Tuesday, August 30, 2005

Criticism of "prior art" provisions in HR 2795

But the bill includes a drastic reform no one is talking about that promises to create far more problems than it solves: Tucked away among other provisions is a hopelessly convoluted proposal to redefine what constitutes "prior art," the body of pre-existing public knowledge for which patents are unavailable. This scarcely discussed redefinition of prior art will have significant and perhaps unintended consequences, because it could allow someone to copy an old idea already in the public domain and claim it exclusively as their own, promoting imitation over innovation.

Existing law defines prior art not just as earlier patents, but also ideas made part of the public domain through public use, knowledge or sale, regardless of whether the underlying operation, algorithm or manufacturing process is also understood. Early computer systems and software programs sold or used publicly are prior art, regardless of whether they were patented or their underlying source code ever made public.

The proposed redefinition would change that and require prior art to meet a perplexing new "reasonably and effectively accessible" standard. As written, the bill would allow someone to patent an old idea unless someone skilled in the field could both "gain access to the subject matter" and "comprehend" its content "without undue efforts."

This proposed "clarification" of prior art is about as clear as mud. It breeds uncertainty and, many will argue, drastically narrows the body of prior art. Software industries, in particular, will be harmed by the proposed change because so much of the prior art in software exists outside the patent system.

Take, for example, the 1996 patent infringement case filed by Quantel Ltd. against Adobe Systems, Inc., accusing Adobe Photoshop's "paint" feature of infringement. A jury ultimately found Quantel's patents invalid because they covered basic software techniques developed in 1977 by Alvy Ray Smith, a computer graphics pioneer and co-founder of Pixar Animation Studios. Smith had sold his paint system to Ampex in the mid 1970s and briefly described his digital paint systems in "tutorial notes" distributed at conferences in 1978. The source code that implemented Smith's paint systems was not itself readily accessible, yet it is fair to say modern computer graphics evolved, in large part, from Smith's groundbreaking work. Under current law, Smith's Paint system is prior art. Under the proposed patent reform bill, however, Quantel could have argued that Smith' s early contributions were not "reasonably and effectively accessible" and claimed his techniques exclusively as their own.

Unlike virtually every other reform in the current bill, the definition of prior art was never flagged as a source of the patent system's problems. The change does not harmonize U.S. patent laws with other countries, and the redefinition is not necessary as part of a first-to-file system. If it isn't broken, don't fix it. Unintended consequences from this change could stifle, rather than promote, innovation.

Whereas the meaning and contours of existing categories of prior art are well defined by years of case law, courts would be bogged down interpreting this unnecessary revision for years to come. Sure, another section of the bill tries to give back some of what the redefinition takes away, but it doesn't go nearly far enough. The bottom line is that the redefinition rewards imitation, and would be litigated for the next decade to show otherwise.

Patent rights are often analogized to rights in real property. Prior art, in this analogy, is like public land. Congress should not award real property rights to someone staking a claim on public land, nor should it give patent rights to someone staking claim to public ideas.

by Kelly Hunsaker in SF Chronicle


Post a Comment

<< Home