Last week I noted that the release of OpenOffice.org 3.0 seems to mark an important milestone in its adoption, judging at least by the healthy – and continuing – rate of downloads. But in many ways, success teaches us nothing; what is far more revealing is failure.
And here we have an in-depth study of just such a failure of OpenOffice.org:
we present a case study on the Belgian Federal Public Service (FPS) Economy which considered the use of OpenOffice.org, but eventually decided not to adopt OpenOffice.org as their primary office suite.
The reasons for that failure are the following:
our results seem to indicate that the adoption of open source desktop software in an advanced, data–intensive organization is still problematic. Most studies on the adoption of open source desktop software have focused on public administrations with rather simple and generic requirements with respect to an office productivity suite.
However, the job description of a large proportion of the users within the FPS Economy was very data–intensive. Several users wrote advanced macros and even developed their own applications in Microsoft Access. Converting these macros and applications to OpenOffice.org would be a time–consuming task. The integration of OpenOffice.org with third party applications (e.g., SAS) also proved to be rather cumbersome.
In other words, the principal reason OpenOffice.org was not adopted was Microsoft lock-in. The difficulty of converting macros, and the use of customised apps in Microsoft Access, were the two biggest obstacles.
This shows again why Microsoft's arguments that players like the European Commission should not intervene when setting procurement policy are based on a fallacious assumption that market forces will allow a fair fight.
Open source effectively has one hand tied behind its back by the legacy code that its tightly wedded to Microsoft's products. The only way to create a level playing field is to insist on completely open standards, where Microsoft cannot simply fall back on the need for backward compatibility with its proprietary formats.
And no, that does not include fairy-tale “open” standards like ISO/IEC DIS 29500 (OOXML), where the fact that the big bad wolf may have put on the grandmother's clothes does not change the fact that it is still a wolf – or that it has eaten the grandmother.
Update: By a happy coincidence, the European Commission's IDABC is about to publish its Guidelines on public procurement and Open Source Software, and there are some interesting comments by one of its authors that touch on precisely the problem noted above:
"Many people assume there is level playing field and that measures to promote Open Source are no longer needed. In fact, there is widespread bias in favour of proprietary applications", said Rishab Ghosh, one of the authors of the Guidelines. Ghosh is a researcher at UNU-Merit, a joint project by the institute of the United Nations University and Maastricht University in The Netherlands. UNU-Merit is one of the partners in the consortium, that on behalf of the European Commission, has set up, operates and maintains the Open Source Software Observatory and Repository.
According to Gosh, software tenders often have either implicit or explicit bias for software brands or even specific applications. Of a thousand government IT organisations, 33 percent said compatibility with previously acquired software is the most important criterion when selecting new applications. Ghosh: "This implicit vendor-lock in means that a tender, meant to last for only five years, leads to a contractual relation lasting ten, fifteen years or more."
To be fair, it's not just Microsoft that benefits from this skew – which means that many other open source programs face the same challenges as OpenOffice.org:
Ghosh and his fellow researchers also found many cases of explicit bias in favour of proprietary software. Of a sample of 3615 software tenders that were published between January and August this year, 36 percent request Microsoft software, 20 percent ask for Oracle, 12 percent mention IBM applications, 11 percent request SAP and 10 percent are asking for applications made by Adobe. According to Ghosh in these cases, a government organisation typically asks for a number of licences or a number of copies of a software application. "It is like requesting the latest Volkswagens and then expecting that everyone can sell these. We all know only Volkswagen dealers can do so."