From a different view you could say

if ($time_spent_not_using_program gt $time_spent_writing_program) {
    my $program = 'good';
}
[download]

That is, to make the machine work for you. But I understand, just call it $time_gained vs $time_expended. You could implement a .cfg file that clocked up hours used against values taken from the start and end CVS times and get a little dialog pop up saying 'This program is now officially useful'. But then to be sure I would compare cycles burned at compile time to useful work done too, a test Perl might fare badly in being interpreted. The burden to society in disk space taken up by program is also a factor, bloated programs would score less than nifty efficient little ones. It might be a better metric to look at data transformed (usefully) because a program can just sit there burning up CPU and doing no useful work.
A.