Investigating bugs seems to me like the only part of computing that could actually be considered a true science. You come up with theories about why a system is going wrong, and then setup experiments based on those theories.