caddemon

At least RTO and WFH companies can co-exist though, and there are hybrid models as well. With school parents are stuck with what their specific district wants to do, and the number of hours and days spent in school is extremely rigid. For the academic year that started in fall 2020, it's not inherently impractical that the two options could have co-existed for both teachers and parents. But obviously it was unrealistic given the state of the school system already pre-COVID.

I think the broader issue with this sort of choice in a pandemic is it does have downstream effects. People going to school everyday could cause an uptick in community cases that then makes necessary trips like the grocery store more dangerous for those who do want to isolate as much as possible. Especially relevant in a city where many people are living together in apartment complexes.

So I agree there's no clear answer here, and I think pre-vaccine I would've kept my kids home regardless. But more generally I am very dissatisfied with how homogeneous and inflexible the school system is, and that is something we should push for change in. Funnily enough, my district has done a full 180 since COVID and this past year stopped accepting doctor's notes to excuse absences! They are quite literally encouraging people to send their kids in even while sick. A friend's son has mono right now and the HS is threatening to hold him back over absences (grades are fine still because ofc that matters less than attendance record.

Their best reported results are a hybrid effort though, here is one of the authors of the paper describing how they used programs generated by the LLM to extract their own insights that then refined future iterations of their workflow: https://x.com/matejbalog/status/1735331210140819938?s=20

It can work by itself too but it is unclear at a glance how well since the main focus of the paper is the new mathematical benchmarks they achieved, i.e. their best results. Will have to read the paper more closely to say anything with high confidence, but based on their summary I'd guess the human in the loop part was pretty important here.

It seems this is essentially an evolutionary algorithm with an LLM generating the pool of new variations at each step. Definitely a very cool idea, but hard to evaluate the results without knowing more about that field of mathematics. Obviously the problem they chose fit well into the FunSearch framework, but I'm curious if this is one of the more popular open problems in that space or if they picked something that was more niche?

Namely, what sort of computational resources had been dedicated to the problem before? Because DeepMind suddenly throwing their weight at a problem that was previously the focus of a handful of random math grad students would make it hard to benchmark the ML advance that was made here -- like would it be possible to find a similar solution with a ton of compute and more traditional genetic algorithms?

I wouldn't be surprised if the answer were no. Protein folding competition was a pretty big space in biology before getting absolutely destroyed by Alpha Fold. But I also wouldn't be entirely surprised if the answer were yes. The amount of hype around LLMs right now is crazy, probably half the news I see about them turns out to be very exaggerated upon further evaluation.

ETA: the headline here is definitely exaggerated, because there was also human in the loop to refine what the LLM was generating. At a glance the technical article doesn't benchmark enough against alternatives to the LLM component in their workflow IMO. But it is entirely possible they tackled well established enough open problems, such that prior work already handled those control cases decently.

I'd love to know what someone in this space of mathematics thinks about the paper! Would it have generated much buzz if they got these same results like 3 years ago? Would it be accepted to Nature if they found they could accomplish something similar using their framework even if it was an RNN in the loop?

I agree a significant amount of work (and often insight too) is needed to translate an architecture idea into something that works in practice, and there are certainly plenty of ideas that are obvious in the abstract. But I also think it's important to avoid dismissing work only on the basis that it doesn't involve "real life datasets".

Deep learning is a relatively unexplored field and there are many open mathematical and scientific questions to ask that involve only model equations or contrived datasets. Novel theoretical results are not just about some architecture idea but about proving facts that can be useful for understanding how the model class would perform in different scenarios. Which in turn can help shape the search space for applied work.

Additionally, I don't think credit assignment should be so discrete. 100% agree that vomiting out vague ideas shouldn't grant claims to credit, but academic science much too often gives only a single author the "real" credit.

Incidentally, in other fields the person who actually makes it work very well may not be the person that receives this credit. Like biology can involve a lot of hard manual work (that isn't really intellectual) in order to realize a project plan. It varies how much of the credit those people receive, and I'm not even sure how much they should receive. This topic is extremely nuanced.

It's hard not to think of the beef when you think of him now though, even if you already knew some of his work. To be clear I don't think that should take away from what he did accomplish, but it's like a Pavlovian association at this point to also have his behavior come to mind.

Samsung frame TV, they're about 33% off this week from the typical price.

