[jl6]

Feels like you are drawing a distinction between harms, and recursive harms.

Harmful things can be analyzed; the risks can be quantified and weighed up against the benefits; the downside can be limited.

Recursively harmful things - which not only cause harm but precipitate further rounds of harm - are harder to analyze and have potentially infinite downside.

[ethbr0]

That's a good, simpler way of putting it: summing over all harms, including future ones.

For single act harms, future terms don't exist.

And for recursive, add in a probability weighting, if one wants to be pedantic about modeling.