This confusion is kind of my point. The lipostat hypothesis is still a bit controversial after 80+ years. If a "set point" for weight truly exists somewhere in the system, it's still not clear what raises this set point, and why lowering it again seems extremely difficult.
There's plenty of studies that indicate that once the set point has been raised, it can't be easily lowered again. This is, funnily enough, contradictory to your 1965 study (unless the 400 lbs -> 200 lbs guy was short, and 200 lbs was still obese). Or maybe only more modern food additives raise the set point permanently? I don't think we know, and almost nobody (relative to how important those questions are) seems to actually test things on large groups of people.
pbmonster
2d ago
Yeah, the entire field of dietary science always reminds me of the factoid that Aristotle thought men had more teeth than women - and then the entire "scientific" establishment believed that for several centuries without ever just... checking.
Hopefully, it'll turn out that we can just measure this. Take a couple of hundred people on a retreat and count what you feed them.
At least with gemini, it should just use patents.google.com
Also, that would be many, many millions of tokens.
No, not all of them, not all the time.
Each local fire department has a big, public summer party here, for fundraising and to say thank you to the volunteers. Those parties are never at the same weekend, and all firefighters visit each other's parties.
Also, after big exercises, sometimes a wild keg appears...
Still, everybody jokes that you could burn all the fire departments down to the ground as long as there's a summer party somewhere in a 50 mile radius. It's probably true.
pbmonster
3d ago
Yes, because this process is the source of 90% of the problems self checkout machines create (the other 10% are alcohol purchases), and an employee has to come and fix those problems.
In my casual survey of self checkout machines across cities and countries, I've come to the conclusion that only supermarkets in extremely low-trust cities use the scales. For everybody else, it's just not worth it, since you can literally increase the machines/employee ratio by 10 if you don't need to use the scale functionality.
My local supermarket doesn't sell alcohol ind is in a high trust region. Tens of self checkout machines are running without employee oversight, if there's a problem the store manager comes to deal with it.
who have asked recent LLMs questions in your area of expertise, how accurate are the responses? What is your field and what models are you using?
For "frontier tasks" in physics/electrical engineering, it's bad. It just doesn't work, even as a search engine.
My most recent request was "Find me patents about the application of concept X at high magnetic field". Should be easy, patents are public by definition. Searching google patents has worked for decades. There's proprietary patent databases with curated keywords. Perfect training data, easy to search.
But all the current reasoning models with web search just give me results at extremely low magnetic field (which is the standard application, there's many patents like that. That's the reason I'm asking an LLM, I don't want to sift through those by hand). So I specify: "Keep in mind that milli tesla and micro tesla are low magnetic fields. Please exclude patents that use those units from your search". I'm already disillusioned, I shouldn't need to do this. A nerdy highschooler would know better. But it doesn't work. It just ignores the request, appologizes, and keeps spitting out patents with those units in the abstract.
Also, I still need to paste every single patent it spits out into my patent database tool, because literally 50% of the results are hallucinated. The patent number is a completely different patent, and the title it prints doesn't exist.
One core weakness of the current models seems to be things that don't exist (as might be the case for the patent I'm looking for). Another example for that is requests like. "I'm using Oscilloscope Y, and I want to change the color of one of the traces on the display. How do I do that?" For my oscilloscope, the answer is "you can't, those traces have their colors hard-coded, fuck color blind people." But the LLM will automatically read the correct manual (good!), link it, and then proceed to hallucinate itself into psychosis. Just flat out invents entire menus and setting dialogs every time I press it harder.
pbmonster
3d ago
My volunteer fire department is currently not accepting more people to their wait list!
There's many small reasons for that, among them: people think it's cool, it's fun training/working with heavy machinery, they party extremely hard, among the best local networks/old boys clubs available, looks good on CV.
There's also more material incentives, like they pay for your trucking licence (so you can operate the heavy fire trucks), power boat licence and things like first-responder medical licences - but in the end, nobody quits after getting those perks.
pbmonster
7d ago
I was in undergrad 15 years ago and teach those courses today. The standard math and physics intro courses have not gotten easier, drop-out rates are about the same. STEM education is pretty conservative, especially so outside the CS departments.
