I have some thoughts on this if you'll forgive a quick thread! (1/) https://twitter.com/DemocracyDocket/status/1381962879612555269

These are great pollsters, and I appreciate the spirit of honesty and cooperation that they're bringing to this process. They're competitors, and this is hard, and it's huge that they're working together. (2/)

The thing that's missing in this quick write-up, though, is the thing that worries me most about polling. This is focused on the accuracy of the horse race numbers - who's winning and who isn't. Understandable, because that's easy to measure against.

But knowing if you're winning is not why campaigns poll. We poll in order to know what to do about it - what messages and tactics work. And that's where I think the crisis in polling is much more profound.

You can make tweaks to weighting and turnout screens and get something right-ish on the topline, but the challenges of non-response and low social trust are magnified with message testing and persuasion targets. The old ways just aren't working.

I loved working in analytics, but one of the main reasons I decided to pull up stakes and go back on the campaign trail is because we need to figure out a different way of running smart campaigns. Our tools aren't working, and I need to be on the front lines to find new ones.

We've seen campaigns completely upend our sense of what's possible in turnout recently. Now we need campaigns to completely upend our sense of what's possible in persuasion. The only other option is giving up.

(end of thread)