The first 2 are very LLM amenable, the last 3 are very mathematical-solution amenable (optimising around issues like that is basically what Ops Research does). I don't see what your argument is here.

The list of people claiming that maths won't work who then get bulldozed by mathematicians is long.