Burn All GIFs Day may be the first time in human history that anyone has ever thought it worthwhile to stage an organized political protest, even a small one, over a mathematical algorithm. An algorithm, broadly speaking, is a step-by-step procedure for solving a problem. In junior high school people learn the algorithm for finding the length of the hypotenuse of a right triangle: square the lengths of the other two sides, add the results together, and find the square root of the sum. Another way of saying this is that one applies the Pythagorean theorem. The GIF dispute is based on another algorithm, this one known as LZW, after the initials of its inventors (Lempel, Ziv, and Welch). First unveiled in 1984, LZW is an algorithm for making files smaller without losing any information. It is simple, effective, and easily understood by computer programmers, some of whom used it to develop the GIF format in 1987.

Because reducing file size is so important on the Internet -- everybody hates to wait while images download -- GIF files are extremely common on Web sites. So are other LZW-based formats, which have names like TIFF and PDF. The LZW algorithm is ubiquitous in cyberspace. Until recently very few people knew it was patented by Unisys. Indeed, relatively few people knew that it was even possible to patent algorithms at all.

For a long time, in fact, patents did not apply to algorithms or to any other mathematical techniques. According to what is known as the "subject matter" doctrine, only physical processes could be awarded a patent -- "mental steps," as patent lawyers called them, were excluded. (If patents had existed in ancient Ionia, Pythagoras would not have been able to patent the Pythagorean theorem.) There were two reasons for the ban on patenting mathematical techniques. First, intellectual-property experts had long argued that mathematical relations already exist in nature. They cannot be invented and thus cannot be patented. Second, patenting mathematical theories would be destructive. Can one imagine the consequences to science if researchers had to negotiate licenses for the use of Newton's laws or Einstein's theory of relativity?

In 1981 the Supreme Court decided Diamond v. Diehr, a patent case concerning a computerized process for manufacturing rubber. The case turned, in part, on whether the use of computer software in the manufacturing process automatically invalidated the patent. The court ruled in favor of the patent. Legal experts quarrel over exactly what the decision means, but the Patent Office decided that Diamond v. Diehr opened the doors to patenting software.

Unfortunately, the Patent Office has had few examiners who knew anything about software. Incredibly, it has awarded patents to such simple processes as putting two windows on a screen, putting a cursor in a window in a way that doesn't erase the data beneath, and recalculating the entries in a spreadsheet -- even though all of these processes were developed by others years before the patent holders "invented" them and are sufficiently obvious to be implemented in a few lines of code.

The activist League for Programming Freedom, one of the organizers of Burn All GIFs Day, argues that patenting software is inherently as misguided as patenting mathematical theorems -- and for exactly the same reason. The counterargument is that Pythagoras could not have started a business based on his theorem. Today, though, abstract symbolic manipulation is the foundation of the world's most dynamic industry. The U.S. Constitution says that the purpose of the patent system is to "promote the Progress of Science and useful Arts." To serve that purpose, is it better to issue software patents or to let code be used without restrictions?

Both sides of this debate could use Unisys as an example. On the one hand, the company has been exceedingly modest about asserting its rights. By and large, it has asked only big companies that make data-compression software to buy licenses -- individual non-commercial users have nothing to fear. On the other hand, it waited a decade to let programmers know the patent existed, forcing them to negotiate terms for products already on the market. Unisys's relative restraint afterward makes no difference, patent-haters argue. No private entity, no matter how benign, should have a legal lock on common ways of presenting information -- it's like giving someone control over the idea of using an alphabet.

In 1984 the writer Stewart Brand observed that "Information wants to be free. Information also wants to be expensive. Information wants to be free because it has become so cheap to distribute, copy, and recombine -- too cheap to meter. It wants to be expensive because it can be immeasurably valuable to the recipient." The result, he said, is a tension that "will not go away."

The battle over the control of information, Brand thought, will only get worse. Its outcome will help establish the ground rules for tomorrow -- the rights of way for the information superhighway, to use the too-familiar metaphor. Because of the growing importance of the Net to political expression, the battle will have important long-range implications for democracy. Gathering on the steps of Unisys, the small band of protesters marking up floppy disks are, improbably enough, the scouts in a technopolitical struggle about which everyone will soon be hearing much more.

--Charles C. Mann