Future Tense

What Google’s April Fools’ Day Snafu Says About Software Copyrights

How Google hoped the world would use the mic drop

Alphabet.

If I learned something from April Fools’ Day last week, it is that the fastest way to send humanity into chaos and confusion is by swapping the forward and reply-all buttons on email programs.

It started as just a silly feature: a button for Gmail that added a cute GIF of a mic-dropping Minion to the end of emails. But rather than putting the button off to the side or in a new location, Google put the mic-drop button in the place of the already-existing send-and-archive button. The resulting reports were comically horrific: people claiming they had lost jobs or insulted clients by accidentally mic-dropping rather than archiving.

Most of the coverage of this incident has been about the tone-deafness of a company in power. But there is another story here: a story about how dependent we as technology users have become on interfaces. That dependence becomes more and more worrisome as companies try to claim ownership over those interfaces.

Interfaces are the in-between space where entities communicate. The most obvious is the interface where human meets machine. A person controls a television by pushing buttons in a certain well-defined order—a remote control interface. That person might control a computer program through buttons and text boxes—a graphical user interface. Or that person might write computer-readable sentences using a set of command words and a grammar—an application-programming interface.

But the human-machine interface is not the only type. As I have explained previously, human languages (and non-human languages like Klingon) are interfaces for communication between people. And computer programs speak to one another through common protocols such as HTML. At bottom, an interface is simply a language.

Interfaces, being languages, demand consistency to work. Words must mean the same thing to both parties the communication, whether those parties are computers or creatures. And the more an interface is used, the more dependent the users become on the assumption that the interface has a consistent meaning. Shorthand words and abbreviations come into speech between friends; muscle memory takes over reading the manual in clicking buttons on a screen.

That means that disasters of various orders happen when one person reassigns meaning to a part of a commonly used language—a mistake in interface. Google changes the action of the button where Send and Archive should be, and people are mic-dropping their dead relatives. Frederic’s nursemaid in The Pirates of Penzance misinterprets the word pilot as pirate, resulting in a very different apprenticeship than Frederic’s father intended. NASA used metric units while its contractor Lockheed-Martin used imperial, and the mismatch in communication between the two sides cost the United States a $125 million spaceship.

So necessary is consistency of interfaces that it is incredibly worrying that companies are claiming ownership of those interfaces via copyright law, saying that no one else may speak or understand the same language. Oracle continues to sue Google for copyright infringement based on Android’s use of the Java application programming interface. I suggested that this theory could lead to ridiculous claims like Star Trek claiming ownership over Klingon; apparently it is no longer a theory, with CBS actually suing over Klingon infringement. And then there is the Cisco v. Arista case.

Cisco makes industrial routers, the computer devices that direct Internet requests from one place to another. Those routers are controlled by a “command line interface”: a set of command words either entered by hand by an administrator or sent to the router automatically through a computer script. Arista makes competing routers, and to save administrators the time and effort of learning a new command line interface language and rewriting all of their scripts, Arista’s routers use the same set of command words. Cisco then sued Arista for copyright infringement.

Arista responded with a fascinating retort: that Cisco, by claiming sole ownership over that command line interface, was monopolizing it in a way that violated antitrust law. Just as Gmail users had become locked into hitting the send-and-archive button from rote memory, router administrators had become locked into the command words that Cisco chose. And if no other router manufacturers can use those command words, the theory goes, there will be a lack of competition in the router market.

This antitrust theory is bold, no doubt, but it makes the key point: Arista cannot just up and create a new command line interface. Any alternative would be inferior, not because it would be less functional or useful but just because of the headaches of word-swapping. If Arista chose different command words, then router administrators could not use Arista devices without substantial effort into relearning and rescripting. Worse yet, if Arista reused some words but changed the meanings, that would invite miscommunications with unpredictable and possibly destructive effects—not something one would want for an Internet infrastructure component.

That risky unpredictability would be multiplied across the hundreds or thousands of interfaces every person comes across every day, were ownership of interfaces the norm. Interfaces must be available to everyone, not subject to sole control through copyright or other means, because we cannot function if they are balkanized and subdivided to all be inconsistent. Because without interface consistency we cannot predict how other devices or other people will understand our words and actions. Because it is only through our shared assumptions about interfaces, about language, that you can understand that I am closing this essay when I recite the words

*mic drop*