Friday, April 20, 2012

Oracle, Google and copyrighting APIs

The second quarter of 2012 has started out with a fairly intense showdown between Google and Oracle.

The conflict?  Java, it's implementation within Google's Android, and if somehow Oracle (the semi-new owners of Java) deserve a chunk of Google's cash over a copyright violation within it's usage.

Whether Oracle does or does not get paid will be resolved in the near future, I'm sure.  But one of the things going on in the courtroom that hasn't been stressed enough is the suggestion by Oracle that the copyright law should be amended to include APIs.

Taking the stand yesterday was Joshua Bloch, Java Guru. Joshua was a former Sun employee, and now currently on the Android team at Google.  Oracle's lawyers were basically trying to get him to admit that his knowledge of Java's APIs (several of which he worked on), impacted the way he designed APIs while working for Google in Android's implementation.  And they were also intent on exploiting the pride that comes with the creativity good API developers possess by getting Mr. Bloch to admit that APIs are a "creative work".  Something that falls within the definition of a copy-writable item.

While I do agree that writing elegant code is a bit of an art form, and can bring about a sense of pride and creativity akin to writing a novel or creating a painting, it is rare that all of this creativity comes from one centralized brain.

Furthermore, while I think that APIs are creative efforts, they are merely components of the bigger picture.  On their own, they don't do much.  If one were to make an analogy to cooking, the ingredients to a recipe would be APIs, and the dish would be the application.  And like ingredients, APIs can be home grown, or bought from a vendor.  But no single vendor has the right to claim they invented tomatoes and that nobody else can grow them, even if they were to somehow convince everyone they were the first to figure out how.  And if they could, how do you suppose that would impact the price of tomatoes?  How many suspicious allegations would get thrown about if someone came up with a fruit that's similar, whether it was done honestly or not?  And how many great dishes that depend upon tomatoes would have never been invented because of the controlled availability and fear of legal action?  Or what if an artist could prevent other artists from using a particular brush stroke because they "invented it".  APIs are nearly on that level of abstraction and expression.

Well written APIs, as I'm sure Bloch would agree, usually follow a number of well established patterns and practices.  In fact, most APIs are mix and matches of already established ideas that have been shared within the development community on a global scale.  This has allowed us to grow from simple monochrome text calculators to satellite powered, complex mapping applications on your cell phone in the short span of 15-20 years.  All of this progress is the result of building upon existing knowledge.  Which is the very essence of API design.

If suddenly entities can claim ownership to this knowledge because it's implemented in a copyright-able API, this brings enormous liability to just about every development effort out there, past, present and future.

I can virtually guarantee that most APIs published today have portions of identical code in them.  Sometimes there's only one logical way to solve a problem, and if methods can be found to improve the process of solving that problem, developers should be free to make those improvements.  Both for the benefit of the end-users and the company providing the service facilitated by the developer's knowledge.

Under a "copyrighted API" law, a huge mess of "I did it first" litigation could become the new way to patent troll.  And it could easily block the progress or birth of the next Facebook or Dropbox.

And not only that, legally responsible release of software would imply proper copyright protection would have to be implemented on it before it sees the light of day.  And to be completely responsible, any changes to improve the process would require a fair amount of research to determine if your improved version doesn't somehow violate someone else's copyright.  And if you had duplicated someone's effort?  You'd have to throw it away.  Wasting the time and money getting you to that improvement.

This law would make software development cumbersome and nearly impossible to do legally.  I know I've personally written code without knowledge of someone else's work, only to find out later that it was nearly identical (differing only by a couple interchanged lines and different naming patterns).  And I'm sure other developers have come up with code similar to mine without even knowing what I'm working on.  Which means a lot of most of us come up with very similar solutions when presented with the same or similar problem.  Guess why?  Education.  Just like how doctors are shown and trained in certain procedures, developers are trained to approach problems in a similar way.  Independently ask a group of developer to write a function/API method to swap two numbers.  I bet you'll find that implementation to be nearly identical from person to person.  Helpful sites like StackOverflow  are potentially a HUGE liability given how many functions exist there that people just copy into their own projects.

So to what degree should an API be eligible for copyright if that's the case?

Such a law will stifle innovation, and slow down the progression of software and systems development we have grown to expect by forcing a legal dependency into the process.  Oracle better be careful.  Especially given how developers jump from company to company.  Suppose a Microsoft employee on the SQL team went to work for Oracle.  Could Microsoft make a case of copyright infringement based on the relatively safe assumption that some of this ex-employee's knowledge cultivated at Microsoft may have landed in Oracle's codebase?

It is a shame that Larry Ellison has lost touch with his roots.  He himself didn't invent the notion of relational databases, and in fact built his company on the ideas of Edgar Codd.  He also built his company based on the knowledge he gained from a database project while working at Ampex.  Surely if an API copyright law were in place back then, his former employer might have suspected some of the code he created there made it into his product at SDL (former name of Oracle Corp)  And he has clearly forgotten the frustration he had over attempting to work with IBM over System R compatibility and how stifling that must have felt at the time.

I think it's safe to say that Oracle may not even exist today if such a law were in play at that time.

As a developer, I'm watching this case closely.  If the end result culminates into a new copyright law, the impact of this case will be far more than Google having to pay some sort retro-active licensing fee to Oracle.  It will affect how we all develop software.  And Larry, if you're reading this (haha.. right).. PLEASE..  Get your check from Google if you must, but realize that APIs have existed outside of copyright since nearly the dawn of computers themselves. And you have benefited greatly from it.  Please don't burn the bridge behind you.


  1. Copyrighting is the process of protecting ones intellectual published or unpublished work.Initially only books were copyrighted but now music, paintings, literature, drama, story also included. If an artist copyrights his work, it gives only him the right to use the work, and others, only if he sells the copyright.

    1. Correct. And that is (was) the biggest issue with Oracle's case. They were trying to claim ownership to some very abstract concepts that would to a large degree put nearly everyone in a position of violation. And with that precedent in place, what would stop Microsoft from doing the same kind of thing? Or Apple? Or Google?

      Slowly development would get walled up to the point where you wouldn't be able to publish any kind of API without running it through some sort of legal entity to protect yourself.

  2. Sadly, someone does own the rights to tomatoes and they are called Monsanto.

    1. Yeah.. I forgot about Monsanto. They are WAY worse than anything we've seen in software development.

      At least if I'm writing an API, to a degree I have control over how much I'm willingly violating some sort of patent. I can even re-do it after the damage is done to not violate the patent. It's not like wind, bees and animals and other things out of my control are pollinating my code with patent encumbered elements and leaving me open to patent trolls. (Though I suppose non-controlled open source projects could end up in a similar predicament.) :)