A good researcher should be an empiricist. They might want to test whether a pet theory applies to a new domain, but they should be challenging that assumption and open to the possibility that it doesn't. The architects who cause problems are those who dogmatically insist on a particular approach in defiance of all evidence.