A private company of any real size isn't plausibly going to choose Atlanta over Chattanooga to win a prediction market bet. This is a good example of the kind of prediction that can theoretically be prosocial, and one strong indicator that it might be is that an insider bet is helpful rather than harmful.

On the other hand, at the point where the prediction market winnings are material enough that they might alter the underlying decision itself, you've clearly got an antisocial structure. Prediction markets that don't want to be seen as mere prop betting venues should refuse to run markets on those questions.