TO SOME PEOPLE, Richard Snyder is a very clever man. Just over two years ago he took the helm of struggling Texan video network company Forgent, and with one inspired decision seemingly turned its fortunes round.
Snyderâs stroke of genius was to trawl through Forgentâs mouldering patent archive looking for buried treasure. What he found turned out to be potentially very valuable: a 1986 patent that appeared to cover JPEG, the leading graphics format used across the World Wide Web. JPEG has been built into many products, including Web browsers, digital cameras, cellphones and scanners, and Forgent now makes a tidy living collecting the licensing fees.
Yet Forgent arguably didnât âinventâ JPEG. The software was developed as a freely available package by a consortium of academics and companies. Forgent, based in Austin, just happened to own a patent on a method for compressing digital video which was similar to a technique used in JPEG. If Forgent can back up its claim of ownership, JPEGâs reign as an open standard is likely to be over. Meanwhile we have uncertainty for consumers, manufacturers and software developers â plus a scrambled effort to come up with a new graphics format we can all use.
Advertisement
If that sounds like an injustice, prepare for more of the same. Ever since the US Patent and Trademark Office (USPTO) started granting patents on software in the 1980s, computer scientists have complained about their stifling effect. Now Europe appears to be following the US lead. A European directive issued earlier this year would make it possible to patent software in the European Union. If it is adopted by the EUâs Council of Ministers and approved by the European Parliament, member states will be compelled to convert it into national legislation.
Europeâs software community is divided over the proposed new regime. Supporters say it would reward inventors and encourage innovation. Opponents say it will do nothing of the sort and that patents actually inhibit progress. They are preparing a last-ditch stand to keep them out of Europe that is drawing support from leading figures on the American software scene, notably free software pioneer Richard Stallman.
The one thing both sides agree on is that Europeâs rules governing the patentability of software contain a huge ambiguity that needs to be sorted out. That ambiguity is found in a 1973 document called the Munich Treaty, which established the current patent regime. In those days, software was primarily an adjunct to the mainframe it was sold with and mass-market commercial software did not exist. One thing the treaty tried to do was state clearly what can and cannot be patented. To that end it contains a list of definite no-nos: scientific discoveries, mathematical algorithms, musical and artistic works, mental acts and âprograms for computersâ. That sounds unambiguous enough. But elsewhere the treaty apparently contradicts itself, saying you can patent anything at all as long as it achieves a âtechnical effectâ â in other words, harnesses technology to do something useful.
Clash of views
Unsurprisingly, this has led to disagreement over what the treaty means. Organisations such as technology think tank the Foundation for Information Policy Research (FIPR), along with open-source software developers and certain parts of the business community, argue that it plainly outlaws patents on computer software. But Britainâs Patent Office, which has been the most vociferous champion of change, insists it doesnât. Sure, âprograms for computersâ cannot be patented, but that only refers to abstract lumps of code that donât do anything useful. âThis is the crux,â says Laurence Smith-Higgins, a spokesman for the Patent Office. âWhen you are looking at a technical effect, that effect could be gained from using a computer.â
The ambiguity in the treaty is mirrored in the way Europeâs national patent offices interpret the law. Most donât grant patents on software, so developers fall back on copyright to protect their inventions. But some 30,000 software patents have been granted, two-thirds of them by the European Patent Office and the rest by national patent offices, primarily those in Britain and Germany. The British Patent Office says that about 15 per cent of patents it grants today are software-related.
Though superficially similar, copyrights and patents are very different beasts. Copyright applies only to the way an idea is expressed: the exact wording of a novel or the precise code of a software program, for example. A patent is much more powerful, though shorter lived. It gives its owner an absolute monopoly on an idea for 20 years. If others want to use it they have to pay.
For supporters of software patents this is a good thing. Because patents reward inventors, they encourage innovation, the argument goes. As intellectual property lawyer Stephen Carter of London law firm Olswang points out, companies invest large sums of money developing new software. âWhat is wrong with their being able to recoup some of that investment in licence fees?â he asks.
Plenty, say opponents. They raise three basic arguments against software patents. The first is that the nature of software development makes patents a totally unsuitable method of protecting peopleâs work. The second is the problem of âbadâ patents whose scope is far too broad. The third is that patents are not actually the best way to encourage innovation, which after all what the system is supposed to be for.
Stallman is one of the most vocal opponents of software patents. When he was in Britain recently, he gave a talk at Cambridge University in which he explained why he believes software simply shouldnât be patentable. The problem, he said, is that a patent is a monopoly on an idea. Thatâs fine in an industry where a single idea can be converted into a single product â pharmaceuticals, for example, where a patent covers a single chemical entity. But new software products are much more than single ideas: they are a complicated construction embodying many ideas, some old and some new, cleverly put together to perform a task. To write an original piece of software without using at least one patented idea is impossible.
Stallmanâs analogy is to the difficulty of writing a symphony if it were possible to patent musical ideas. Even a latter-day Beethoven would find it impossible to write a symphony that didnât infringe any patents, not because of a lack of original ideas, but because you have to use a lot of existing ideas in order to make recognisable music. The same applies to software. âNobody is so brilliant he can reinvent all of computer science,â Stallman said. âIf he did, he would make something that the users would find so strange that they wouldnât want to use it.â
That leaves software developers with little option but to cough up licence fees for the patented ideas they use. But this is no real solution. A decent piece of software will incur so many licence fees that it quickly becomes a financial liability.
For large companies, the easiest way to navigate through the patent minefield is âcross-licensingâ. If you want to use an idea that has been patented by another company but donât want to pay, go find a patent in your portfolio that is being infringed by them. Then you strike a mutual agreement not to chase one another for licensing fees.
