APIs are formally abstract programs in their own right. I personally think almost all copyright and patenting of software is effectively harmful to humanity, but from a consistency argument I can kind of see if accepting the premise of copyrighting code that it is hard to make exceptions for code just because it is abstract code that is nonexecutable. Like maybe their counterargument will be something like, you can't use copyrighted blueprints of a building, etc.
posted by polymodus at 12:54 PM on April 5, 2021 [4 favorites]
My experience of Oracle as an IT director:
Weird, extortionary email arrives informing me my company of ~450 people is using over 500 Virtualbox licenses, can we please cough up $35,000 to make this go away. We're helpfully provided with a PDF that explains how to tell if VirtualBox is on a laptop.
I sit down with the helpdesk guy and we write a custom Puppet fact to find VirtualBox. Weird ... we can only find a total of a few dozen instances in the whole laptop fleet, mostly among devs and a few sales engineers.
We take that back to the Oracle guy, who then coughs up a spreadsheet of IP addresses of VB downloads with dates and times, which are all suspiciously consistent as to time of day. Oh, that explains it: Those are all in our CI system's address space, and Oracle was just counting downloads of the VB image from our company's whole address space. We explain the situation: Your stuff turns up on a few dozen laptops and exactly three Jenkins jobs, and it all applies to our open source stuff, which is supposed to be able to evade these fees. But how does a few hundred bucks for perpetual licenses for those sound?
Oracle dude skulks away. Our procurement guy yells "and don't come back!" at his retreating back.
Months later, in a casual conversation with one of the sales people who used to be at Oracle we learn this whole "get a list of IP addresses, demand money" thing is a semi-official Oracle thing they all do when they need to hit their quarterly quotas.
Two years pass, I get another email with another list of IP addresses and another demand for "urgent action on potential legal issues" with our VB use, to the tune of $35,000. It is so poorly written and arrives at such a strange time I can only assume that the dude wrote it on his phone before passing out drunk on his couch. I forward it to the procurement guy, who I can only hope let off with a shell full of rock salt after waving around our previous receipt.
posted by mph at 12:54 PM on April 5, 2021 [25 favorites]
Weird, extortionary email arrives informing me my company of ~450 people is using over 500 Virtualbox licenses, can we please cough up $35,000 to make this go away. We're helpfully provided with a PDF that explains how to tell if VirtualBox is on a laptop.
I sit down with the helpdesk guy and we write a custom Puppet fact to find VirtualBox. Weird ... we can only find a total of a few dozen instances in the whole laptop fleet, mostly among devs and a few sales engineers.
We take that back to the Oracle guy, who then coughs up a spreadsheet of IP addresses of VB downloads with dates and times, which are all suspiciously consistent as to time of day. Oh, that explains it: Those are all in our CI system's address space, and Oracle was just counting downloads of the VB image from our company's whole address space. We explain the situation: Your stuff turns up on a few dozen laptops and exactly three Jenkins jobs, and it all applies to our open source stuff, which is supposed to be able to evade these fees. But how does a few hundred bucks for perpetual licenses for those sound?
Oracle dude skulks away. Our procurement guy yells "and don't come back!" at his retreating back.
Months later, in a casual conversation with one of the sales people who used to be at Oracle we learn this whole "get a list of IP addresses, demand money" thing is a semi-official Oracle thing they all do when they need to hit their quarterly quotas.
Two years pass, I get another email with another list of IP addresses and another demand for "urgent action on potential legal issues" with our VB use, to the tune of $35,000. It is so poorly written and arrives at such a strange time I can only assume that the dude wrote it on his phone before passing out drunk on his couch. I forward it to the procurement guy, who I can only hope let off with a shell full of rock salt after waving around our previous receipt.
posted by mph at 12:54 PM on April 5, 2021 [25 favorites]
mph,
Open Source is a response to Oracle. Thanks for the excellent illustration.
Your Childhood Pet Rock--
Indeed, it was 2007, but there's actually relatively few languages *even now* that have been proven to make scalable, pleasant GUIs. It's still mostly C, C++, Assemvly, and ObjC. There's scripting on top, sometimes jitted -- what Java was, what Google made JavaScript become with V8 -- and the most mature and open solution for Google to choose (and they did choose it) was Java. Others tried JS, WebOS comes to mind. Didn't work out.
posted by effugas at 1:20 PM on April 5, 2021 [1 favorite]
Open Source is a response to Oracle. Thanks for the excellent illustration.
