spontaneous symmetry breakdance

8. Do NOT neglect your writing skills. Even if your school doesn't make you take a lot of gen-eds, you still need to be able to communicate.

If enrolling in an extra rhetoric & comp class is what it takes, just take the class.

7. If you need to learn a new kind of math, physics-y sources are usually clearer and more pedagogical than than mathy ones. The best linear algebra textbook is the first chapter of Shankar.
...

6. Get in the habit of using Mathematica.
...

5. Learn a small amount of at least one programming language, enough to translate a numerical algorithm into code. (Yes, even with LLMs. It's fine to get Claude to write the plumbing between the parts that actually do physics, but you shouldn't blindly trust them to choose the right integrator)
...

4. If you want to understand something in one of your classes, go to office hours. If you want to understand something *outside* one of your classes, find someone to recommend a book (or teach it to you). Also, just read random shit.
...

3. Your department probably has a weekly lecture series aimed at a general audience (of physicists), probably called the "colloquium" or "departmental seminar" or something (my grad department just called it "Physics Club"). Go as often as you can. You'll understand more and more as you progress
...

2. Point 1 will require emailing some PIs. If you're not sure how to approach this, ask one of your profs or TAs. And if you're worried about balancing research with classes/work, you should be aware that many (not all) labs will pay undergrads. Some schools have also funding you can apply for.

...

Not sure how generalizable this is to other fields (or to people who aren't planning on grad school), but my advise for physics majors is:

1. Get involved in research *early*. Unless you've literally never taken a physics class before, it's probably not too early to start.
...

(and these are the things you understand to have changed since based on my interactions w profs on platforms line this). Is there a go-to hidden curriculum you have ready to share with such or have some advice of your own? Tagging #ITeachPhysics and #ITeachMath TIA

my preferences at this point are Collins < Platner < Mills <<<<< sortition from registered maine democrats over 30

you're right, I sincerely apologize to all 10 remaining South Park fans

If you're the sort of person who is temperamentally inclined to "douche vs turd sandwich" or "giant meteor 2016" posting, but too smart to equivocate between Republicans and Democrats, I have some great news about an upcoming senate primary.

🧪⚛️ From the Nature article by @dangaristo.bsky.social et al., we know that NSF only got its OMB authorization last week. Still, this is very worrying. If awards fall far behind last year despite a budget down ”only” 4%, it will be a basic defiance of congressional appropriations.

okay, I need to read this, because what the actual FUCK

seriously though it was insane how much pro-Dubai propaganda zoomers get

I'm genuinely at a loss trying to figure out what *Katy Perry* would be using that many Claude tokens for

Max Boot is like “idk, the justification for this war looks pretty flimsy.” This is the most unsold war in the history of wars. That is like Cat Fancy calling a cat less than mid.

I have also found that the Sonnet vs Opus difference is night-and-day for Haskell but not super noticeable for Python, so it could be that more powerful abstractions require a higher threshold of model reasoning ability.

one of the reasons I like this passage a lot is think it captures a kind of small c conservative sense of duty/heroism that's really different from the kind of contemporary understanding of those concepts where it's journey of self-discovery/self-actualization. not, it fucking sucks, but here we are

For human languages, idk, but anecdotally model architecture seems to play a big role. My GF reports that GPT-4o is way better at parsing Latin than any of the Claude 4.5s.

Stress testing Claude's language skills Polyglot or poly-not?

For programming, still an open question, but it seems like corpus size and language properties are both factors. Claude struggles a little with Haskell, for instance

vivsha.ws/blog/stress-...

"Woo is harmless", they said. "Stop imposing your scientism on marginalized people", they said.

i need a fucking drink

A Cookie for Dario? — Anthropic and selling death - Anil Dash A blog about making culture. Since 1999.

Woke 1 logic is that someone who does what you want becomes a target for criticism. Yes, people who do the right thing get cookies. Not "it's the minimum standard" bullshit

www.anildash.com/2026/02/27/a...

i say as if i'm not one myself, lol
(though I was exclusively a mosquito-nets-and-vegetarianism guy until *very* recently)

say what you will about EAs, but at least they have principles

solidarity to Anthropic and the neat little guy they've built

Scofield v. Guillard, 3:22-cv-00521 - CourtListener.com Docket for Scofield v. Guillard, 3:22-cv-00521 — Brought to you by Free Law Project, a non-profit dedicated to creating high quality open legal information.

If you’re following the pro se tiktok psychic Idaho murders defamation case (which you should be), she’s now trying to disqualify a juror

www.courtlistener.com/docket/66679...

If you assault someone, and in the process they receive a traumatic brain injury that makes them a happier person, it's still assault.