I have a question: where would we put single-arm trials in the evidence pyramid?

(Tim you're now an expert on evidence pyramids!)

The vast majority of the time we're dealing with trying to improve existing treatments, so effect sizes are much smaller and more elusive.

I'd disagree with the premise really. There are some situations where huge effects happen e.g. antibiotics versus none, vaccines versus none. These are where science has understood a mechanism and created an intervention that targets it effectively.

Coz they can't get anyone better

Serious point is the level of innumeracy among top.level politicians.