not a stupit guestion IMHO.
I would suggest to backup classic (it's about 8MB .zip file in my case) and then uninstall. It enables you to run again (perhaps you like THC slowdown experiment which will not be available for BOINC) need with no need to install again and re-register. Note that BOINC and classic can run paralel - which makes sence on P4 with HT. But you would propably decise to run 2 BOINC models...
Also, archive folder can be moved to a CD of DVD for futher analysis (IDL, CPView or perhaps specially ad-hoc utilities provided later by CPDN team to test some hypotesis).