Who the what?

A. Suppose there's a standard recipe for people who want to make coffee: harvest and prepare (or simply buy) some coffee beans, grind them up, boil them for a few minutes, and serve.

B. Suppose a company — let's call it Feurig — declares a patterned approach toward following this recipe:

  • Provide penetrable cups of a certain size containing prepared, ground beans.
  • Provide a ring sized to hold the cup, a mounted pin to puncture the bottom of the cup, a mounted injection nozzle to penetrate the top of the cup, and a hinged apparatus to automate these penetrations when a cup is inserted into the ring and covered by depressing a handle.
  • Provide an encompassing container capable of heating water, detecting its temperature, and injecting that water into the cup at a rate suitable for cooking the bean dust.

C. Suppose Feurig then implements this patterned approach toward following the recipe by making cups and a device to accommodate and process them.

D. Suppose further that a competing company with an interest in making coffee notes Feurig's success in the marketplace and creates a different machine — made from different materials, employing a different heating, monitoring, and injection facility, and penetrating the cup differently.

E. Suppose even further that yet another company makes a cup different from Feurig's but consistent with the scale of the holding ring  on Feurig's machine and capable of being refilled with arbitrary contents (such as tea or sympathy).

What is the API?

The API is not the standard recipe (A) for making coffee: that's an obvious practice deeply embedded in the common culture and widely exercised in industry and among hobbyists.

The API is not the device that Feurig made as an implementation (C) of the patterned approach that Feurig had declared, and it is not the competing machine (D), and it is not the alternative cup (E).

The API is B: a patterned recipe-following approach capable of being realized in a concrete implementation.

F. Suppose now that a complex culture of innovation and competition has arisen around the API defined in B, and that a company — let's call it Deploracle — comes along and buys Feurig.

Deploracle argues that its newly acquired intellectual property extends not just to the physical brewing device its wholly owned subsidiary invented, but also to the abstract pattern to which that device and its successors (and many knock-off devices) conform to ensure interoperability, substitutability, and some other seven- or eight-syllable word.

That's sort of like claiming IP rights not only over the particular car you manufacture, but also over the general idea of exposing a latch to open a door, providing access to a seat, and presenting a wheel, some pedals, and a feedback display to enable intentional control of a driving machine– a contingent set of conventions that declare a patterned approach to the general recipe for driving a car. (Adherence to those declared conventions of capability and method ensures that many automobile manufacturers can make a car, that many people can learn to drive a car, and that people who learn to drive a car can thereby drive any car that conforms to the expectations implicit in that training.)

So Diabetes-Benz lays claim not only to its actual line of cars, but also to the very idea of doing a car in that way, simply because they declared that convention when implementing their car.

Does that seem right to you?


Last 5 posts by David Byron


  1. DaveK says

    For your API analogy to be complete, you really need to specify that the size of the ring is part of the API: that everyone makes cups of that exact ring size, so that any cup can be used with Feurig's original machine; and that when someone else tries to make a machine with a ring of the same size, so that it can use Feurig's or anyone else's cups, Feurig claims that the exact size of the ring is a work of great creative imagination, and that nobody else should be allowed to make a machine with rings of that precise size.

    Then the Federal Circuit weighs in, agreeing with Feurig, on the grounds (heh!) that anyone could make a machine with a differently-sized ring, and they would be free to try and persuade everyone in the coffee industry to adapt to their new-and-incompatible ring size.

  2. says

    I had a feeling this was going to end badly as soon as it was filed. For one simple reason: Oracle bundled the patent claims along with the copyright claims in one case.

    That might not seem very important, until you realize that the patent claims caused the case to fall under the CAFC. If it were just a copyright case, then it would probably have been heard in the 9th Circuit.

    Oracle got smoked, of course, on the patent claims. But the copyright appeal was heard by the CAFC. And the CAFC has a long and storied history of expanding the shit out of IP rights and generally disregarding guidance from the Supreme Court.

