“RE: Ugly Secrets of Content Management Systems”

What I think they miss is that it's a lot of work to write the underside of a CMS. Sure, Vignette's system has a lousy reputation among those who actually use it, mainly because their workflow system sucks... but what so many companies will find is that it's an awesome amount of work to try to develop the services that Vignette provides in your own in-house product... and then you have to try to support it. If IT staffs were static, that would be fine, but they're not.

That's the point. In my opinion, the questions to ask when you consider developing your own CMS are: "When do I want to start using it?", and "How much resources can I put into it?".

If you've got unlimited resources and don't plan to use it for months, then you could go for it. Develop a system that is built from the very beginning to be very easy to maintain and improve (using adds-on or plugins). And make sure the code is well documented. Also make sure your system is less expensive and is better (or fits your needs better) than every concurrent third party product (because you don't want to spend more to get less).

OTOH, it's almost certain that if you try to build such a system little by little and following your needs, you'll end up with a lot of pieces hacked together. Soon or later, that will be hell to maintain and even worse to upgrade.

In fact, the better question may be: "Will I get soon a better product than what's on the market, and will it cost me less?".