mwt

> If progress is as fast as LLMs

Is this the new "x technology is a year away" ?

Could you provide examples of healthcare executives held personally liable for harm resulting from reckless decision-making? I have never heard of such a thing happening in healthcare so framing CEO responsibility as a solution to the problem sounds like a stretch to me.

Some examples: Elizabeth Holmes got canned for lying to investors, not harming patients. Purdue Pharma plead guilty to misleading regulators and giving doctors kickbacks, not causing some hundreds of thousands of opioid deaths, but no Sackler family members were personally tried.

Pharmaceuticals are heavily regulated, the "we vibecoded a therapeutic and released it without testing" hypothetical has no basis in reality

Would Doom give a sociopath instructions on how to commit a mass shooting if asked? I don't remember that part of the game

Only through the first two paragraphs but a little turned off by the "everybody else is wrong, we are right and it's this one specific thing" attitude when it the topic is understanding something as complex and opaque as the global economy

The cheminformatics I do (mostly drug discovery/biophysics) definitely requires bonds!

I'd be really interested to know of anybody making money with those topics (and doesn't already have their own domain-specific practice for the problem)

This code is jibberish to me, but it appears the target is just parsing how many atoms are in a molecule string of some representation. That's cool, but to do just about anything useful in chemistry we need the bond graph (and often more - bond orders stereochemistry, plus much more for biopolymers).

Much like the classic "our department in [big company] works a lot like a startup"

I'm a little surprised to see you dismiss type-checkers disagreeing with each other - there are more than a few cases of mypy and pyright disagreeing and brushing them aside as rare enough to be irrelevant is in conflict with my experience over the years.

I'm happy you believe that the community can converge on unified standards - I wish I shared the optimism - but that's not where we're currently at and years-long efforts don't help me now. (For example, I'd love to use PEP 695 generics since I found a use case for them around 18 months ago but I can't until I can get away with not supporting 3.11, which is years out.) Maybe everything is perfect in 2028 or so - I'd be thrilled - but that doesn't help the pitch for annotations being a value add for people who are worried about the current jobs.

> as for the split in type checking tools, it is not as bad as you think

I'm surprised you already know it's not as bad as I think - have you been able to use it? Working on a team that mixes mypy and pyright is pretty frustrating since they don't agree on everything (i.e. when one changeset passes on one and fails on the other) and I see no reason to believe the inconsistencies will become more rare when the number of opinions goes from two to three

Unfortunately that doesn't answer the question

thanks, I didn't know it was a thread because of the login wall

kinda mixed on this

- why not just contribute to the community tool?

- there's already a major split in Python type-checking tools, if there's a third that doesn't agree with either of them it'll be a mess for projects to deal with

- astral has been hiring like mad recently and has yet to communicate that they can actually make money ($5 million doesn't last forever)

- does it actually exist? is this currently a closed-source codebase, or is "we're building" future tense?

> Is that related to some legal/privacy issue?

Possibly in some medical or social science fields, I don't know. I know there is not such an issue in chemistry and materials science. There also may be some complications for collaborations with industry, but that's kinda a different situation. For people whose career development is not strongly tied to reproducibility of their work (a.k.a. everybody) it's just another step in the overly complex process of publishing in for-profit journals. Funding agencies generally aren't going to punish people for using this excuse and the watchdogs/groups concerned with reproducibility have no teeth.

Not an excuse, but journals don't make it easy to share files, as hard as that is to believe. Some will only take PDFs for supplemental information and many have garbage UIs, stupidly small file size limits, etc. Just uploading to a repo (or tagged release) on GitHub is common these days because there is much less friction.

> I wonder if there is any requirement for researchers to at least publish their data set for statistical analysis and further research.

Not generally, though the tide is slowly turning in the right direction. Unfortunately many laws/policies pushing for openness and transparency in research are sidestepped with the classic "data available upon request," a.k.a. "I promise I'll share the Excel files if you email me" (they will not).

I'm not claiming there are no overly stubborn PIs. But we should not be lazy and paint so a brush that we view the field as a monolith.

I think there is truth to the general principle you refer to, but I don't think it accurately describes what I saw skimming experts' comments in the linked thread. I'm an outsider to medical research but have experience in other parts of STEM research at universities. Here I saw a plenty of nuance, documentation of historical skepticism, concern over broad perception, and plenty disagreement over technical points. Far from a unified kool-aid drinker sort of situation. And I think there has been plenty of changes of opinions in the Alzheimer's field in recent years given the number of failed drugs - which goes against the idea that these scientists are following their career over the evidence.

Hopefully so, but there will probably always be a back-and-forth between frauds and journalists in the same way security is always a competition. At least the easy frauds are more likely to be caught today.

