Barron's Business Weekly, Dow Jones, September 7, 1992

Software Patents Block the Path of Computing Progress

By Thomas G. Donlan

Imagine the plight of a musician in a country where certain chord progressions were patented, or a poet confronting a patent on metonymy, or a politician who could not get a license to use hyperbole. Now imagine computer programmers in a country that awards patents on basic principles of software engineering. Unfortunately, you don't have to imagine it; the country is the United States.

An overwhelmed and confused U.S. Patent Office has been handing out patents on crucial software routines. Though the routines have been independently invented many times and now are incorporated in many complex programs, the Patent Office has turned them into property.

The Patent Office has issued thousands of software patents and has thousands more are under consideration, but this minefield is largely unexplored. For one thing, it can cost more to search the patent system for protected parts of a program than it does to write the program. Also, the Patent Office takes years to consider patents, and does not disclose applications until it makes a final decision. In the rapidly advancing world of computers, it is almost inevitable that someone will invent a software technique after the first claimant files and before he receives a patent. The holder can then enforce his rights against any independent inventor.

The owner of one such patent, Refac Technology Development of New York, is actually in the business of pursuing royalties on every spreadsheet program ever sold. Another "litigation company," Cadtrak of San Mateo, Calif., collects royalties on an even more basic concept that makes the cursor work on video displays. Other patent holders are attempting to enforce claims in voice mail, database management, word processing, and, of course, computer games.

Idea Monopoly

The Patent Office only recently began to hire computer science graduates. Most examiners have awarded software patents without expert knowledge of the "prior art" that would invalidate awards, or understanding of what would be "obvious to a worker of ordinary skill in the field." So nobody knows how many patents would withstand exhaustive court challenges. Adding to the uncertainty is a change in the odds. Since patent law practices were overhauled in 1982, patent enforcement cases are easier to win and easier to uphold on appeal because they don't go to the regional federal courts of appeal. Instead, the U.S. Court of Appeals for the Federal Circuit, filled with judges "expert" in arcane distinctions of patent law, hears all patent appeals. The result: two-thirds of patent appeals are unheld, rather than one-third.

In computer software, this is not a promising trend. It does not lead to fair rewards for inventors; it leads to more lawsuits and to retroactive punishment for inventors. Rather than encouraging invention, it discourages creativity with the threat of prolonged litigation. How many programs will be written if programmers assume that they will be sued?

Patents are far tougher obstacles to competition than copyrights. A book on the fall of the Soviet Union can be copyrighted but anyone else can write a competing book of his own. If the book could be patented, the first author would enjoy a 17-year monopoly on the underlying idea. This never happens because books are protected by copyrights, not patents. Copyrights confer a lifetime of protection against specific copying of artistic expression, while acknowledging that there cannot be "ownership" of the underlying ideas.

It took only a few months for programmers to find ways to write a basic input-output program that worked like the copyrighted one in the original IBM-PC, thus opening the market for PC-clones. But Hayes Microcomputer Products, the modem maker, won a patent for the idea of software -- any software -- to switch a modem from transmit to receive. It may be that every modem maker owes Hayes royalties or damages, no matter how the competing gadget works.

There's real power in patents, and real money. Minolta is supposed to pay $127.5 million to Honeywell for infringing a patent on automatic camera focusing; Texas Instruments has earned more then $1 Billion in royalties since 1986, and the big bucks are still to come on its Japanese patent for the integrated circuit.

Software used to be governed more by copyright law, and many critics, such as the League for Programming Freedom in Cambridge Mass., would have it so again. Software copyrights even allow special protection for trade secrets inside programs. But even copyright law can be too slow and abstruse for an industry whose essential nature is change.

Silly software copyright cases are cluttering the courts. Those pitting Apple against Microsoft and Lotus Development against Borland International have centered on "look and feel" of a program because there are many different expressive combinations of words and symbols that can produce the same effect on a computer screen. Is the expression contained in the computer code or in the screen image?

Lotus objected to Borland's Quattro Pro spreadsheet because it offered a layout of menus almost identical to the Lotus 1-2-3. (The name of the Borland program virtually promised that it would, and the program delivers.) Apple sued Microsoft because the Microsoft Windows programs makes IBM and its clones operate like the Apple Macintosh, using a graphical interface, icons, and a mouse. (Apple ignored the damaging irony that founder Steve Jobs lifted many of the Mac ideas from Xerox's Palo Alto Research Center.)

Another case, Computer Associates v. Altai, produced a ruling in a federal appeals court as least as complicated as the software at issue, but one that seems to strengthen the right of programmers to use pieces and methods from other programs. And Sega won a case claiming than an upstart should not have to decode a chip in a game machine to search for clues on constructing compatible games.

The Real World

As much as we believe that the fruits of profits and invention belong securely to the inventor, as much as we believe that one of the most important roles of government is to secure private property, both tangible and intangible, we observe that patent and copyright law aren't working right in software.

The real point of the software business is, or should be, to satisfy customers' needs as quickly and efficiently as possible. No good programmer invents more techniques than he must; no sensible customer wants to software made entirely from new cloth. Progress in computing, and thus success in every industry, requires rapid and painless diffusion of programming ideas.

Companies that spend all their effort defending monopolies on their old inventions won't have the energy to to develop new inventions. Fairchild Semiconductor is a good example. About the same time the it earned all its operating income from royalties, a dearth of new products weakened it to the point of takeover. Sadly, Apple may be another good example: The Macintosh interface is nice, but is it worth defending now? What has Apple done for the user lately?

A new Commerce Department commission is supposed to propose major revisions to the patent, copyright, and trademark law sometime soon. We hope the commission recognizes that the world has turned. New laws must be crafted for software that provide appropriate rewards for creativity while fanning the flames of competition.