Ah! That explanation did the trick, now I finally understand what you were talking about.

This is something my mind had automatically skipped over, as in my model it's "obvious" that harm is "wrong" merely because the human brain is hardcoded to reject/dislike some things by default, like pain and such. I had taken "harms" as being things that fit this definition, which would make the whole thing incompatible with your arguments unless theft were to also be preprogrammed in the brain.

Basically, I had assumed one specific example of this source of wrongness, and made it incompatible with "theft" or other complex behavior models, skipping over the rest. Assuming I now understand this correctly, that is.