    So this ruling presents an enormous problem: Google will not get any relief from the CAFC on the copyrightability issue. Even if they consider it en banc, it's not going to get reversed. And if Google prevails at district on fair use (fairly likely), then that does little to solve the enormous uncertainty with interoperation that the CAFC created. And, of course, there will be a good chance that Oracle will appeal that and the CAFC will further gut fair use on that appeal.

    The only way to "fix" it is either through the Supreme Court or copyright reform through Congress. Both of those organizations appear to be completely inept from their recent (8-10 years) behavior.

    So it wasn't surprising, but it's still a huge blow to interoperation. Expect a lot more anti-competitive behavior in the software world, further strengthening the various walled gardens that already lock consumers too heavily into their various ecosystems.


  3. says

    Then comes the fun part: Feurig had picked a particular height of cup that'd conveniently hold the standard amount of grounds, but it turns out the Free Coffee Foundation had been making cups of that same height for years before Feurig started. The FCF claims by the same argument that Deploracle is infringing on the FCF's valuable IP. But if the cups aren't that height they won't work in Feurig machines, so now Deploracle's in the unpleasant position of not being able to make cups that fit their own machines. Zany hijinks ensue.

  4. Xtifr says

    But. but, it was really ha-a-ard! Surely anything that's hard to do deserves copyright protection!? (I have this ditch out back that took me days to dig, and I really don't want anyone copying it!)

  5. Michael P says

    I'm waiting for people to realize that if this recipe/design is protected by copyright, then people need permission to create derivative works — like, say, that cup of piping hot beverage — and that getting this permission might require the maker to agree to all kinds of unexpected and disagreeable terms.

  6. Whit says

    Just wait until the Pepsi/Coca-Cola fight on who owns the pattern of opening a soft drink.

  7. Jimmy C says

    Isn't the Oracle case about copyright, not patents? Google copied the headers files. That's not in dispute. Isn't the issue whether or not header files are worthy of copyright?

  8. says

    The issue just decided was whether the APIs are worthy of copyright. The secondary issue, to be determined, is whether (if so) Google's adoption of them is kosher under fair use doctrine.

    Amusingly, the guy who wrote many of them for Java while working at Sun (which was purchased by Oracle) is also the guy who, while later working at Google, made use of them in creating Android.

  9. says

    @Jimmy C: It's not that Google copied the headers (you can't, because Java doesn't have header files). They wrote their own classes that used the same class and method names, with the same parameters of the same types in the same order as specified in the Java API Specification. Which is what you need to do to implement your own Java libraries compatible with the originals (ie. code written to work with the originals will work with yours). Oracle's claiming that merely doing that, without copying any of the implementation code, is still copyright infringement.

    IMO this is going to come back to bite Oracle if it stands. As decided, simply submitting a query directly to Google (generating the URL yourself, rather than going through Google's home page and entering the terms manually in the search box) constitutes an infringement on Google's copyright on the API for their search service. Their API defines the base URL and the names and meanings of the query-string parameters used to submit a search request, and your program is copying the structure of that API to generate the submission URL without a license to do so from Google. And then we have things like eg. a large chunk of the java.net.Socket class API being identical to the socket API defined in the FSF's glibc, and since Java code doesn't dynamically link to glibc (the JVM might, but the Java library .class files don't) it can't take advantage of the LGPL exception.

  10. gramps says

    A couple points, possibly related.

    Is Feurig worth this hassle? Mrs. Gramps brought one home from Costco last year (thankfully they – Costco – have an outstanding return police) for a test drive. First, yours truly was skeptical about a coffee maker that limited the availability of the raw material; then it became apparent that it needs to remain plugged in to keep the water in the reservoir hot (does Al Gore know about this?); it made one cup at a time, there are two of us who have a cup of coffee before going out to tend the livestock.

    Included was a generic "cup" that supposedly allowed using someone else's coffee in the machine (Folgers, Hills Brothers, Yuban??).

    After farting around with this thing for several hours that night and again the next day, we gave up, put it back in the box (however with the water still in it as there was no way to drain the thing) and put it in the garage to await the next Costco run.

