Survey researcher here. It's really time we had a come to Jesus moment on political polling in the US, both the quality issues and how they're presented for public consumption (I'm looking at you 538).

To give a relative comparison, most social science empirical work is done on surveys with above 60% response rates. Political polling has ~6% response rates. Response rates are not perfectly deterministic of sample selection and representation, but single digits are not good

We also need to take seriously the consequences of political polling becoming an *industry*. The push for more polls, sloppiness on quality, and the unwillingness to consider the costs to democracy, are likely linked to the *industry* part of this