People really need to take these polls seriously, and some misunderstanding of the win probabilities really contributed to everyone’s shock in 2016 and I think their disbelief in current polls.

A 70% chance based on the summation of multiple polls per five thirty eight was Hillary’s chance of winning. Considering the end result was an extremely slim electoral college victory only for Trump, that’s pretty reasonable. I think the problem here is just a misunderstanding of 70% probability, a lot of people thought that implies something way more sure than it actually does. That’s just a slight favoring of Clinton, closer to a 50/50 chance than a sure thing. A 30% chance is like saying, I’m going to get at least three heads when flipping a coin four times. Or pretty close to getting a pair on the flop in poker. It’s really not that unlikely, happens all the time.

It was the NY times upshot trying to copy five thirty eight that had some really bizarre math creating numbers way up in the high 90s of percent that clearly couldn’t be right and especially didn’t help with the false confidence.

It also doesn’t help that those win probabilities often get mentioned in the same breath as polling numbers. 70% in a poll is an insane advantage that would translate to a basically 100% win probability, while a 70% win probability is just a slight edge. I think some people that see those numbers close together can’t help but unconsciously conflate them.

Another important thing to consider is when polling errors happen, they tend to be correlated with each other, not independent. And it just so happened that the polls across multiple upper Midwest states were consistently underestimating Trump’s support. Not to mention a bunch of last minute news events that took place after many of the last polls that could have moved them.

Anyways, it still would be much better to be up in battleground polls than down. We shouldn’t be complacent when there’s in actuality only a slight advantage, and we definitely shouldn’t be complacent when we’re down. These numbers should be a cause for major concern.

Lol, when this first came out I joked about them skipping ammendments they didn’t like. I wasn’t actually expecting them to do it. I mean no one who buys this is going to be spending any time reading the thing.

Hold on let me have chat gpt rephrase that for you.

I’m not exactly sure of the source, but there was a statement suggesting that language models offer three kinds of responses: ones that are too general to be of any value, those that essentially mimic existing content in a slightly altered form, and assertions that are completely incorrect yet presented with unwavering certainty. I might be paraphrasing inaccurately, but that was the essence.