Bear Left!

The Copyright Machine
Tim Francis-Wright

29 July 2001

The news this month of the arrest of Dmitri Skylarov, a Russian computer programmer, confirmed for many mathematicians and computer scientists that copyright law in the United States is fatally flawed. Skylarov faces criminal charges after arriving in the United States to attend a computer science conference. He had written a computer program that circumvented the copy-protection scheme on Adobe E-books. The law that allowed for Skylarov's arrest is the Digital Millennium Copyright Act (DMCA) of 1998. It is just one example of an intellectual property system that protects weak ideas, inhibits intellectual progress, and lessens the rights of consumers.

The DMCA allows civil actions against offenders, as well as criminal proceedings, with fines of up to $500,000 and jail terms of up to 5 years for a first offense. Adobe dropped its civil action against Skylarov after protests from thousands of scientists, but Skylarov still faces criminal charges. One notorious section of the DMCA makes it a crime to circumvent a password or other protection scheme even to make a copy normally protected under fair use doctrine. One example of prohibited copying would be copying an Adobe E-Book to a format that a blind person could use, either through a text-to-speech program or a Braille printer. Another would be copying an E-Book from the computer that downloaded it to another computer, and then erasing the E-Book from the first computer. A third would be downloading a DVD movie onto the hard disk of a computer system if that system lacks a DVD driver. All three examples were permitted under existing copyright law, until the DMCA made them each illegal.

It is more than reasonable for information industries to be worried about digital copies. Digital copies of music, books, and movies have higher quality than analog copies. Digital copying is often faster than analog copying. Digital copies are inherently more usable than analog copies.

But those worries do not require the bludgeon that is the DMCA. To allow information providers to sell their products without fear of rampant copyright violations, the federal government could simply have relied on existing copyright laws. These laws prohibited most copying, except for strictly defined instances of fair use by purchasers. For example, fighting Internet sites from trafficking in E-Books or DVD movies requires someone to watch for pirated material. It does not require extra protection for those files.

Those worries should have encouraged companies like Adobe to welcome attempts to crack their encryption schemes. Companies that want to allow limited copying of their products should want to make sure that the algorithms used to encrypt and decrypt the data are impossible (or at least very difficult) to circumvent. When Skylarov was about to demonstrate to a computer conference that Adobe had used weak encryption schemes, the law allowed Adobe and the federal government to stop him. Adobe appears to be embracing a result in which it is left with a poor encryption scheme that will not stop any real miscreant for very long.

The most chilling consequence of the DMCA in crytpographic circles is that good mathematics might lead to arrest. Although the DMCA allows for encryption research without penalty, in the real world research and commerce can and do meet. Publication of an article that shows how to defeat an encryption algorithm might be tantamount to showing how to defeat the encryption scheme of a commercial product.

The Skylarov case is not the only recent case involving the DMCA. The Secure Digital Music Initiative is a consortium of almost 200 companies and organizations that are trying to limit copying of music by technical means. A team of computer scientists and engineers from Princeton and Rice University participated in a challenge by the SDMI to remove encoded digital "watermarks" from music files. A watermark would allow the user to make one copy of a CD or another medium, but not to copy the copy.

The Princeton-Rice team succeeded in removing each of four proposed watermarks without "significantly degrading the audio quality of the samples." When the scientists submitted their findings in a paper to a respected computer science conference, their paper was accepted. They did not, however, present the paper at the conference because the RIAA, the SDMI, and a corporation that developed one of the watermarking schemes threatened to sue the universities and researchers.

The SDMI is using DMCA to prevent the light of day from shining on the mathematical facts that show that its current products are flawed. Rather than accept that digital watermarking needs to work well in order to prevent music piracy, the SDMI would rather stick with the flawed system that it has. Meanwhile, the researchers have had to resort to a lawsuit to get their paper admitted to another conference later this year.

The DMCA is part of a constellation of laws that are the result of Congress doing what big media wants. The two political parties are jointly responsible for a system that makes certain mathematical algorithms criminal in nature, attaches absurd lives to copyrights, and makes snippets of computer code patentable.

Article I, Section 8 of the United States Constitution requires that Congress "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." Originally, copyright for written works lasted for fourteen years, with the possibility of renewing copyright for another 14 years. Eventually, both of those timeframes were doubled. The Copyright Act of 1976 made a mockery of the "limited Times" phrase in the Constitution by extending the copyright terms for works created after 1977 through 70 years after the death of the author. The terms for works for hire, anonymous works, or pseudonymous works were the shorter of 95 years from publication or 120 years from creation. The Act created some enormous lives for new works. This article, for example, cannot enter the public domain for another 120 years, assuming that I live until the reasonably ripe age of 85 years.

Imagine Shakespeare writing his plays under a copyright scheme like that in America today. Cymbeline, King Lear, an Macbeth all rely on the 1577 publication of Chronicles of England, Scotland, and Ireland by Raphael Holinshed, who died in 1580 or so. Under American copyright law, Shakepeare could use Holinshed's work, but not before 1650 (when he would be 84 years old, but dead for 34 years). Many of Shakepeare's historical works rely on The Union of the Noble and Ilustre Famelies of Lancastre and York by Edward Hall. Hall died in 1547, one year before the publication of his chronicle, so his chronicle cannot enter the public domain until 1617, one year after Shakepeare has died.

Congress compounded its error in 1998. The Sonny Bono Copyright Extension Act of 1998 extended copyright terms to 95 years for most pre-1978 copyrights. (Expired copyrights that were issued between 1950 and 1963 for a 28-year period but were not renewed were not affected.) The Extension Act made a further mockery of the Constitution. Extending existing copyrights did nothing to "promote the Progress of Science and useful Arts," because every affected work already benefitted from the pre-1978 copyright scheme. The Extension Act did do one thing: it continued the Congressional practice of keeping all of the old Walt Disney cartoons out of the public domain.

The Patent and Trademark Office is notorious for issuing patents that are silly at best and dangerously flawed at worst. James Gleick wrote an article for the New York Times Magazine last year that excoriates the state of the patent system. Some patents are just plain stupid--for example, 5443036, "Method of Exercising a Cat" by having it chase the dot made by a laser pointer. Fortunately, it would be difficult for the patent holder to assert patent infringement in court and get away with it (for example, cats have played with flashlight beams for decades.)

The recent rash of software patents and business-method patents has the potential, as Glieck explains in detail, not to encourage, but rather to choke, innovation. Many software patents go to companies or individuals who were astute enough to patent the software process first, and who then shake down other companies for licensing revenue. Even when these patents have merit, the existence of the software industry belies their real usefulness. Microsoft started selling software shortly after its founding in 1975 but did not receive its first software patent until 1988. One software patent out of thousands is the one held by Unisys for the LZW compression scheme, used in the GIF files on many websites. LZW is just one of many compression schemes, so the only effect of the patent is to make some money for Unisys and to raise the hackles of some computer scientists.

Even worse, in 1998, an appeals court overturned a lower court's ruling that a business method was just a mathematical algorithm and could not be patented. As a result, one could patent abstract ideas, as long as the ideas were useful. Through this awesome act of reification, the appeals court opened the door to such jaw-dropping inventions as Amazon.com's patent on the one-click ordering system. It should be ludicrous that the patent office is granting 20-year monopolies on trifling pieces of computer programs, really nothing more than mathematical algorithms.

America does not need this ridiculous host of intellectual property rules. The personal computer industry started and flourished in America without the benefit of software patents. The American economy grew strong and nimble without the benefit of patents on business methods. American authors wrote and ate well without the need for royalties over a lifetime plus 70 years. For once--and perhaps only this once--I say roll back the clock.