I wrote about the economics of the free software business awhile back (which does not necessarily have anything to do with OSS - OSS is often associated with free, but it doesn't have to be). Anyway - here's the original article I commented on. At the time, I got a bunch of comments, many of them disagreeing with me (but not well, IMO). Anyhow - James Governor sent me a pointer to this article on the subject. Stuart Yeates tries to claim that Forbes "doesn't understand" business.

While I don't worship at the altar that Forbes like to put in front of itself, I'd say that it's stretching things to say that they don't "get" business. Based on this drivel, I'd say that Stuart can't do math:

It maybe true that "only 3% to 5% of JBoss customers buy support contracts" but because it only costs to make the first copy of software, the 95%-97% who don't buy support can easily be seen as either free advertising or as potential clients who are locked into JBoss and locked out of WebSphere.

The cost of copying the software doesn't enter into it. What matters is the following:

1. What kind of overhead is JBoss carrying (staff, buildings, etc)?
2. What are the actual revenues from the small percentage that pay?
3. Are the actual revenues (2) higher than the costs (1)?

The important part is (3). If the costs are higher, then the model isn't working. The original Forbes article makes the assertion that JBoss isn't profitable - which is a strong indicator that the model isn't working for them.

Anything else is an answer to a question no one asked. A loss leader only works if it directs people at something you are making money on. Like, say, GlueCode being a loss leader for WebSphere...