Having recently read Why I Don’t Like Databases, I find myself agreeing wholeheartedly with Stuart.
Further to his second point on a databases good points (the power of SQL), I don’t think that is necessarily a clear advantage now that XPath is available. Given that XPath can run across any XML file, it’s easy to imagine XML files in a filesystem that can be queried with as much power as SQL.
As time goes on, I hope to see the differences between database backends and filesystem backends become more transparent. There’s little reason for it not to happen, and plenty to gain from it.