All but the best reasoning models hallucinate so much on standard problem sets that they're not very useful to students (who overwhelmingly only use free models). Also, those problem sets mirror closely what will be on the exam, where using an LLM is only possible on a bathroom break. The students who don't drop out by year two usually have learned that they need to do the problem sets themselves for their own good.
Students will cheat on lab reports, but they always did that. Today, it's ChatGPT, 15 years ago it was a Dropbox with old lab reports to copy from. Proficient cheaters will only "get help" on abstract, conclusion and the theory section - which is hard to proof, so we look the other way. Bad cheaters will copy the data analysis section, or even the experimental data. This is extremely easy to prove, and those get nuked in public.
would be allowed to house students in conditions no better than those junior enlisted in the army experience(food would have to be absolutely identical down to coming in boxes labeled 'not suitable for prison use')
Why? Is the cruelty actually the point this time? Because I see absolutely no gain here.
Palatable meals (produced at scale at a stable location) are not expensive, especially not when compared to education. Barracks bunk beds might benefit unit cohesion, instill obedience/submission and be easier to supervise/police, but that's far less necessary for the next generation of academics, and the trade-off in privacy and independence is absolutely not worth the price difference.
I'd go the other way. Kill mandatory "all-you-can-eat" meal plans (also makes the "freshman twenty" less of a thing) and mandatory on-campus dorm life. Have private businesses operate the dorms and cafeterias (plural, they need to compete) and let students live off-campus the moment they want.
And if you want to safe money, start cutting at the admin building.
pbmonster
10d ago
Any purported educational benefits of video games are bogus. All research on transfer on learning is that it doesn't exist.
As an anecdotal counter point: I played through the "Age of Empires 2" campaign at around 13 years old. Not only did it teach me more about history than the entire middle school curriculum, it also awakened my love for history. The combination of both lead to perfect grades in all my history classes till I graduated high school.
AoE2 is an exceptionally well done game, and arguably stands undefeated in almost every metric used to judge a real time strategy game, even today - so this is by no means an automatic result of video gaming. But it is one possible result.
Kerbal Space Program (which reignited my passion for rocketry long after studying mechanics at university) and Dwarf Fortress (which is the source of 99% of my knowledge about minerals, ores and metal smelting) are two other examples I would have no trouble letting my kids play, once they are around... 12?
I mean, sure. And the Marines are part of the DoD. Does that facilitate ideological capture, especially one that differs between administrations?
My personal experience in both cases says no, not really. The government can't significantly change the ideological makeup of either the national labs or the Marines, and both are ideologically not significantly different than the median of the population.
pbmonster
15d ago
Are the current national labs "part of the government" in any meaningful way?
pbmonster
15d ago
You can allow an almost arbitrary amount of academic freedom in biochemistry and expect that there will be at least some valuable and true information that is eventually produced
Agreed. I proposed a solution for that downthread: move all university STEM research to national labs, and have them train grad students. Undergrads stay with their "teaching professors" at the current universities.
pbmonster
15d ago
I would argue that you are treating academia as a single thing when it is clearly made from a lot of different parts. STEM ideally has both feet planted in reality, and is not very subject to ideological capture.
Yes, absolutely. And the only reason I'm doing that is because the current culture warriors rampaging through universities and science funding are doing the same.
I have a pretty radical solution for that: move all university STEM research (including all the grad students) to national labs. All undergrads stay at the current universities, where only "teaching professors" remain. Even in STEM, most undergrad classes don't benefit greatly from having an active researcher teaching them - but graduate level classes do.
I would also contest a bit that research (e.g. in fundamental physics) is genius constrained. The biggest discoveries in physics in the last two decades were the Higgs boson and gravitational waves. Both LIGO and LHC were massively collaborative efforts. The bulk of the work was done by PhD students who were smart, but not super-geniuses.
