The Living Thing / Notebooks :

Justice

Joshua Briond, Navigating justice for sexual abuse survivors, when you’re a prison abolitionist and a survivor

Social isolation is not going to solve the prevalent issue that is toxic masculinity, cis-heteropatriarchy, and rape culture. It’s not going to make predators feel less entitled to the bodies they’ve already exploited or wish to exploit.

Scott Alexander has many of the rationalist ethics amusements scattered through his work. Here’s a recent one:

Sophisticus: What about this? I think that people with low motivation sometimes can be helped by reinforcement – including negative reinforcement. But other people think they should be punished. There’s a big difference between simple negative reinforcement and punishment. If you’re just using negative reinforcement, you’re trying to use as little as possible to get the result you want. But when you’re judgmental and you divide people into good and bad, you usually add that the bad people deserve to suffer, regardless of the effect.

Simplicio: This is a strange distinction. Suppose I beat up my wife and threaten to do it again. Shouldn’t I go to jail?

Sophisticus: I think we shouldn’t be excessive about it, and I don’t support mass incarceration, but I don’t want you to get off scot-free, because it seems like that would encourage future domestic violence.

Simplicio: If only there were a word for the sort of thing where we made sure people didn’t get off scot-free in a way that encouraged future crime!

Anti social punishment

Herrmann, Thöni and Gächter found out that participants in some societies were engaging in what they’ve called “anti-social punishment”. They were punishing cooperators![…]

antisocial punishment is harsher in participant pools from societies with weak norms of civic cooperation and a weak rule of law.

statistical justice

The presence and absence of frequentist guarantees in the application of law. Like everyone in a specialised field I nurse suspicions that my field would make the world a better place if only everyone would listen to us.

However, I don’t really have enough expertise to say anything about how we should redesign juriprudence to support probabilistic guarantees.

OTOH, because of the increasing automation of everything, the question of how you should design statistics to support justice (e.g. unbiasedness) is getting urgent - see model interpretation.

See also trolley car problems.