Why do many projects in CS research remain closed source?

I really believe in a "No code, no publish" policy. How can I verify that you aren't a fraud if I can't reproduce your work?