“The Order of Move in a Conversational War of Attrition”, Christian Decker2023-07 (, )⁠:

[Twitter; bad news for AI-debate?] This paper investigates computationally when and how the order of move matters in a Conversational War of Attrition (Meyer-ter-Vehn et al 2018). Switching the first mover flips the debate’s outcome for certain type-realizations and triggers two potentially opposing forces on jurors’ ex-ante expected costs.

In the finite-horizon version of the game, a last-proposal advantage prevails if the jurors’ bias dominates their impatience, and a first-proposal advantage prevails if impatience dominates bias. In the infinite-horizon version, there is an unambiguous first-proposal advantage.

These mechanisms are reminiscent of the Rubinstein1982 sequential bargaining game.

[Keywords: debate, Conversational war of attrition, order of move, Rubinstein bargaining, first-mover advantage]