Harboring skepticism of the work people did with a seemingly fraudulent researcher is a good idea. Dismissing everybody in a field whether or not their work is fraudulent is disrespectful approach (not to mention useless).

I'm not asking for people to turn off skepticism or blindly trust researchers. It's not disrespectful to be skeptical.

What is disrespectful is not bothering to read what people have to say before dismissing them as liars who are too vested in "the current research direction" and/or money for their perspectives to matter. It only takes reading a few comments to see that's not happening - for starters, people were skeptical of this group's work for a while now.

> basically all images in scientific publications are not scrutinized for photoshop

This was (mostly) true back then, but it is definitely not true today. People tempted to commit fraud now have to be worried about people like Elisabeth Bik exposing them and ruining their careers. In my experience, the type of people known lie in papers overlap strongly with those that are career-minded/money-driven. So having a few journalists with the skills to detect fraud is an obvious win. Some of the frauds will just get better at, that's just how it does, but it's not like there are no imaging experts in the field.

It's pretty disrespectful to signal (without evidence or elaboration) that researchers are not credible (or worse, broadly lying) in order to keep their research grants flowing. A hypothesis that turns out to be wrong is something both industry and government are obviously not going to invest further in. The people working in the field have skills to transfer to other departments and projects; these aren't the sort of scientists and engineers that are out of ideas or work to do.

It's also plenty obvious that there is no single, monolithic "current research direction" or even that this researcher's work was of fundamental impact when it was published - not to mention the number of people that were highly skeptical from the beginning.

> ... corporate american enabling this will not be seen positively.

I don't think this is true in red states or for the minority views that determine what is and is not illegal in this country, or at least it won't play out like you wish. Certainly I don't see this playing out along this optimistic path. Politicians in Texas and Mississippi will frame it as companies helping dutiful law enforcement investigate evil crimes and ergo it's fine that Google was tracking everybody all along. Single-issue voters would be happy to see abortion providers or seekers behind bars and couldn't care less if the fourth amendment was violated along the way.

Just look at how conservatives (ok, well, a good chunk of all mainstream politicians) and voters viewed the FBI vs. Apple conflict from a few years ago - it was "we have to make sure the cops can get the bad guys so of course the cops should have access to whatever information they need" not "I have a right to privacy - time to stop using these services until they respect it"

> There's no actual storage density information, is there? Nothing in KJ/m^3 units.

Even if there was (to be honest, didn't read the journal article) this is something that can easily be hacked. Energy storage research papers regularly hack energy density numbers by reporting the kJ/cc values of a tiny (like order 1 g) fleck of nanoparticle dust, which totally misrepresents the physics that matters are scale (i.e. in an EV).

Scaling up stuff is hard, including when you're moving from micron scale to cm scale.

It's just diffusing the problem from urban centers to mid-sized cities elsewhere in the united states. Say for simplicity that the core result of the problem is that people not making astronomical tech/finance/etc. salaries can't live in NYC because people making those salaries are scooping up supply and driving up rent ($5,000/month and higher). Then say the solution is to let some amount of people move to smaller cities and work remotely, enough that NYC prices somehow magically drop 20%. Well, where are those jobs going to go / how much of a salary penalty are people going to take to move from NYC to STL, Austin, Nashville, Columbus, Denver, Ann Arbor, and Raleigh? Because they're not going to suddenly be making the same salaries that engineers already in those cities make, they're going to want more. And they're going to scoop up housing supply and put pressure on those housing markets. The same forces keeping NYC rent at $5,000 have caused places like Nashville to become unlivable for the lifers; your grocery store workers and baristas can't get by with $3,000/month rent there either.

It would be lovely if the solution for skyrocketing rents in the big 2-3 urban centers didn't simply shift the problem to every other city, but there's no indication at all that will happen.

> everyone wanted to play like a Karpov or a Kasparov.

I wonder if there's a modern analog to this with how super GM styles have mostly converged. Trying to play like Magnus is as silly as trying to play like an engine. Even the most aggressive players (Nepo, Shak?, Rapport) aren't so wildly different in style.

Maybe the 2010-2020s version of this is bandwagoning onto popular theory, like all the Najdorf lines I know I'll never understand.

But for the vast majority of players there will be a difference in computer lines and human evaluation; even if you're playing a strong game there is still a world of difference between 2500 lichess and ~3300 FIDE stockfish. This is more true for the median ~1500 lichess player. Even if you put two GMs up against each other and give one a computer, some portion of games would include an obvious computer line that a ~2800 FIDE human wouldn't evaluate the same way as an engine.

Thanks for the context - I haven't spent a ton of time looking around but it does seem like there are several moving parts (and evidently some important stuff isn't in the same github org)

Thanks! I didn't know about this page.