    The Mr. Coffee returned to its position of prominance on the counter. It uses any kind of coffee we buy. It uses any kind of basket filter (or a paper towel if the inventory control fails and we run out of filters). It makes up to 12 cups at one time. It makes the 4 "cups" that fill two mugs in about the time it took Feurig to make one cup, small at that. I thing you see where I'm coming from.

    I am not equipped to discuss the API, and that may be a concept that will impact a wide variety of things. I will leave that to you young-uns to resolve, just don't f**k with my morning beverage.

    [Sorry for the wall of words, it happens some times.]

  11. David says

    For a pretty good discussion of the legal decision itself, see here.

    FWIW, I was a little uneasy with Alsup's initial ruling, as much as I liked it. I really hope that the final outcome is more in line with that decision but I'm not optimistic.

    Maybe Congress could apply eminent domain to the Unix API if the court decisions are unfavorable …

  12. says

    @Todd, the litigation includes use of the expressions "headers" and "method headers" to reference method signatures and concomitant return types. That's where this semantic confusion comes from.

  13. says

    P. S.: The entire intertubes is infringing on that masterpiece of post-hoc creative expression, Roy Fielding's thesis. Oh, and anyone sending information via binary switching is answerable to the offspring of noted dramatist Claude Shannon.

  14. DaveK says

    This stupid ruling ignores the long-standing precedent set by the cloning of the IBM PC BIOS back in the earliest days of the PC market. If it hadn't been legitimate to write your own implementation of the BIOS that conformed to the same API (interrupt numbers for particular functions, registers used to pass arguments to those functions) as IBM's original BIOS, there would never have been a PC-compatible market; software written for one brand of PC wouldn't work on any other, and the entire industry would be fragmented, and a fraction of the size and state of advancement that it is today. We would all be incalculably worse off.

  15. Stephen H says

    Why has nobody ever challenged the ridiculous blowout of IP laws in the US, based upon the clause in the Constitution that allows for those laws?

    "To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries", does not reconcile with "life plus 70 years".

  16. gramps says

    And now you know, if you did not already, that we all do not default to programming and its methods as a first choice.

    However, thanks for the stimulation. A little internet research and several re-readings of the post turned into an educational experience for me. Its tough keeping up with this stuff as the shadows continue to lengthen.

  17. Castaigne says

    Does that seem right to me? It seems neither right nor wrong to me. I see no moral component there whatsoever. It's value-neutral.

    My only question is whether it is legal or illegal. If legal and you don't want it so, then the law needs to be changed.

  18. SPQR says

    The District Court's opinion was a tour de force of copyright law. The appellate court's opinion is a confused hash of nonsense.

  19. says

    Stephen H, got a couple hundred thousand you're not doing much with? That would be the start of what you would need to challenge the Patent Office. Last I looked, the EFF was thinking along the lines of 3-10 Million.

  20. Pablo says

    I really like this analogy. Lol @deploracle.

    I was very impressed by the lengths Alsup went to educate himself on java and programming in general before ruling. Seems like this court skipped that step.

  21. Peter B says

    My take on IP protection for engineering innovation —

    Consider the following to be unprotected IP. My question at the end is “should it be eligible for IP protection?”

    “A Method of Measuring Airflow Using One Thermistor”

    Background: thermistors are resistors whose resistance varies substantially with temperature. In order to measure resistance it is necessary to pass a current through the resistor. That current produces a temperature change in the resistor. This self heating can be a problem or a useful feature.

    Prior art: use two thermistors. One is shielded from air flow, the other is exposed to air flow. Both thermistors are given enough current to experience heating. The thermistor in the air flow runs cooler because much of its self heating is removed by the air flow. A little arithmetic later and air flow is measured.

    This engineering innovation solves the same problem using only one thermistor.

    Step 1) Measure ambient temperature using as little current as possible.

    Step 2) Increase the current such that sufficient self heating takes place. This step may either use a timer or temperature measurements as the thermistor heats. Optionally, both time and temperature may be used together.

    Step 3) Using as little current as possible measure the cool down profile.

    Step 4) Calculate airflow.

    Should the above be eligible for any kind of IP protection?

