Pure open source companies are in the preventive maintenance business, too. Providing a Linux distro, for example, means rolling out security patches, providing installation support, running compatibility tests, producing documentation, pressing media. And you bet every customer takes advantage of these things -- if they just wanted the software, they wouldn't have paid.

On top of that, every so often a customer has an unforeseen, serious problem, and that's when the meter starts running. That's when you hope the vendor has calculated its subscription premiums accurately enough to be able to afford to solve every issue that comes up.

Then the vendor has to pay its development costs: putting out new versions, adding most-requested features. With open source you get a bit of a reprieve, because the community pitches in. But Fleury's thesis is that you need a ringleader: a central body that guides the development process and contributes most of the code. That costs money.

Plus, simply adding features isn't innovation. Innovation means taking risks. It means R&D, which might lead to marketable products or it might not. It's a gamble. In short, innovation is expensive. It's the something extra, like those catastrophic costs dental plans hope they never have to pay.

So it seems to me that, unless I'm missing something, Gates might be right. Without additional revenue from software licenses, "pure" open source companies have two options: either charge support fees that are substantially higher than those of proprietary software vendors, or else forget about innovation.