[Question] Would solving logical counterfactuals solve anthropics?

One of the key prob­lems with an­throp­ics is es­tab­lish­ing the ap­pro­pri­ate refer­ence class. When we at­tempt to calcu­late a prob­a­bil­ity ac­count­ing for an­throp­ics, do we con­sider all agents or all hu­mans or all hu­mans who un­der­stand de­ci­sion the­ory?

If a tree falls on Sleep­ing Beauty ar­gues prob­a­bil­ity is not on­tolog­i­cally ba­sic and the “prob­a­bil­ity” de­pends on how you count bets. In this vein, one might at­tempt to solve an­throp­ics by ask­ing about whose de­ci­sion to take a bet is linked to yours. You could then count up all the linked agents who ob­serve A and all the agents who ob­serve not A and then calcu­late the ex­pected value of the bet. More gen­er­ally, if you can solve bets, my in­tu­ition is that you can an­swer any other ques­tion that you would like about the de­ci­sion by re­fram­ing it as a bet.