So could economists compensate by going out of our way to punish murderers, rapists, and thieves, since we agree with ordinary folks that these are non-cooperators? Or will we then be evil for punishing such folks too much?
Arnold goes on to trip over his positivism, leaving his head in the sand:
I am not as comfortable as other Masonomists are with using signaling explanations. The opportunity for "just-so" stories strikes me as too great. "Counter-signaling" seems to me to take signaling completely out of the realm of testable theory and into the realm of nonfalsifiability. If a peacock growing a useless tail can be explained as a signal, then what would falsify the theory of signaling?
So broader evidence strongly suggests that signaling is behind lots of human behavior, but Arnold is reluctant to explain any particular behavior as signaling, since signaling processes tend to be too messy to offer as clean strong "scientific" evidence. It seems Arnold would rather focus on less-likely-to-be-true hypotheses whose simplicity allows him to make clear rigorous "scientific" tests.
This seems a clear choice between the status and truth; you can either study the clean hypotheses that let you show your rigor, mastery of sophisticated tools, and affiliate with high status "scientists", or you can study the messier hypotheses that you think are more likely to be true. I choose truth.