Regarding http://www.onlamp.com/pub/a/onlamp/2005/01/06/svn_homedir.html, I have kept most config files and scripts that i take from job to job in CVS, but have now switched to subversion (for its better handling of directory and file renaming) and offline work. Unfortunately, if not everyone uses subversion, you will be maintaining multiple repositories, or have to choose a main one and a secondary one that doesn’t get updated as often.
However, I wouldn’t synchronize offline mailboxes between computers, and neither the web browser cache. Also, I don’t intend to put binary objects in general in the repository. If they’re important enough to archive or backup, they should be explicitly backed up onto multiple media and periodically verified for media failure. Digital baby photos have a different level of archival importance than a .cshrc for instance. I.e. one you wouldn’t ever want to lose over dozens of years, while another you could lose and still survive.