Cross-licensing isnât always so cosy. Sometimes you will find your competitor is infringing so many of your patents you can hound them out of the market. Of course, that works best if youâve got a large patent portfolio to choose from. The overall effect is that software patents can enable large companies to smother small ones. Large companies can better afford the resources required to research what parts of a project in progress might be covered by patents. Then they use their patent portfolio as a weapon.
Ross Anderson, co-founder of the FIPR and a scientist at Cambridge University, says: âWhatâs now happening is that companies amass large patent collections which they use to intimidate new market entrants, and to prevent free software that might compete with their products.â As a result, he says, the patent system provides weak incentives to inventors but strong tools for large firms to stifle innovations they dislike.
The second objection to patenting stems from experience in the US, where âbadâ patents with excessively broad scope are a persistent problem. The most blatant example is a patent issued in 1992 to Comptonâs New Media of Carlsbad, California, which covered retrieval of text, photos and video from a CD-ROM. Competitors objected, saying the technology was no more innovative than indexing a book. The USPTO eventually agreed and revoked the patent in 1994. But other over-broad patents have since been issued. In particular, companies have successfully applied for patents on âbusiness methodsâ, which basically means software for automating long-established business practices.
Oneâclick warfare
One of the more contentious business method patents related to online bookseller Amazon.comâs âone-clickâ shopping, which claimed ownership of the idea of buying goods with a single click of the mouse. Just before Christmas 1999, Amazon.com won an injunction against a rival, Barnesandnoble.com, which grudgingly added a second click to its check-out. When the Japanese Patent Office found âprior artâ the two companies settled out of court, but the US patent technically still stands.
Europeâs patent offices claim they can avoid the problem of bad patents, partly because the European system is more restrictive than that in the US. âThe difference is colossal,â says Jeremy Philpott, a spokesman for Britainâs Patent Office, who is also a former senior examiner. For one thing, the EU is more rigorous in its insistence that the technical innovation to which the patent applies is ânon-obviousâ. Europeâs system also includes a nine-month period after the application has been published for public opposition before the patent is granted, unlike the system in the US, where anyone opposing a patent has to go through the courts.
It also helps that the European system is inherently slow. It can take 7 to 10 years to get a patent granted, compared with just three years in the US. In 10 years a software product could go through as many as 20 generations. That sluggishness, Philpott says, effectively makes most software unpatentable. âA large amount of software development is small, incremental steps that cannot be protected in the timescale in which the patent system operates,â he says. âThe increments are not worth bothering with.â
Even so, Philpott admits there has been a gradual drift in standards in Europe, with the result that companies apply for patents they know wonât be granted on current criteria in the hope the system may have changed in their favour by the time the application is considered. Meanwhile, the application nails a date to the idea, and prevents competitors from getting in first.
In the end, the argument comes down to how best to encourage innovation. Supporters argue that patents promote innovation by rewarding inventors. But according to Anderson, the financial rewards of owning a patent are less than you might think. He says that only three software patents have ever caused serious money to change hands: the cryptography algorithm RSA, which is widely used in e-commerce; the multimedia compression standard MPEG; and âvirtual memoryâ, which allows computers to run more applications than their memory chips can support alone. Itâs difficult to verify Andersonâs claim, as patent offices do not follow up the patents they grant to see what the results are. But it is clear that merely having a patent is no guarantee of financial reward. Itâs how you exploit the patent that matters.
It also seems that patents are not the only â or perhaps even the best â motivator for inventors. In the computer field, the inventions with the most rapid take-up and biggest impact over the past decade-and-a-half have arguably been the PC, the Internet, email and the World Wide Web. None of these were stifled by patent claims. Stallman claims that in the days before software patents took off in the US, everyone disclosed their important discoveries in academic journals, encouraging these ideas to be used freely and built upon. Now hardly anyone does. If anything, he says, patenting has made computer science less open and less innovative. âI saw no particular speed-up in progress after 1990,â he said during his talk in Cambridge.
Lauren Weinstein, director of the California-based pressure group People for Internet Responsibility, suspects that developers are shunning whole areas for fear of infringing patents, stunting possible progress. âWe donât have any real way to measure how much damage this whole environment of patents has created,â she says. âItâs like the dark matter in the Universe â itâs probably most of the matter out there, but you canât see it and you canât measure it.â
Against this backdrop, campaigns have sprung up to stop the new law. The FIPR is actively campaigning against software patents in Europe, as are the Foundation for a Free Information Infrastructure and the Eurolinux Alliance. Eurolinux is organising a Web-based petition against software patents which has attracted about 120,000 signatures so far. The organisation expects to present it to the European Parliament soon, and hopes to sway its final decision.
But one way or another, the campaigns look almost certain to be defeated. Even if the EU directive gets bogged down in parliament, the British Patent Office has another strategy lined up: it plans to lobby for the Munich treaty to be amended to make it clear that software can be patented. âWe would like to see whichever can be done fastest,â says Philpott.
And in any case, companies can just ignore the European system. In 2000, around 1 in every 6 US patent applications was filed from Europe, and itâs not hard to see why. The US is the biggest market, and the USPTO promises much faster turnaround on applications: response within 14 months and issuance within 3 years. So if Europe maintains stricter standards, people can just run off and use the US system to get their patents. Then they can invoke international trade agreements to ensure their US patents are honoured in other countries. âYou create a lowest common denominator situation, where people file where itâs easiest and then use international treaties and agreements to enforce it.â How long before Europe has a Forgent of its own?