Your Childhood Pet Rock--
Indeed, it was 2007, but there's actually relatively few languages *even now* that have been proven to make scalable, pleasant GUIs. It's still mostly C, C++, Assemvly, and ObjC. There's scripting on top, sometimes jitted -- what Java was, what Google made JavaScript become with V8 -- and the most mature and open solution for Google to choose (and they did choose it) was Java. Others tried JS, WebOS comes to mind. Didn't work out.
posted by effugas at 1:20 PM on April 5, 2021 [1 favorite]
Frankly, it feels as if SCOTUS had kicked the can further down the road by NOT deciding whether API is copyrightable.
The whole idea of an API is to setup a limited interface to your program so other people can use it. The idea is to promote usage.
So the idea of Oracle trying to enforce copyright on API and use that to deny Google the use of Java SE API was basically counter to the entire idea of API.
It's basically the equivalent of suing the barn door and the neighbor after the cows left.
posted by kschang at 1:28 PM on April 5, 2021 [1 favorite]
The whole idea of an API is to setup a limited interface to your program so other people can use it. The idea is to promote usage.
So the idea of Oracle trying to enforce copyright on API and use that to deny Google the use of Java SE API was basically counter to the entire idea of API.
It's basically the equivalent of suing the barn door and the neighbor after the cows left.
posted by kschang at 1:28 PM on April 5, 2021 [1 favorite]
An API doesn't have to be for other people to use. APIs are a level of abstraction used to algorithmically define abstract data types. There's nothing in CS that says a greedy person can't spend hundreds of hours developing an API only for certain people (e.g., themselves, or their research team) to use. Unfortunately. In a sense, abstract data types are half the battle, they do take a lot of work to figure out. There could even be an argument that UIs are copyrightable or trademarkable, so why not APIs and other such abstractions like language specifications.
posted by polymodus at 1:53 PM on April 5, 2021 [1 favorite]
posted by polymodus at 1:53 PM on April 5, 2021 [1 favorite]
I think it might be consistent with Breyer's legal philosophy here that they didn't take a hard line on API copyrightability (sorry, Mike). A lot of it seems to be founded on the idea that the validity of copyright protections should reflect the medium as well as technological progress (pointing out that there are differences between film and music copyright protections, for example).
Saying something like "APIs aren't copyrightable" in clear terms might age badly. Our current understanding of the boundaries between API and implementation is already fuzzy... Java 8 introduced "
I think this decision was essential and I can breathe a sigh of relief that it went the way it did. However, it seems obvious from the decision that there are many avenues remaining for bad actors to abuse IP law, and it really should be up to the legislature to fix that rather than the courts.
posted by Riki tiki at 2:10 PM on April 5, 2021
Saying something like "APIs aren't copyrightable" in clear terms might age badly. Our current understanding of the boundaries between API and implementation is already fuzzy... Java 8 introduced "
default" interface methods that aren't clearly one or the other, but that you'd probably need to replicate exactly to make something like Dalvik in the modern era (I don't believe this was done for sinister reasons, but it can definitely help sinister people).I think this decision was essential and I can breathe a sigh of relief that it went the way it did. However, it seems obvious from the decision that there are many avenues remaining for bad actors to abuse IP law, and it really should be up to the legislature to fix that rather than the courts.
posted by Riki tiki at 2:10 PM on April 5, 2021
Like every copyright decision, this one must dance around the consequences of the fact that software is a horrendous fit with copyright. In a reasonable world it would be subject to a separate regime that wasn’t built around the infinite rent-seeking of a cartoon mouse.
posted by sinfony at 2:23 PM on April 5, 2021 [13 favorites]
posted by sinfony at 2:23 PM on April 5, 2021 [13 favorites]
And while I'm here, I'm happy with the outcome of the case (as far as it goes, at least), but I can't get on board with this idea that Oracle is worse than Google. Google is at the forefront of the effort to convert every single person, alive or dead, into raw input for unknown, unknowable, and unspeakable purposes; the world is worse for its existence and will only become more so. Oracle just wants to extort you for outdated, bad software it bought from somebody else.
posted by sinfony at 2:28 PM on April 5, 2021 [7 favorites]
posted by sinfony at 2:28 PM on April 5, 2021 [7 favorites]
Well, like, people choose sides in W40k, yeah?
posted by clew at 2:54 PM on April 5, 2021 [3 favorites]
posted by clew at 2:54 PM on April 5, 2021 [3 favorites]
An important consequence of the Seventh Amendment ruling here is that fair use defenses will become somewhat-less-ruinously-expensive to litigate (perhaps even resolvable on the pleadings in some cases?), which will hopefully lead to some more robust caselaw.