    I believe the answer is no. The only critical step in this whole process is the asking if it’s possible to perform air flow measurements using only one thermistor. Then consider a sampling of those engineers that understand “Prior art: use two thermistors.” I suspect 95 out of 100 of those engineers will come up with something substantially similar to this “engineering innovation”. Assuming that to be the case, I would say that it’s obvious to one skilled in the art”. OTOH, if only 3 out of 100 could solve this problem then it’s likely worthy of some kind of IP protection.

    Some of that 95% may rather use two thermistors. Perhaps because the “prior art” approach gives continuous air flow measurements.

    In the range between 95% of engineers can solve some problem vs. only 3% who can is a vast unmeasurable gray area.

    (I also say the answer is no IP protection for this because I thought of it and I am only a simple firmware engineer who doesn’t do electronic hardware.)

  22. Jonah says

    Steven H:

    Why has nobody ever challenged the ridiculous blowout of IP laws in the US, based upon the clause in the Constitution that allows for those laws?

    They have. See Eldred v. Ashcroft. (Spoiler: the anti-copyright extension side lost.)

  23. mcinsand says

    I have multiple problems with this, particularly based on Sun's behavior prior to Oracle's purchase. They were happydancing over Google's activities, as well as doing everything that they could to make Java open. Then again, Google seems to have done a lot of work to try to avoid any IP pitfalls. Reword: Google worked hard to avoid any areas that might be potentially encumbered with remaining 'IP' flags.

    Although I am hoping for a successful appeal, since I fall firmly on the side of believing that API's are not IP-able, perhaps Oracle's behavior could drive people to dump a horrible language. On a second pipe-dream wishful thought, maybe they can obtain IP rights to imperial measurement units (feet, pounds, drams, …) and drive people away from them, too.

  24. CJK Fossman says

    Judge Alsup acknowledged having prior knowledge of programming. This helped enable him to see the bogosity of Oracle's claims.

  25. Dan T. says

    As I recall, one of the claims in the successful suit by the makers of the ARC archiver over PKWare (makers of "ARC clone" PKArc, and later introducers of the competing and now-dominant ZIP format after losing this suit) in 1989 was that PKArc emulated the command structure and file format of ARC, thus infringing an alleged copyright over the API details. (There were also other claims including trademark rights to the name ARC, but the user-interface/file-format was among the claims.)

    Some relevant info/links here: http://fileformats.archiveteam.org/wiki/ARC_%28compression_format%29

  26. CJK Fossman says

    @Dan T.
    The suit was apparently settled out of court, so I don't think there's binding precedent.

    According to comments I found, the copyright portion of the suit was over "look and feel." I believe the MS v Apple lawsuits established that look and feel are not copyrightable subject matter.

  27. Steko says

    "Amusingly, the guy who wrote many of them for Java while working at Sun (which was purchased by Oracle) is also the guy who, while later working at Google, made use of them in creating Android."

    More amusingly, Google claimed a clean room implementation with that guy on their Android team.

  28. says

    @Steko And even more amusingly, when called to testify as a representative of Google, that guy had the integrity to throw them under the bus and say, in effect, "Yeah, I probably did copy my old work into my new work. Sorry."

  29. says

    Late is the hour when I got around to this rather brilliant take (capped with the amazing Richard Brooks; Richard Brooks makes [i]everything[/i] better).

    Also, another bouquet of lols for "deploracle".

  30. CJK Fossman says

    And even even more amusingly, judge Alsup declared the few lines of code in question as something a middle-schooler could write.

    The clean room issue is a red herring. Oracle is claiming that the structure and sequence of its APIs are copyrightable subject matter. Those APIs have been public knowledge since the beginning of Java, so clean room issues do not apply.

    Also I don't think Google actually raised a clean room defense. Google mentions the Apache Harmony clean room implementation of Java and admits that some Harmony code is in Dalvik. The Apache Harmony story is too long to relate here. The main point is that Oracle changed direction 180 degrees on the topic of Java openness.

    All of this makes me want to investigate if Scala will run on OpenJDK.