Like so many things in the general OB/LW project, it's always possible that an otherwise sound practice will steer you wrong. If you come up with a good reason to downweight certain types of bad arguments, you might end up using a similar reason as justification to ignore good but unpalatable arguments. That's why it's hard.