Free Speech Rights for Programmers
Communications of the ACM, Volume 44, Number 8 (August 2001), pp. 23-25

David S. Touretzky

Note: this version of the essay corrects a copyediting error that appeared in the ACM's version, and adds an illustration I created that the ACM chose not to use. The ACM's official version is available here.

Misuse of copyright law is an affront to the rights of computer scientists.

The Digital Millennium Copyright Act (DMCA) was supposed to protect publishers from electronic piracy. Instead it's being used to curtail the speech of computer scientists. Congress had been warned this would happen, to no avail. Now, with our First Amendment rights under attack, we must rise to defend them. But first, some history.

A century ago, many forms of speech were suppressed on the theory that they threatened the social order. Women's rights pioneer Margaret Sanger was indicted under the Comstock Act in 1914 for distributing information about birth control. Political activists who protested U.S. involvement in WWI were serving time for sedition. But beginning with the founding of the American Civil Liberties Union in 1920, a series of successful First Amendment cases gradually broadened speech protections to include virtually all subject matter. As technology progressed, this extended to new modes of expression as well, including audio recordings, motion pictures, and Web pages. Today even the most inflammatory ideas can be voiced with impunity—and in the case of some artists, at considerable profit.

How ironic that the most recent attempt to limit free speech comes from Hollywood. And it's directed at computer programmers. In 1998, Congress made a series of revisions to copyright law known collectively as the DMCA. Section 17 USC 1201(a)(1) makes it illegal to circumvent a technological measure that controls access to a copyrighted work, and section 1201(a)(2) makes it illegal to "traffic" in any circumvention technology, product, service, or device. Congress was assured these changes were necessary to protect works such as DVD movies from unlimited copying. But as Jessica Litman explains in her book, Digital Copyright, since 1909 Congress has been enacting whatever copyright provisions the lawyers for the major content-producing and distributing industries negotiate among themselves [1]. No one is looking out for the public's right of fair use.

The technology that protects DVD movies is an incompetently designed stream cipher known as Content Scrambling System (CSS) [3]. The law, not the cipher, provides the real protection. But movie studios had other uses in mind for the new law. They could eliminate fair use because the law prohibits access to a work in a manner not approved by the copyright owner. They could control which movies customers watch through a region-coding mechanism requiring North American DVD players to reject discs intended for European or Asian markets, and vice versa. They could even force viewers to sit through commercials, by locking out the fast-forward function. Consumers would naturally prefer DVD players without these enhancements. But under the DMCA, using an unauthorized player to watch a movie, even in the privacy of one's own home, carries a fine of up to $2,500. And selling such a player could result in imprisonment for up to five years.

Software DVD players for the Windows and Macintosh operating systems created new markets for DVD movies, but also undermined the studio's access control scheme. Programmers quickly devised tools to gain access to digital content, initially by capturing frames from the video board during playback by an authorized player, and later by decrypting the DVD files. When a decryption program called DeCSS began to propagate on the Net, the studios filed lawsuits to suppress it. They claimed the program was an illegal circumvention device. But in the view of most programmers, code is speech.

On July 25, 2000, I testified for the defense in Universal City Studios, Inc. v. Reimerdes1 (also known as the 2600 case. U.S. District Judge Lewis A. Kaplan had issued a preliminary injunction in January 2000, prohibiting the defendants from distributing DeCSS source or executable files, but not from publishing discussions of the algorithm. The judge recognized only the discussions as protected speech. In reaction to this, I created a Web site called the "Gallery of CSS Descramblers" ( to illustrate the absurdity of trying to draw meaningful distinctions between computer code and other forms of expression.

