that is, it's not a reflection of what useful, collaborative technical work *has* to be like, but rather a reflection of what organizations run according to modern management practices value and support

I will die on the hill that if "the median score of people who use a particular accommodation" is that much higher than "the median score of people who don't use a particular accommodation" it is a sign that the instrument you are using is measuring the wrong thing

(In fact if you see a major bump in score outcomes via accommodations what you are very likely seeing is a gatekeeping event, not a knowledge test.)

CDP itself has some remarkably poorly designed interfaces but I spent a bunch of time working with it at CXScore, so I should still have a general intuition for how to use it.

If you do want to give a try, I'm more than happy to collaborate. It's one of those projects I'll never really do on my own, but would be a lot more motivated with someone else involved :P (Classic ADHD things...)

I started by playing around with making my own CDP library, but looks like there are some bindings already: hackage.haskell.org/package/cdp

One thing I've wanted to try—even started, but got derailed :P—is a simple web automation framework like Playwright. With CDP it would not be too bad to implement, and Haskell could provide a much nicer library design and API than Selenium/Playwright/etc.

Bit late, but just heard about Tom Stoppard passing away :(

sounds like there was room for more procrastination

probably getting back to The Blind Assassin

only book I've read so far is Drive Your Plow Over the Bones of the Dead, so I don't know how it compares to her other works, but I thoroughly enjoyed it and would recommend

Not sure about companies but I used to think this about government programs, and learning some history—as well as current events like the Snowden leaks—disabused me of that notion.

Turns out you can keep juicy secrets secret even with thousands of people involved.

cat

that's basically how I see "intellectual property"

or, at least, certainly commercial copyright and some if not all kinds of patents—trademarks, trade secrets and so on are pretty different

a joyful discovery today: prefix sum `sum[l,r]=sum[r]-sum[l]` can be generalized to arbitrary group, some of which are good-enough hashes which can be used in subsequence searching. this is exactly the rabin-karp algorithm, but online materials about it seldom mention group theory at all

I remember reading some quote about how business organizations don't exist to maximize returns; instead, they exist to do whatever leadership wants, with making "enough" returns as a practical constraint

a profitable business makes less of a constraint

Previously you could quickly filter out low-effort or mediocre work based on qualities of the writing. LLMs break these fuzzy mental filters people have developed, for better or worse.

I expect to see the same thing for cranks in areas where those are a problem.