There's something about Google's side of this story that has always made me very skeptical -- it seems like some sort of shell-game being played between "using" an API in the sense of interacting with it and "using" an API in the sense of just copying the whole darn thing, as Google is alleged to have done -- but I lack sufficient knowledge to have anything more than vague qualms.
posted by Not A Thing at 3:07 PM on April 5, 2021
There's something about Google's side of this story that has always made me very skeptical -- it seems like some sort of shell-game being played between "using" an API in the sense of interacting with it and "using" an API in the sense of just copying the whole darn thing, as Google is alleged to have done -- but I lack sufficient knowledge to have anything more than vague qualms.
posted by Not A Thing at 3:07 PM on April 5, 2021
For a layman, think of the API as a blueprint that shows the overall footprint of a building, detailed specifications of the entrances and exits, and nothing else.
Depending on the specific building, this could be a great deal of information, representing many years of work. Or it could something simpler, like a shed (or even a very big shed).
Google absolutely copied the Java 6 SE API nearly in its entirety.
The question is whether or not you can sue for damages if somebody else copies your doors and exits.
While it seems like you’d have a good case to make against another architect that builds an exterior replica of the Chrysler Building, it would be hugely problematic for Sears Roebuck and Company to start suing the pants off of anybody who’s ever designed, built, or owned a bungalow.
API development is serious work that requires serious effort. While one could reasonably make an argument that this kind of work should be protected under copyright, it’s very difficult to draw the line where an API becomes sufficiently complex to be worthy of those protections. It creates a huge opportunity for tedious and frivolous lawsuits that would make any kind of software-development a legally perilous endeavor.
While this may sound like a slippery-slope argument, the world of software-patent lawsuits provides plenty of precedent to show how API copyrights could go completely off the rails
posted by schmod at 3:32 PM on April 5, 2021 [8 favorites]
Depending on the specific building, this could be a great deal of information, representing many years of work. Or it could something simpler, like a shed (or even a very big shed).
Google absolutely copied the Java 6 SE API nearly in its entirety.
The question is whether or not you can sue for damages if somebody else copies your doors and exits.
While it seems like you’d have a good case to make against another architect that builds an exterior replica of the Chrysler Building, it would be hugely problematic for Sears Roebuck and Company to start suing the pants off of anybody who’s ever designed, built, or owned a bungalow.
API development is serious work that requires serious effort. While one could reasonably make an argument that this kind of work should be protected under copyright, it’s very difficult to draw the line where an API becomes sufficiently complex to be worthy of those protections. It creates a huge opportunity for tedious and frivolous lawsuits that would make any kind of software-development a legally perilous endeavor.
While this may sound like a slippery-slope argument, the world of software-patent lawsuits provides plenty of precedent to show how API copyrights could go completely off the rails
posted by schmod at 3:32 PM on April 5, 2021 [8 favorites]
I think a broad argument that APIs are not freely copyable could be that APIs are formally abstract programs in their own right.
Are they, though? Suppose I write a library in which everything above the level of raw machine code is implemented via public APIs; I literally don't even increment a number except by calling JiAmaths.increment(number). I then publish the API, all the way down to my stupid increment function. How is that API a program except in the sense that it can be compiled? The copyrightable element is surely the implementation of the functions: knowing that I implemented an "increment" function doesn't tell you whether or how I called it in JiArepetition.fornextloop(). And if my mindblowingly–detailed API doesn't tell you anything about my program logic, why should a higher–level API be any different?
posted by Joe in Australia at 3:33 PM on April 5, 2021
Are they, though? Suppose I write a library in which everything above the level of raw machine code is implemented via public APIs; I literally don't even increment a number except by calling JiAmaths.increment(number). I then publish the API, all the way down to my stupid increment function. How is that API a program except in the sense that it can be compiled? The copyrightable element is surely the implementation of the functions: knowing that I implemented an "increment" function doesn't tell you whether or how I called it in JiArepetition.fornextloop(). And if my mindblowingly–detailed API doesn't tell you anything about my program logic, why should a higher–level API be any different?
