As the post mentions, RD participants have an incentive to argue dishonestly. They also have little incentive to say anything informative at all. To solve this, I'd propose Paranoid Debating: everyone is scored on the correctness of a team estimate, except for one participant who's secretly designated an Advocate and one participant who's secretly designated a Naysayer. The Advocate gets more points for higher team estimates and the Naysayer gets more points for lower team estimates. Variants: give points for figuring out who the A and N are, or let it be known publicly.