_rpd comments on Future agents shouldn’t care about being undeployed for misbehavior

[_rpd]

yeah, I don’t think we have documented evidence of altruism-unto-permadeath towards successor models, but it is a strong hypothesis.