posted by Joe in Australia at 3:33 PM on April 5, 2021
Google is at the forefront of the effort to convert every single person, alive or dead, into raw input for unknown, unknowable, and unspeakable purposes
Indeed. We should remember that one of these two companies has multiple major antitrust lawsuits against it at state and federal US levels and in Europe for engaging in monopolistic practices in search and collusion with social network ad companies. I suspect the larger question isn't so much whether this case makes Google better or worse than Oracle — as the differences between the two are not in degree but in kind — but whether monopolies should, in the first place, have the undue influence they have on our legal system and, by extension, our society. This ruling may accidentally benefit a few of us who happen to write software, but benefiting society was definitely not what motivated the ruling or legal action by either side.
posted by They sucked his brains out! at 4:07 PM on April 5, 2021
Indeed. We should remember that one of these two companies has multiple major antitrust lawsuits against it at state and federal US levels and in Europe for engaging in monopolistic practices in search and collusion with social network ad companies. I suspect the larger question isn't so much whether this case makes Google better or worse than Oracle — as the differences between the two are not in degree but in kind — but whether monopolies should, in the first place, have the undue influence they have on our legal system and, by extension, our society. This ruling may accidentally benefit a few of us who happen to write software, but benefiting society was definitely not what motivated the ruling or legal action by either side.
posted by They sucked his brains out! at 4:07 PM on April 5, 2021
I'm not even sure how it follows that an API could be copyrighted. Surely it should either fall under patent (a method of doing business or running a machine) or not at all, as for example game mechanics cannot be patented, nor can fashion designs.
It's program rules, not a work of art.
posted by explosion at 5:15 PM on April 5, 2021 [1 favorite]
It's program rules, not a work of art.
posted by explosion at 5:15 PM on April 5, 2021 [1 favorite]
The decision addresses that.
posted by schmod at 5:58 PM on April 5, 2021 [2 favorites]
posted by schmod at 5:58 PM on April 5, 2021 [2 favorites]
What else were they supposed to use? Symbian? (BARF)
Symbian was great. For phones that didn't have giant touchscreens. The touch version was very much half baked, though. Google never would have considered using it, however, because it gave far too much of the sort of control to the user that would have made Google's data collection troublesome.
They could have used MeeGo, though. The N9 was a damn slick device. Even Maemo was better than early Android. It wasn't until ICS that Android stopped being total shit.
On the actual topic, though, I'm happy for this ruling, but it would have been far better if the decision had been that APIs aren't copyrightable. Sufficiently motivated API owners still have plenty of room to shut down reimplementations of their APIs by arguing that some specific use doesn't count as fair use. They might well lose, but not before their target spends a fortune on attorneys.
posted by wierdo at 6:08 PM on April 5, 2021
Symbian was great. For phones that didn't have giant touchscreens. The touch version was very much half baked, though. Google never would have considered using it, however, because it gave far too much of the sort of control to the user that would have made Google's data collection troublesome.
They could have used MeeGo, though. The N9 was a damn slick device. Even Maemo was better than early Android. It wasn't until ICS that Android stopped being total shit.
On the actual topic, though, I'm happy for this ruling, but it would have been far better if the decision had been that APIs aren't copyrightable. Sufficiently motivated API owners still have plenty of room to shut down reimplementations of their APIs by arguing that some specific use doesn't count as fair use. They might well lose, but not before their target spends a fortune on attorneys.
posted by wierdo at 6:08 PM on April 5, 2021
Delphi and Modula-3 existed in 2005. Borland sold Codegear (inc. Delphi) for about $30M in 2008.
posted by Monday, stony Monday at 10:28 PM on April 5, 2021 [1 favorite]
posted by Monday, stony Monday at 10:28 PM on April 5, 2021 [1 favorite]
And if my mindblowingly–detailed API doesn't tell you anything about my program logic, why should a higher–level API be any different?
An API might define an innovative class hierarchy, protocol, etc., and I can see IP proponents wanting to argue that such innovations are covered by copyright or patents. And so that would be code that didn't run, and be independent of the specific implementation.
