Take the Alfresco survey findings with a lot of maple syrup - Open Source Software Up the Stack

« Why just celebrate the 10th anniversary of the Open Source Initiative when we can argue about it | Main | Red Hat rolls out a wider, deeper JBoss middleware line-up »

February 13, 2008
Take the Alfresco survey findings with a lot of maple syrup

The second edition of the Alfresco Software Open Source Barometer was released on February 12 and its claims and some misleading possible non-Alfresco interpretations rapidly spread across the open source blogosphere. The most interesting information I found in blog posts about the survey was that IHOP gives away free pancakes on National Pancake day. That was useful data.

But don’t necessarily blame Alfresco for any misinterpretations. I have cautioned before about putting too much stock in web-input, “opt-in” (Alfresco’s term) select-audience, apparently non-weighted statistics such as these. Alfresco itself recognizes this is an issue and describes the methodology in the second slide in the presentation on the survey that it made at the JBoss user conference.

The Alfresco survey results actually come from two different sources and the “n” and methodology of the most interesting finding is different than the more widely publicized findings based on the “Barometer Survey.” Alfresco’s separate research is based on what Alfresco calls the “Group Deployment Survey.” It confirms a long-time statistic in the IT industry, which predates the modern era of open source software (OSS). That finding: Developers like to evaluate on Windows and deploy to Unix/Linux.

Overall in my research experience that statistic is moving in Windows’ favor over time as more and more Windows servers have become available. But the press is reporting that Alfresco found the opposite. That would truly be interesting but I cannot find a time-series comparison for the Alfresco “Group Deployment Survey” data so I cannot analyze the claim. And to be fair, what Alfresco is quoted as saying is that

“In the 2006 survey, for example, 41 percent of respondents reported that they deployed on Linux. A year later, that number jumped to 51 percent.”

If the switch is from UNIX, that would be consistent with my experience over the years.

Most important, the other publicized findings (Ubuntu beating Red Hat, MySQL beating Oracle, Ubuntu and Red Hat beating Suse because of the Microsoft-Novell patent agreement) can only be applied to people that sign up for the Alfresco Community. These are not necessarily even Alfresco “users” as is widely reported. North America is grossly underrepresented in those that sign up for the Alfresco community (presumably because it is a UK company) as are Oracle database users (which years of research confirm equals 75% of the universe).

It would be interesting to see a cross-tab of just the Oracle database users and just the North American users and see if they differ greatly from the net of the rest.