What I wish was understood about election forecasting, and this didn’t make sense til grad school for me but it’s much easier to make make sense if you are a sf nerd and I should have known earlier, is that the “chance of winning [x]”number and the “wins by [x]” #s aren’t related

Or, if they are, they’re deeply and distantly related, like mathematical derivatives

If a Senator has a 57% chance of winning their race, that means 57 out of 100 times if the model is right they get whatever is needed to win their race. That could be 50.00… to 100% of the vote

And if Biden has a 2/3rds chance of winning nationwide as one outlet is reporting, that doesn’t meant he is particularly likely to get 66% of electoral or popular votes

People really need to think about this as a sci-fi many worlds scenario.

I’m not saying every electoral outcome happens in a possible world or that there’s multiple worlds, but that’s the best way to conceptualize that there’s a giant probability tree that’s going to butterfly out, and the quality of the models isn’t 100% based on if they are “right”

Because, and this is what people still miss in 2016, but maybe wouldn’t if they pictured Nate Silver as Tilda Swinton in Doctor Strange, the models *did* predict a Trump victory,in like 47 or something possible worlds, and we happened to open the door or ride the butterfly to one

Since we can’t leap timelines or go back to change things, we don’t actually know which butterflies got us onto a certain path or through a certain door, and so it’s hard to evaluate these models

But I really hesitate to say anything was necessarily even skewed in 2016

The other thing to keep in mind about this that should comfort ppl while also terrify them,but I think mostly comfort relative to the doomscrolling, is 1)the philosopher who said we live in the best possible world bc of God was an idiot, and 2)so is anyone who says it’s the worst

is it possible we live in an ultimately deterministic universe?on a number of philosophical levels, most of which will literally never matter to anyone without a time machine, sure;for all knowledge based purposes we have,we do not live in a deterministic model of knowledge

This is why heroic epics for a very long time but ESPECIALLY in the current era with Star Wars and the like encode their satisfying narrative resolutions with “fate,” and why some of us feel that “grim and gritty,” anyone-can-die stories reflect experience better.

It’s not because everything sucks necessarily - when things are already bad enough, random chance contains the seeds of hope. See Inglorious Basterds (2009) for a literal elseworlds example of this that’s about as blunt as you can get.

(It’s also why Terminator: Dark Fate isn’t in any way a betrayal of the hopeful ending of Terminator 2, because if there “no fate but what we make,” Sarah and John could have screwed up, and they did, making the Dark Fate.)

Anyway, if there’s no mystical energy field guiding our destiny, what we have are the predictions of statistics, and more colloquially, of our own expectations and experiences - but the latter can be colored by trauma, so I personally like sticking to stats.

Getting technical again for a moment, I was confused for a very long time about how “probability” could be a mathematic domain in the way that calculating, like, the number of oranges if you add 2+3 is, bc it seemed to imply time travel. I have an answer I want ppl to get.

(And I want to be clear that this is the actual statistics-field answer, not me just going wild with metaphors or whatever)

“Probability”in math and stats is the likelihood that something is *currently true*.The reason we can use it to make future forecasts is in the “something”

A classical probability claim would be if you say that the Cubs have a 26 in 1000 chance of winning the World Series. Strictly speaking, this is a lie, because you don’t know, for instance, if an apocalyptic plague will prevent the World Series from happening. Just, you know, if

What you really mean, if you’re not Nate Silver (who I met when he still was known more for baseball more than politics, I liked him more back then) &are doing, like, college-level stats and not wizard-level stats, is that if the future is like the past, the Cubs have that chance

Now, you *can* add additional factors to the model, which requires either extremely complex actuarial math by hand (which is why bookies have always been historically good at predictions, bc their predatory business depends on it) or, today, computer-run algorithms

“Big data,” the thing I’m making into my career, is trying to predict as much as possible by incorporating as many factors into your model as possible.

It’s very easy to see its flaws, and even fiction that semi-glorifies it, like Age of Ultron, ultimately concedes it can only have true predictive power (Vision) with literal magic

For instance, not to bang on a drum, but a 2019 big data predictor-man like Nate Silver probably had a complex model of the 2020 World Series which had a different and more “accurate” model of what the likelihood of the Chicago Cubs winning the World Series was.

That model would have been created with things like individual player stats and coaching data at the very least, weighting more recent performance, performance of teams that new players are from, etc., and possibly stuff outside baseball

That stuff would most easily have been developed from correlating his analysis of politics with his analysis of baseball stats, because those are the two things he has information about existing in the past relevant to the Cubs’ performance.

My guess is that Silver has done a lot of work determining the extent that politics impacts baseball, and that baseball impacts politics, too, because when you’re a data scientist you work with the data you have.

So here’s where this gets fun

The Cubs have actually won a fair number of World Series championships *for World Series championships they made it to*. (My mom is from Chicago, don’t ask me why I suddenly know about baseball.) It’s not a *good* record but it looks less abysmal if you know they won 3/11 games