Abstract programs are programs that can be mathematically or algorithmically analyzed for some purpose, but the programs themselves can't be run. In some parts of engineering it's used to do things like prove properties about behavior of a system, etc. So I was thinking of it along those lines; but, even if not, software copyright by definition could apply to a fragment of software, not a whole program.
posted by polymodus at 11:17 PM on April 5, 2021
An API might define an innovative class hierarchy, protocol, etc., and I can see IP proponents wanting to argue that such innovations are covered by copyright or patents. And so that would be code that didn't run, and be independent of the specific implementation.
Abstract programs are programs that can be mathematically or algorithmically analyzed for some purpose, but the programs themselves can't be run. In some parts of engineering it's used to do things like prove properties about behavior of a system, etc. So I was thinking of it along those lines; but, even if not, software copyright by definition could apply to a fragment of software, not a whole program.
posted by polymodus at 11:17 PM on April 5, 2021
>as for example game mechanics cannot be patented
https://patents.google.com/patent/US4738451A/en
https://patents.google.com/patent/US5662332A/en
posted by Heywood Mogroot III at 1:56 AM on April 6, 2021
https://patents.google.com/patent/US4738451A/en
https://patents.google.com/patent/US5662332A/en
posted by Heywood Mogroot III at 1:56 AM on April 6, 2021
For a layman, think of the API as a blueprint that shows the overall footprint of a building, detailed specifications of the entrances and exits, and nothing else.
I think of an API as a standard to allow interoperability. Like the way standardized electrical outlets mean you could just buy any appliance, plug it in, and use it; you can just plug your software into their API.
posted by mikelieman at 4:10 AM on April 6, 2021 [2 favorites]
I think of an API as a standard to allow interoperability. Like the way standardized electrical outlets mean you could just buy any appliance, plug it in, and use it; you can just plug your software into their API.
posted by mikelieman at 4:10 AM on April 6, 2021 [2 favorites]
I'm not even sure how it follows that an API could be copyrighted. Surely it should either fall under patent (a method of doing business or running a machine) or not at all, as for example game mechanics cannot be patented, nor can fashion designs.
It's program rules, not a work of art.
---
The decision addresses that.
The dissent addresses it (wrongly, in my view). The majority expressly declines to answer the question. The reason the question is open in the first place is that the law says software can be copyrightable so long as it meets the basic requirements of originality and expressiveness. Oracle's arguments on this include such delightful notions as (I shit you not) we could've called Math.max anything we wanted, like Numbers.largest, so it's original and nobody gets to copy it without paying us. Legally, to my dismay, that argument isn't that ridiculous.
My enthusiasm for the decision is thus tempered, because it could end up having very narrow effect. It took ages and enormous effort (the District Court judge learned fucking Java for this case) just to litigate the issue of whether the API is copyrightable and the answer we're left with after the SCOTUS opinion is still "well, uh, maybe?"
Also, unless I'm hallucinating, the court finds that fair use is a mixed question of law and fact, which, again, does not seem to go as far as one might hope. Still going to need evidence and testimony on, e.g., the effect on the market, which means experts, which means $$$.
posted by sinfony at 6:18 AM on April 6, 2021 [2 favorites]
It's program rules, not a work of art.
---
The decision addresses that.
The dissent addresses it (wrongly, in my view). The majority expressly declines to answer the question. The reason the question is open in the first place is that the law says software can be copyrightable so long as it meets the basic requirements of originality and expressiveness. Oracle's arguments on this include such delightful notions as (I shit you not) we could've called Math.max anything we wanted, like Numbers.largest, so it's original and nobody gets to copy it without paying us. Legally, to my dismay, that argument isn't that ridiculous.
My enthusiasm for the decision is thus tempered, because it could end up having very narrow effect. It took ages and enormous effort (the District Court judge learned fucking Java for this case) just to litigate the issue of whether the API is copyrightable and the answer we're left with after the SCOTUS opinion is still "well, uh, maybe?"
Also, unless I'm hallucinating, the court finds that fair use is a mixed question of law and fact, which, again, does not seem to go as far as one might hope. Still going to need evidence and testimony on, e.g., the effect on the market, which means experts, which means $$$.
posted by sinfony at 6:18 AM on April 6, 2021 [2 favorites]
« Older from a personal responsibility to a public good | Mastery of blaseball is in the blood. And the... Newer »
This thread has been archived and is closed to new comments
Well, I guess one of them had to win.
posted by Mayor West at 7:57 AM on April 5, 2021 [13 favorites]