Why I'd rather roll my own CMS

Just when I was feeling silly about building my own blogging CMS I found two posts that reminded me why I like to keep my data in my own system. David Weinberger recently completed the "upgrade from hell" with MoveableType and Scoble is throwing a hissy fit over Wordpress' RSS feed. This is why I don't feel comfortable with other people managing my data. I'd rather do the work upfront than get caught later in somebody else's database.

This brings up a key problem with web-based apps. When a desktop app fails or a major bug shows up, I can at least find some product or write some code to get to the data and transfer it to some other tool. What happens when my data is in a database on some server somewhere? Of course, you have to rely on remote hosting for lots of things. I don't want the hassles of running my own webserver at my house, but I keep a complete working mirror of any site I build, including all the supporting scripts on one of my machines at home. Once my blogging system is moved to Ruby/MySQL I'll certainly host it remotely, but it will still be running my code.

The average user has to rely on other people's code, and it will take years before that trust is solid. Yes, I know that people use remote blogging systems every day, but that is still the leading edge of users. Before we all move to web versions of Office tools, a lot of thought has to be given to recovery. I'm going to assume that any reputable host will keep back-ups, even though that has been proven false also. If users discover that upgrades and new "features" cause them to lose data, there could be a backlash that sets Web 2.0 back a few years. When Word fails, people may curse, but they don't give up on Windows. If a web version of Word fails, they may decide to take their data back where they feel more control.