Gods of the Upper Air by Charles King! a.co/d/fbbUrTa
25.08.2025 06:47
👍 1
🔁 0
💬 0
📌 0
I've found that it often fails / refuses to read the contents of PDFs, especially for longer papers. If there were a way to force the model to ingest the full text, maybe it would be better at summarization?
😅 I think about it as training wheels for human children
Replying to posts on bluesky because that's apparently when my cats want breakfast
My intuition is that additional differences would slightly hurt if the DD is approximately valid. In that case, triple difference is an unbiased estimator + extra noise. I can imagine it would help if you were doing Jon + Ashesh's honest inference and the triple dif had better pretends
I'm currently at some combination of (1) ask other questions, (2) choose more restrictive models, (3) tolerate more uncertainty bc, as you say, what else can we do?
That whole interview was shockingly ahistorical
Idk, definitely not a clean instrument
It also seems like we just threw in the towel, so 🤷♀️
Pretty sure Newton requires the number of units in new development to be prime 🤷♀️
Do you know if there's easily accessible data?
A big chunk of total wages is overtime:
While looking for a job last year, I neglected to consider being a police officer in in the Bay Area (where wages are competitive w/ PhD econ positions) 😶 publicpay.ca.gov/Reports/RawE...
So better than the median report?