ymb: I'd tend to agree, but I think for my part I was just curious. Scientific polling has had a bad reputation lately, and I wondered if maybe an "opt-in" poll like at AB is worth taking more seriously than you might think. For 2010, 2015 and the EU referendum I think AB's poll wasn't that bad in the end, or at least "missed" the true result in a very consistent way, but that pattern didn't hold for 2017.
Still, if nothing else I found it fun to mess around and see what information I could get out of AB polls.