But that doesn’t justify using unweighted responses!

These are the unweighted proportions for years to graduation from the FIRE dataset: would Kaufmann infer that the true proportion of college students who are about to graduate surged by 74% and then collapsed by 30% since 2020?

The jump in the weighted proportion between 2023 and 2024 could very easily be downstream of a change in the weighting protocol. I don’t know the details here but I do know that since 2020 a number of surveys moved from 2 to 3 potential responses for gender ID

hey thanks (1) for saying so, I appreciate it (2) in a roundabout way, for your help--I haven't used my personal site for much but I remember referencing at least one post of yours when I first put it up

Short write-up here:

jacobeliason.com/posts/2025-1...

One on hand I agree with this and on the other I’ve been conditioned to feel like these accounts are an important part of the short form text post platform ecosystem

There are tons of these. Letterboxd, Craigslist, Goodreads

The world is much more complex than we can individually comprehend. If systems don't have ways of simplifying that ("attenuating variety" per Stafford Beer) then the variety will still be attenuated, just in a completely arbitrary way that leads to us prioritising completely the wrong information.

In a scenario where FDIC-backed assets are compromised I’m not sure how useful your cash is going to be

apps.apple.com/us/app/5-cal...

Any way we could get the same range on the choropleth shades as the other one?

Also, this is just a weak form of the general observation that a billion is a surprisingly big number. You lose a lot of the unintuitiveness when you start counting at 200m. Two hundred million is a big number too!

I get irrationally annoyed by posts that begin like the quoted one. Don’t tell me what I can’t fathom, I’m good at fathoming

👁️

tsdf?

a guy like that doesn’t worship god, he worships power