It's very unclear to me how you seem to expect medical research to be done in the cloud. Overhead is crucial to create and maintain research infrastructure necessary to advance medicine and science.
I work in pharma and we do tons of medical research in the cloud and it was done before in academia; it typically wasn't my own university that developed (on its own from direct funding) the necessary infrastructure, but universities under contract, or NIH itself, or a consortia funded through grants. The university IT department itself struggled with just keeping wifi up, or giving us 10BT networks on a hub in an age when gigabit switches were common.
Note: I was not typical. I set up as a digital nomad a long time ago. A scientist with a lab on campus who pulls power, AC, and other resources from central services, must have some way of compensating the university, and you could argue that instead of a prenegotiated (and amortized) overhead, it could instead be a cost center billing (scientist gets contract, uses campus resources, gets billed directly for their consumption). Don't forget also that because of Bayh Dole, the university gets monetize the IP generated by their scientists (who typically do not see any direct revenue; instead, they have to make a startup that licenses the tech from the university).
My experience with overheads showed that the universities I worked at had the highest (50-75%) while less prestigious regional universities were more like 30%.
Oh! This is new: I just looked up Berkeley's rates now and they explicitly have an off-campus rate that is 25%, while the on-campus is 60%. Apparently the federal government was already capping off-campus at 25%! I also see that funding agencies like Gates, Chan-Zuckerberg, etc, already cap at around 10-15%. Man, the deans must be furious
The rules for off campus rates are pretty strict, and not just “I’m not in my office.” They get used for Extension, stuff like that.
Ironically, prohibiting “breaks” to Gates, Chan-Zuckerberg, etc. and saying the rate they get is the same as your NIH rate would probably go over well. Most people I know in research admin are perpetually annoyed that private foundations act like being subsidized is their right.
> I just looked up Berkeley's rates now and they explicitly have an off-campus rate that is 25%, while the on-campus is 60%. Apparently the federal government was already capping off-campus at 25%!
Off-campus is capped because the grant is then paying for rent on the lab space from the private landlord. Universities don't charge research labs for power, cooling, rent, janitorial, Internet, etc. Maybe they will start now.
I wrote a small novel but lost it due to user error, so here's a hot take with much less nuance from a college math professor. Just like calculators meant we needed and could pivot the skills we ask our students to learn, AI will also shift the paradigm. But unlike calculators, which are very good at what they're meant to do, AI produces BS in the academic sense [0] and if used uncritically frequently produces work that's just nonsense. So I tell students I don't care so much if they use it, but that the students who do tend to end up looking foolish.
As an math educator, I think there's a huge flaw in this study. The investigators failed to follow up to see why the mistake was made. They leap to assuming the player is trying to "deny the antecedent", but I think there's a much simpler explanation: the players aren't reading the instructions carefully.
There's two reading errors I would expect someone to make given this experiment:
1. The instructions that each card has exactly one letter and exactly one number are before the big cards. I bet many players just skipped that instruction.
2. Mistaking the P→Q as P←→Q smacks more of a reading comprehension error than a logical error.
the kids are alright. Reminds me of when I organized a real life version of Pac-Man on the quad in 2006, inspired by this game ran on the streets of Manhattan in 2004. https://www.pacmanhattan.com/
The blog illustrates that such assumptions about what's a sufficient attribution are fraught with danger, so "the smallest professional courtesy" can expose you to a $150k risk
Nice to see JRMF trending on HN. JRMF puzzles and festivals are a great way to engage general audiences of grade school students in mathematical inquiry and problem solving.
I like the phrasing used by https://houseofgraphs.org/ — while there are an infinite number of mathematical objects in any category, there are only "a few thousand that can be considered really interesting."
Happy to have authored A094777, Number of legal positions in Go played on an n X n grid (each group must have at least one liberty), [1] and A269417, Number of Go games on n X n board with no repeating position and suicide allowed. [2].
