I agree that a rationalist taboo would also work. But you can't always rely on people being willing to taboo their words. It is also best to avoid infer that someone is doing something majorly wrong. By arguing, "You've proven offensive-1, but you need do extra work to prove offensive-2" we allow people to save more face. What's nice is that instead of trying to get someone to retreat from their previously held position, you've extended the goal so that the onus is on them to find a way to bridge the gap. I need to do more experimentation with both techniques though, to determine which one is more effective.