Thanks for this! Looks very useful - will look more in depth later this week. Worth noting the method for weighted quantiles in my blog post you cited is not the same as the version in ggdist (never updated it b/c I wanted to do more testing first, so this is quite welcome)

add the jalepenos to the roll

oh yeah shacket it is

do we call this a shacket or a jirt

Sure but it's not just trolling. There are people who vote like this.

Where did the global post-covid collapse of incumbents come from? Sadies who don't understand the relationship between policy and outcomes but who vote for the incumbent when times are good and against them when times are bad.

I could imagine picking people to show based on difficulty to answer, like folks where a model of vote preferences gives party probability closest to 1/2 (or dem/rep/no vote closest to 1/3)

call it "maximum entropy stupidity detection"

like, fuck, anyone voting based on policy decided who they're voting for before the campaign began

nominate someone exciting and popular who can win on vibes