One of the gallery exhibits was the C source code for the DeCSS decryption routine. What was the court's basis for objecting to publication of this code? Perhaps it was the knowledge that the code could be compiled and executed, while a mere discussion of the algorithm could not. A second gallery exhibit was some GIF files containing a screen dump of the code. These weren't compilable; they were only pictures of the code. Was it legal to publish such pictures? If not, then I had another exhibit to offer: a version of the code expressed in a computer language of my own devising, but with C semantics. Since no compiler existed for this language, should this exhibit be considered "code"? And if I published the algorithm in this form and someone later wrote a compiler for the language, what would have changed? Perhaps what the judge intended to ban was any description of the algorithm in a rigorous formal notation. In that case I had translated the C code line by line into English. Surely this version was protected speech. But any beginning programmer could convert the English description back to C again. So what was gained by banning the C code itself? Finally, I offered a picture of a t-shirt with the decryption code printed on it, purchased from an organization called CopyLeft. I wondered whether the shirt itself qualified as a "circumvention device," and suggested that to afford the studios the protection they sought, the court would have to prohibit my wearing the shirt in places where others might see it.

Media coverage of my testimony resulted in a flood of contributions to the gallery, from visual art to music to haiku, all expressing the decryption algorithm in creative ways. Someone set the code's English translation to music. The song was immediately banned from out of fear of Hollywood's wrath (and not unreasonably, since the studios had by that time sued CopyLeft over the t-shirt.) Later, programmers at MIT produced Perl and C implementations of the algorithm that were only about 450 bytes each—short enough to jot on a cocktail napkin. One of these programs has now been encoded as a prime number. Perhaps this is illegal too.

We lost the 2600 case, although Judge Kaplan said he was persuaded by the portion of my testimony showing that both source and object code have expressive content, and thus deserve First Amendment protection. Code really is speech. But even protected speech may be censored if the government demonstrates a compelling interest, and a judge can be convinced that less restrictive measures won't suffice. In rejecting the defendants' free speech argument, Judge Kaplan declared that the functional aspect of computer code made it more dangerous than other forms of speech, hence more deserving of suppression. Kaplan declined to offer an opinion on the other gallery exhibits.

As of this writing, the ruling is currently on appeal, and the case will probably end up before the Supreme Court. No matter what the courts decide, the studios will not soon regain control over DVD movies. Improved decryption software is available from Web sites in countries that don't (yet) have a local equivalent of the DMCA, such as in the U.K. [4]. And people are beginning to trade movies online, just as they have traded music. Such noncommercial sharing, which Hollywood denigrates as "piracy," might one day alter the economics of the movie industry.

Of more immediate concern is the chilling effect the DMCA has on scientific speech. Already, a faculty member at the University of North Carolina has been forced to take down online class materials dealing with DeCSS. And on April 26, 2001, at a workshop a few blocks from my office, a group of computer scientists led by Ed Felten, a Princeton University professor, was prevented from publicly presenting its research on breaking music watermarking algorithms. The recording industry had threatened the group with a lawsuit under the DMCA [2]. This aggressive misuse of copyright law is an intolerable affront to the rights of computer scientists. If the DMCA's prior restraint on the publication of "dangerous" knowledge is upheld, the precedent will surely bolster the efforts of other would-be censors with their own agendas. To protect its income stream, Hollywood would drag us all down a slippery slope.

Congress never intended the DMCA to diminish anyone's First Amendment rights. Nor did it wish to make fair use impossible. Before the EU states adopt similar legislation, we should demand that Congress repeal not only the DMCA's anti-trafficking provision, but the anticircumvention provision as well.


1. Litman, J. Digital Copyright. Prometheus Books, Amherst, NY, 2001.

2. Markoff J. Record panel threatens researcher with a lawsuit. New York Times, Jan. 15, 2001.

3. Stevenson, F.A. Cryptanalysis of contents scrambling system. (1999);

4. Wilson, D. Who really is getting ripped? Los Angeles Times, Mar. 15, 2001.


David S. Touretzky ( is a principal scientist in the Computer Science Department at Carnegie Mellon University.


1111 F.Supp.2d 294 (S.D.N.Y 2000)

The First Amendment to the US Constitution, superimposed over a prime number encoding of the DeCSS decryption algorithm. In the background is the Bill of Rights, which contains the first ten amendments to the Constitution.

©2001 ACM  0002-0782/00/0800  $5.00

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

The Digital Library is published by the Association for Computing Machinery. Copyright © 2001 ACM, Inc.