It's hard not to see open source usage surveys in the same light as any other assay of this kind: you can make the numbers say anything you like if you're careful (or not careful). I cracked open Actuate's Annual Open Source Survey for 2008 with this in mind, and while it has its limits it's still an interesting read.

Biggest limitation: the survey only covers four geographic areas -- North America, the United Kingdom, Germany, and France. No mention of Brazil or India, the former rapidly becoming a major public-sector open source adopter and the latter being a massively significant tech economy all its own. I'd guess this limitation was self-imposed to keep the survey focused on comparisons between the above four territories -- e.g., that Germany and France are major leaders when it comes to governmental adoption of open source, compared with the U.S./Canada and the U.K.

On the plus side, the way the survey was compiled shows they asked some of the right questions. When respondents were polled about the use of open source software in their organization, one of the possible answers was "We have a policy of not using OSS", which helped break out the small but significant percentile of outfits determined not to use it. Other questions followed in the same vein, including a specific question about why they chose not to use OSS. "Consider it to be a high risk" was the No. 1 reason. Problem is, the wording of the answer invites debate: does "risk" here mean the legal kind -- as so egregiously misinterpreted by folks like Edmund Walsh -- or the technical kind?

The most interesting survey question and answer for me was "How do you keep up to date with the latest open source software developments?" By far the biggest answer: colleagues, by a whopping 42.9%. "Don't keep up with developments" was the next biggest (32.5%), which implies that there are a great many people who will use open source when they get a good recommendation from someone they trust, but don't really elect to stay on top of what's happening in that space on their own. The numbers for this particular question total up to more than 100%, meaning people were able to supply more than one answer.

Finally, the least surprising answers were in the "what are you using" category. All the usual suspects showed up here: Apache, Linux itself, MySQL, Tomcat, the Mozilla family, and so on. JBoss and Eclipse "have higher penetration in financial services" than in most other business types, although I'm not surprised to see Eclipse taking a full quarter of the pie all for itself -- only slightly less than Apache and Linux proper!

Side gripe: the only way to download the report is to register with Actuate's Web site and provide a bunch of personal information. Aren't we past this kind of thing by now?

Follow me on Twitter: http://twitter.com/syegulalp