Microsoft has become a sponsor of The Open Source Census, a project started earlier this year that aims to track and catalog the use of open source software in enterprises worldwide, the group announced Monday.

The company's "customers, partners and developers are working in increasingly heterogeneous environments," so participation in projects such as the census is relevant to the "ecosystem" in which Microsoft operates, said Sam Ramji, Microsoft' senior director of platform strategy, in a prepared statement.

It is the latest gesture by the Redmond software giant toward the open source community, which has long regarded it as a bogeyman due to actions like its claim last year that open source software violated more than 200 of its patents.

Ramji, who could not be reached for comment, is seen as a major driver behind Microsoft's gradually warming attitude -- at least publicly -- toward open source and interoperability.

It is important to balance open-mindedness with skepticism when thinking about Microsoft's open source strategy, according to one observer.

"I've met with Sam and there's no question those guys are smart with what they're doing with open source," said Jay Lyman, an analyst with the 451 Group. "They definitely have changed. Is it genuine? Some of it is and some of it may be less so."

Microsoft's involvement could help the census gain interest from larger enterprises, Lyman noted. But at the same time, it may also draw ire from Microsoft's many critics, he added.

In addition to Microsoft, ActiveState, EnterpriseDB, Oregon State University's Open Source Lab and OSAlalt.com have also joined the effort, which provides a tool, from vendor OpenLogic, which a company can use to scan computers and spot installed open source code. The scan data can then be pushed in anonymous form to the OSC's database.

Contributors can get reports that summarize their own use, as well as comparative data based on similar companies' results. Aggregated data untraceable to any company is available publicly.

More than 220,000 open source packages or installations have been found during the two months since the effort launched, according to the site. But as of June 12, only about 1,300 machines had been scanned.

Lyman's firm is watching the census' progress closely to see whether it turns up enough data to provide a useful representative sample. "The theme from all accounts is that open source usage is wildly underestimated," he said. "Maybe we'll get a better sense of that."