Stuart Cohen of Business Week says that the common services/support model for open source just doesn't work - because most of the time, you don't need support for open source software. If you don't need support (and thus, don't buy it) - that pretty much demolishes the model that most people assume works.

However, it seems like it's more a matter of degree - he takes RedHat as a successful example of this model:

Red Hat (RHAT), arguably the most successful open-source company, has also found ways to add value beyond supporting the Linux kernel. It adds substantial layers of software on top of the kernel, a solid piece of software that needs little support, in order to provide additional value to its customers. If Red Hat relied on supporting the Linux kernel, it would go out of business simply because the code is so sound.

That sounds exactly right to me - it's not enough to support open source, you have to create additional value around it if you expect to make any sales. That's what we are trying to do with Web Velocity, as it happens - we support Seaside on Cincom Smalltalk, but we are adding additional value in terms of the development model and in terms of database support.