I know it has mixed reviews with the tech crowd but everyone I know that's gotten one for primarily aesthetic reasons loves it. I have an open floor plan apartment with not a ton of wall space, so I'm optimistic it will be a good way to have a TV when needed without making the TV such a focal point of the room. I'm also looking forward to getting exposed to new artwork in the shuffle mode.

The problem is moreso trying to maximize profit after claiming to be a nonprofit. Profit can be a good driving force but it is not perfect. We have nonprofits for a reason, and it is shameful to take advantage of this if you are not functionally a nonprofit. There would be nothing wrong with OpenAI trying to maximize profits if they were a typical company.

At longer timescales it is important to be able to recognize mistakes and reverse course, but this happened so fast I'm not sure that's the right characterization. There's no way they could already decide that firing Sam was a mistake based on the outcomes they claim to prioritize. Reversing course this quickly actually seems to me more like a reaction based directly on people's negative opinions, though it may be a specific pressure from Microsoft as well.

Wall Street non-competes are only enforced if they pay you your base salary during the period. They can also opt to not do that and waive the non-compete, in which case you can work anywhere. I think it's ridiculous that Subway has non-competes, but with regards to the article I doubt anyone is forced to choose between working for a trading firm and starving. There also are some firms that do not do non-competes. So it's closer to voluntary than involuntary I'd say, though very few decisions are truly 100% personal choice.

They make sense where there are specific internal secrets, it's not just about IC skillset necessarily. But it's true they're definitely overused.

Nicotine specifically is also more likely to be used by people with psychotic disorders than the general population, as a form of self-medicating. Many Schizophrenic people report smoking cigarettes to help their symptoms in the short term, though I'm not sure what research says about long term mental health effects of this practice.

Psychosis is a potential symptom of lupus, albeit uncommon. Some people with lupus have been misdiagnosed with Schizophrenia, and more generally there are signs of immune dysfunction in many cases of Schizophrenia. I'm not sure about this particular scenario, but psychosis is also a potential symptom of meningitis, and brain inflammation can definitely cause weird behaviors (e.g. the account in the book Brain on Fire). It's not "severe damage" in the traditional sense, but depending on illness and treatment timeline there can be irreversible life-altering psychiatric consequences from brain inflammation.

If someone missing a left arm managed to get really good at baseball, they could still play. A one armed dude lighting it up on the field would be drafted, because objective results.

Of course this is highly unlikely because it handicaps skills related directly to baseball, but it's not completely impossible for a pitcher or DH. The Minnesota example is not comparable, because the ability to do good research is not inherently impacted by staying at one institution, and the grant evaluation is directly penalizing the staying part rather than indirectly.

I can't speak to whether the university or this guy's research were actually any good, but it hurts science to have such rigid proxy judgements.

Hot take, but I don't think it should be such a taboo if individual institutions have "insular culture". As long as there is heterogeneity amongst relevant institutions, the synthesized literature can still be very productive even if e.g. MIT has a bias towards a particular way of thinking.

Conversely, I think it's problematic for scientific creativity how homogeneous behavior across institutions has become. I don't believe it is a result of this cross pollination requirement, but rather causality probably goes the other way. Many old professors at MIT were true lifers, from undergrad to PhD to faculty. Postdocs were often skipped entirely, or looked more like a 1 year visiting researcher gig.

Regardless I don't feel there should be a viscerally negative reaction to individual institutions having clear characteristics. It's true that objective evaluation is hard, but it's hard anyway honestly.

Who is the FTX representative at this point?

It was largely other biology professors, not admins. It's a bad look no doubt, and there are other issues with modern university admins - but this is a direct consequence of those in the field, not typical bureaucracy crap.

You don't value and promote a single or even a handful of archetypes. You value and promote a healthy diversity of approaches. Not just diversity amongst individuals but also between institutions. If MIT, Harvard, Stanford all had meaningfully different research philosophies, it wouldn't be quite so obvious how to game the system.

If the NIH gave some grants like they already do, some grants that would target someone like Karikó, some grants that would target reproducibility experts, etc. (in addition to some that may be a combination of such criteria) it would be a much more complicated political game to try to be an ideal imposter. You'd still have some but I think there would be many more filtered out, and more true positives kept.

I largely agree with your complaint but I don't think the causation is quite that simple. "Bean counters" didn't show up to a highly competitive field and decide to redefine the competition. It used to be relatively easy to get a faculty position, and there used to be a relatively small number of people/papers to keep up with. As academic science grew, both in the complexity of open questions and the size of the aspiring faculty population, there was going to necessarily be some changes to academic output and its evaluation.

