Scirus is owned by Elsevier Science and is powered by FAST. It searches across the Web and journal literature, purportedly with a bent toward scientific data. Infotrieve is a document delivery company that specializes in science, technology, and medicine. The long-term viability of both can be called into question, as Scirus is close to being a demonstration site for FAST's technology coupled with Elsevier's content. Infotrieve, as a privately held, smallish company, is subject to the vagaries of global economic conditions. Did the government consult with these companies before citing them as the reason to possibly shut PubSCIENCE down? "No," says Wes Crews, Infotrieve's CEO. "We didn't know about it until we saw the notice, but we're pleased that the government recognized we have a better value proposition than they do."

PubSCIENCE launched in October 1999 with the mission of providing free Web search capabilities for journal article abstracts and citations in the physical sciences. Reading the abstract is free, but hyperlinking to the full text generally involves paying for the article. The collection contains over 1,200 journal titles from 35 publishers, including both professional associations (American Association for the Advancement of Science, American Meteorological Society, American Physical Society, American Society for Microbiology, Royal Society of Chemistry, and the Society for Industrial and Applied Mathematics) and private publishers (Blackwell Science, Kluwer Academic Publishers, Nature Publishing Group, Springer-Verlag, and Taylor & Francis Publishers, Ltd.). A few university presses also contribute to the database. Clearly modeled after PubMed, PubSCIENCE wanted to attract scientists and the general public to its information. Noting that the U.S. federal government funds 80 to 90 percent of scientific research and development, DOE touts PubSCIENCE as a significant taxpayer benefit.

The private sector never saw it that way. Since its inception, PubSCIENCE has been a target. Database producers and some scholarly publishers felt threatened by the free availability of peer-reviewed scientific information. The Software & Information Industry Association (SIIA) is on record as opposing what it sees as unfair competition from the government. In testimony on July 11, 2001 before the Senate Committee on Governmental Affairs on the E-Government Act of 2001 (S. 803), SIIA said, "The Department of Energy's PubSCIENCE presents an ongoing example of the inappropriate role of government in providing access to non-government information."

David LeDuc, SIIA's director of public policy, reacted to the imminent closure of PubSCIENCE with pleasure, adding that his association "looks forward to the resolution of this issue." He was unclear, however, about the decision process, stating that since there were still several weeks to go in the comment process, it's too early, he believes, to unequivocally declare PubSCIENCE inoperative.

On this issue, LeDuc is in agreement with the DOE itself. Walter Warnick, OSTI director since 1997, is waiting to make a final decision until all comments are in. However, he didn't sound optimistic about continuing the project and had a tendency to use the past tense in describing it. The language of the notice echoes SIIA's concerns about competition. Acknowledging that PubSCIENCE was always a public/private sector collaborative project, the notice goes on to say: "More recently, private-sector information products have emerged that freely offer bibliographic records to Web patrons. Provider systems such as Scirus and Infotrieve have progressively increased the availability of freely searchable citations, and this trend is anticipated to continue."

Warnick explained the tests that OSTI ran to determine overlap between PubSCIENCE and the two private-sector companies. Although he didn't have details of the actual searches, he said it was a combination of a comparison of journal title lists and the actual queries in which the results were compared. As a result of what Warnick calls "extensive" testing, OSTI found almost 90-percent overlap, making it very difficult to justify the continuation of PubSCIENCE.

"It's an evolutionary process," said Warnick. "When we started, the private sector wasn't as involved. Now they are. When we looked at Scirus and Infotrieve, we had to ask if the government should do this." Warnick points out that there had already been a cooperative arrangement with Scirus and Infotrieve. There are links at the PubSCIENCE site to both.

If PubSCIENCE goes away, OSTI has other projects in mind. Some already exist. DOE Information Bridge contains the full text of DOE report literature. Its PrePrint Network does the same for scientific and technical pre-prints. Since these databases contain documents owned by the government, everyone agrees it's an appropriate role for the government to make these freely accessible over the Internet.

Warnick has expansive plans for other types of information as well. Gray literature is high on his list. The GrayLIT Network (http://graylit.osti.gov) includes the searchable full text of gray literature from the Defense Technical Information Center, the DOE (from Information Bridge), the NASA Jet Propulsion Lab, NASA Langley, and the Environmental Protection Agency. The Federal Research and Development Project Summaries (http://www.osti.gov/fedrnd) contain information about research projects from the DOE, the National Institutes of Health, and the National Science Foundation.

Did the private sector win its battle against competition from the public sector? Neither Warnick nor LeDuc is willing to put it in those terms. However, the ultimate worth of any information product is in its usage. Is PubSCIENCE valuable to the scientific research community? Warnick cites the number of university library Web sites that link to PubSCIENCE. However, phone calls to numerous science librarians in academic and corporate environments failed to find extensive use of the database. If a link exists and nobody follows it, should it count as a success metric? Probably not. If a product isn't useful, it needs to change. DOE has a long history of abstracting and indexing. Although PubSCIENCE may disappear, the agency's commitment to information seems unlikely to become extinct.

Marydee Ojala edits ONLINE: Exploring Technology & Resources for Information Professionals.