London

Log off: Gaussian fears that rivals could use its programs to “catch up with less effort”. Credit: C. CUTHBERT/SPL

Chemists who have been banned from using a leading software package because they use competing products have taken their fight online.

The group has set up an anonymous website to draw attention to the behaviour of Gaussian of Wallingford, Connecticut.

The site, which appeared about three months ago, provides a list of scientists who are “banned by Gaussian”, and says it aims to “shed light on some practices of Gaussian, Inc. that can undermine basic scientific ideals”.

Gaussian's software is used by theoretical chemists around the world to predict molecular properties such as bond length and reaction energy. Thousands of site licences for Gaussian have been sold at about US$2,000 each for academic use.

But a small number of prominent researchers have been banned from using the software. Since the early 1990s, Gaussian has denied licences to chemists who work on software that replicates parts of Gaussian's functionality.

Researchers who are issued with licences must also agree not to allow access to banned chemists. At least ten scientists have been denied licences.

“It's like a worker from Ford not being able to buy a Chrysler car,” says Peter Gill, a banned chemist from the University of Nottingham, UK.

The banned researchers say that the company's actions go beyond normal competitive practices. They argue that the restrictions prevent them from acting as referees on other papers, and from properly checking PhD theses, as they are unable to use the software to replicate their colleagues' results.

“I have only reviewed about half of the 40 to 50 papers I normally review in a year,” says Mark Gordon, a theoretical chemist at Iowa State University. Gordon found out he was banned when he tried to buy Gaussian software about 18 months ago. He works on a theoretical-chemistry package known as GAMESS, which is distributed for free.

Gaussian's president, Michael Frisch, says that providing licences to competitors would allow rivals to learn about Gaussian's features without having to develop those features themselves. “Having a competitor's program allows them to catch up with less effort,” he says. He also says that some researchers with competing products have previously published inaccurate comparisons of the speed of their programs compared with Gaussian.

Frisch acknowledges that it is good practice to check calculations during peer review, but denies that it is essential.

“Provided that the method is written up and published I don't normally redo calculations,” agrees Mike Robb, a theoretical chemist at Imperial College London, who has worked on the development of Gaussian.

Gaussian says it has no plans to change its policy. Some researchers point out that the problems faced by banned researchers could be reduced if the many individual programs that replicate parts of Gaussian were integrated into a single product.

Researchers funded by the US Department of Energy are working on a software package that combines some of these programs. They declined to be interviewed.