> While it's a nice idea, I don't think it works in practice. You're
> trying to measure indirect effects, and if that were in any way
> possible, marketing would be an engineering science.
I agree, measurement of FOSS use in business is very difficult.
The Internet has clearly had an enormous yet difficult to measure impact on many buying and selling activities. For example, there is anecdotal evidence that many eBay users consult items for sale merely to estimate the going price for something, before placing a classified ad in a local newspaper. Ebay countered this by limiting access to terminated transactions to members only. Similarly, it is said some visit Amazon simply to obtain the ISBN number of a book so they can go to their local bookstore to buy it (or even their local library to borrow it). In both these cases, no sale was made benefiting the Internet sites which provided the information; yet there was an (unmeasurable) impact on a local economy.
In the case of FOSS projects, it's notoriously difficult to measure what's going on in companies; the usual reliable indicator is a press or consulting firm survey of CIOs in which they remain anonymous and they are asked about current installed base and future purchasing intentions, etc.
The Internet itself is not so reliable for measurement: for example, Netcraft tracks web server usage and publishes statistics on Apache versus IIS, but doesn't take into account parked domains. A recent switch to IIS from Apache of a large US service provider caused some flutterings in the blogosphere until it was revealed that a large majority of the domains were merely parked, which means of course the web server is not even used.
The current situation is a clash of ideas. We are still in a period where education is necessary -- how FOSS is structured compared to traditional proprietary offerings, the danger of software patents to innovation, the DRM mess. We can agree that the rapid growth of GNU/Linux in business is incontestable; after all, Red Hat, Novell/Suse etc. can count the number and size of support contracts sold -- so it's clear that CIOs are getting the message. As for *how* they are getting the message, I think it's most likely from the Internet, probably from the specialized press illustrating business cases, possibly their usual integrators making the transition, maybe even from consultants (who in general I think have been slower than business to understand the industry changes underway).
Microsoft built their empire in business (aside from cheating through predatory and illegal business practices) by supporting developers, ISVs, and integrators, the front lines of business IT solutions and support. However, as a marketing company, tempted by the easy margins of license-selling, they made a critical mistake -- they left *all* business process specialization to third parties, a mistake IBM, HP, Sun etc. never made. As a result, Microsoft is overdependent on its business partner ecosystem. In this context, perhaps a useful metric would be to calculate percentages of self-described FOSS-oriented integrators over the past three years; I believe integrators have faced, and proposed, the choice between FOSS and proprietary sofware for longer than many non-IT businesses.
Sean DALY.