good point,
I did it already this way once because we where producing HTML and PDF files. And to be honest I did it in this way because I wanted to do so. I wanted to try it, if it works and I made a nice frontend to it for the administration. But I "spread" the files because I don't want anyone to edit a document when all his changes would be overwritten with the next update of the content.
Ok, the lost cronjob is a very strong concern, but are there anything alike "triggers" on a unix machine to invoke an action or script - like you have in some highly advanced RDBMS ???
So that is why I would prefer to keep those files in separate places. Ok, another reason is that automtically updated files will/might irritate the readers of cvs logs.

Have a nice day
All decision is left to your taste