Unfortunately they've hyper-formalized in an utterly homogenous way. It's such a disaster that perhaps a nepotism free for all wouldn't be quite so bad. But I do think there's a deeper discussion here about how academic research can scale in a better way, both to accomplish larger/more expensive projects and to handle increased application pool sizes.

Biomedical sciences are pretty bad in general. There are truly tough problems but I also think that industry is way too intertwined with academia right now, which causes cascading effects and perhaps parts of the cultural rot to spread. Obviously the economic incentives/legal concerns you've described are a different story, but the initial mistake you've mentioned that enabled the whole scenario I'm guessing is part of a larger pattern with the quality of pharma research.

At least profits have some rough relationship with reality in most industries. In academia it is the worst of both worlds. You need to be a start up hype man type (or something adjacent anyway) to get the grants, but you only need to deliver a paper at the end - not an actually functional product or organization. So we do not see the same upside that a market driven approach might have for eventually finding a stable state, only the downside of being ruthlessly metric driven.

Yeah any increase in funding needs to go to meaningfully different grant processes. Giving more money to the existing funding structures will not fix a damn thing. Because the present system is just woefully unscaleable. More money to current NIH -> more PhD students -> tenure track competition remains fucking ridiculous.

It's funny that you use "academia is like the mafia" as an example that we are supposed to find outrageous. The labor structure of academia has a number of similarities with how gangs operate: https://blogs.lse.ac.uk/impactofsocialsciences/2013/12/11/ho...

I don't think it is quite that intentional, but just as human economic systems regularly create bubbles there are blind spots in the academic science collective. People desperately want to get in on LLMs right now.

This guy is a science journalist, the authors of the source research paper are more relevant if you're trying to infer anything about the material from the authors: https://www.nature.com/articles/s41586-023-06668-3

FWIW I think even the original paper is more of a master class in riding an academic bubble than anything of important substance. I'm just not sure what looking at other articles by this journalist is supposed to convey.

I've seen restaurants with great food close because of terrible business sense, and I've also seen a few closings recently of long-standing places where the owners wanted to retire and couldn't find anyone they'd trust that would take over. COVID amplified these effects for a time too.

I'm sure a higher fraction of closing restaurants are mid (or worse), but by raw numbers there are many places legitimately worth memorializing. Associated personal memories are a thing of course, but there are multiple closed spots that I feel were inherently very good and have not properly been replaced in my area.

The primary purpose of an internship at most major companies is hiring - you can screen candidates much better before extending a full time offer and the candidates can also get a better read on how much they'd enjoy working there. Even the second order effect of drumming up hype for others to want to work there is probably more important to many internship programs than whatever the interns might tangibly produce.

At least I can say for sure that people at Google aren't shy about how it takes a few months to get up to speed with internal tooling and not much is actually expected from a summer intern. But they built themselves into a brand on campuses, which became so mainstream there was a stupid movie made about how fun it is supposed to be to intern at Google.

And as far as finance goes, trading interns aren't even legally allowed to trade, so I wouldn't be surprised if the culture surrounding internship at Jane Street just leans more heavily into this for SWEs as well.

Jane Street isn't an especially big player in HFT, though even with a longer timescale your overall point still stands.

Sugar can be problematic, but those rat studies are quite contrived. They follow an intermittent fasting schedule where rats don't have access to any food for 12 straight hours of each day and then they are presented with the option for unlimited sugar drink for a period, with this cycle repeated for weeks to get the reported results. If the rats have 24/7 access to the sugar drink they don't develop all these addict-like behaviors. Such research can be mechanistically interesting for biologists but I don't think it says much about how we ought to assess human behavior in practice.

Here is a review article discussing in more detail relevant literature on evidence for/against sugar addiction: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5174153/ Certainly questions remain, but I think it's pretty clear sugar and cocaine are not actually in the same stratosphere here.

I don't see how this is necessary collateral damage for the prevention of drunk driving. I think most folks in this thread would agree that current implementations leave an awful lot to be desired, regardless of what they believe the best approach to be theoretically.

That's a problem with specifics though, not with the general existence of DUI laws. Weed will be hard to enforce but BAC is more cut and dry, I don't see a problem with having a BAC cutoff for driving.