That’s absolutely fascinating, thanks for sharing. I am keen to do more research to see whether polling companies have changed their approach since 2016. In their defence, when you consider how close Clinton was to winning states like MI, WI, PA, and FL, they were well within the MoE of any poll.
But, from my very limited knowledge, I tend to only pay attention to state wide polls of >1000 responses. All the polls I’ve used in my article are minimum 1100. The Morning Consult one polled 27,754 and claims to only have a MoE of 1%!