Comments on the “Banned by Gaussian” Website
By Michael Frisch, President, Gaussian, Inc.
This web site is a collection of half-truths and outright misstatements. To address every misrepresentation would produce a very long document, so I will limit myself to addressing some of the more outrageous statements and implications.
Anonymity Masks Hypocrisy and Cowardice
In this piece, I will simply refer to “the authors of the website” because they have chosen to remain anonymous. The website has no signature(s) and leaves the implication that everyone listed participated in its construction, but they avoid actually making this statement or otherwise confirming the authorship. Of course, even though they have neither the courage to stand by their malicious statements, nor the integrity to admit their own interests in competing programs, I must respond to them.
All 11 Listed Universities are Using Gaussian Right Now
In cases where there is directly competitive work going on at a University, we always offer that University a license which provides for access by anyone except members of the competitor's research group. The statement that we refuse to license entire universities is an outright lie.
In fact, most of the sites listed on the website have numerous users of our software, including the current version, Gaussian 03. 10 of the 11 sites listed on the website have Gaussian 98, and 10 are already using Gaussian 03. The only remaining site, Iowa State, was offered a license for Gaussian 03 which would provide access for everyone outside of the one competing research group, but has elected not to return it.
While there have been two occasions when we have refused to offer any license at all to an organization, neither of them was a university, and each of them had flagrantly violated their previously licenses. Even Northwestern University, the subject of an article in C&E News after they breached our license, was offered a new arrangement, and researchers at Northwestern are currently using Gaussian 03.
Gaussian Results are Reproducible without Gaussian
The website makes the implication that papers reporting results using Gaussian should not be accepted by referees because they are not reproducible. This is another completely false statement, for several reasons:
Most Calculations Can Be Run in Other Packages. Almost every calculation which can be run by Gaussian can also be done in some other package, albeit not as conveniently or as quickly. Thus, in general it is possible to directly reproduce the results without access to our software.
All Models are Published in the Literature. More specifically, there are a handful of areas, representing no more than a few percent of calculations using our software reported in the literature, in which the authors of Gaussian are leading the field and for which the methods are not yet available in other packages. However, the scientists at Gaussian, Inc. and their academic collaborators take pride in publishing all of their new methods and algorithms in the open scientific literature, subject to peer review. Gaussian's competitors are welcome to code these models and algorithms, but we are under no obligation to help them do so by providing our program to them.
The Actual ACS Standards are Fulfilled. In fact, Gaussian authors' publication practices fulfill the very standard cited in the website from Pure Appl. Chem., which states explicitly that the program used to perform a calculation need not even be available for the results to be publishable. Rather, it requires only that “sufficient technical details should be provided to allow a reader to reproduce the results of any calculation.” Previously, the website's excerpt from these articles included the complete standard definition. Recently, the excerpt has been substantially truncated in a way that misrepresents and materially alters its actual statements. The full text is available here.
Some Groups Choose Not to Distribute Their Code. If access to the specific software package were made a requirement for publication, then no paper using software written in a group and not distributed would be accepted. Do the authors of this website really advocate rejecting all papers from theoretical groups which choose not to distribute their programs?
Publication Before Release Is Standard Practice. Similarly, the policies advocated on the website would also prohibit publications about any new developments put in a package but not yet released. Do they really advocate requiring that all theorists hold up publishing their new methods and results until the corresponding software is publically available? None of the people on the website's list follows this practice now, nor does anyone else.
Sadly, the Cited Standard is Frequently Ignored. For example, journals now accept routinely papers with results from molecular mechanics calculations using proprietary and unpublished force fields. These really are results which are not reproducible without access to a particular software package. Yet the authors of this website choose to complain only about Gaussian, for which all methods have been published in the open literature and whose primary results can be reproduced quite closely by other programs.
Reasonable Protections are not Draconian
The “draconian restrictions” mentioned on the website for these licenses simply amount to a requirement that people acknowledge that the competing group is not to be given access. The licenses for competitor sites allow the same research use and publication of results as our licenses at the thousands of other institutions using our software.
Using Gaussian is Not Harmful
Finally, the instructions on “How to be banned by Gaussian” are downright silly. More importantly, even the misrepresentation on this website cannot manage to imply that applying our software to chemical problems and using or publishing the results would get anyone banned. In fact, the activities they describe are of interest primarily to the rival software package developers listed on the website and not to the thousands of chemists using our program as a tool in their research.
Last update: 18 September 2012