I kind of disagree. There's a lot of small stuff happening behind the scenes at universities and then silently creeping into products all over the world. There's two relatively recent prices for lasers, those came from "classic-size" groups. Advances here (independent from those prizes) also still frequently make it from universities into (e.g. telecom) products. In biochem, CRISPR/CAS9 was an incredibly small team. In material science, I expect small university groups making big contributions to high-entropy metal alloys and to further improvement of semiconductors.
pbmonster
15d ago
So then the question is if we really want to mess with the system that globally is the best at driving fundamental research in the hard sciences - because of a pretty small number of people spreading garbage on TV or through their private blog. Which, arguably, those people would be doing anyways (there's plenty of idiots with normal jobs successfully spreading garbage through the same channels). So really, we would only be changing their title from "Professor" to "Doctor".
pbmonster
15d ago
"Academic freedom" sounds good and all, but what happens when it's implemented in real-world universities?
Nobel prices and fundamental research that changes the world a few decades later.
As the "classical liberals" freely admit, the results are often not stellar. So what's their solution? Doesn't seem they have one.
Research (and upstream activities of future research, like teaching and mentoring) are strong-link problems. Your end results only really depend on the very best that do it, the "not stellar" don't effect overall outcomes much. The problem is - as in most strong-link problems - that you don't know who the very best will be in advance. So having a lot of "not stellar" people have academic freedom (and have little to show for it 30 years later) is just the price of doing business.
So, what's the solution? Don't worry to much about it. Hire already successful mentees from the previous generation of strong links, punish outright fraud (Alzheimer research scandal, ect.) and leave them alone. And then don't trust any of them to much - if you're making policy decisions on their advice, you need a large meta study anyway. Usually, a strong link contrarian will appear.
What you shouldn't do is have a new crop of elected representative fight their way deep into the system every 4 years and topple everything. That actually affects outcomes.
pbmonster
16d ago
one of my roommates, who never practiced polyamory per see, but always had a "rotation" of girls going (maybe this is the cool chad version of poly, idk)
There's a joke along the lines of "Ah, so you sleep with a bunch of different girls, who each also might or might not sleep with a bunch of different guys - but you really like them, and one of them might be your housemate? Back in my days, we used to call that 'being single in college'."
Depends on how they do it. I've read so many tariff stories from industry insiders, and they all go "We used to import this for $80, shipping, handling and fees brings it to $100, and to account for our overhead we sell it for $300 plus tax. But now, we import it for $80, tariffs brings that to $110, shipping and handling brings that to $130, and to maintain our margins, we have to sell it for $390 plus tax."
They ignore that because of tariffs, they now make $260 of profit instead of $200, and half-heatedly claim that this is because of "margin" and because of "overhead". This insistence to preserve the gross margin percentage (instead of preserving the dollar amount margin per product) makes absolutely no sense to me, but they literally all do it. Margin is a percentage, end of story.
So, you probably can't learn much from looking at what Amazon claims are tariff costs.
pbmonster
28d ago
- Day 1: Approach day. Drive into the mountains, ascend to camp/the mountain hut by evening.
- Day 2: Summit day. Spend the day hiking and preferably climbing. Chill at camp/hut to soak in the views, or (ideally) move to a new location.
- Day 3: Summit secondary peaks, take a detour back to the car.
pbmonster
28d ago
If I were queen, my hard rule would be no sexuality in school for any kid under 12. At 13 or so, obviously you need to explain sex and how babies are made
Bit late, isn't it? By that time, most of the girls will be menstruating and a few of them might be sexually active already.
And sure, you can argue that the parents should have prepared them for all of that, but realistically, some haven't. I'd prefer it if the schools make sure they have some idea what's going on, at the very least the biological reality of their situation.
pbmonster
1mo ago
While we have advanced pretty far on the "reacts to sensory inputs" front lately, the "autonomous reproduction" is still sorely lacking.
pbmonster
1mo ago
US phone manufacturing capabilities have little to do with "AI killbots"
How do you imagine the first generations of killbots to look like? I imagine a state of the art mobile processor, a lithium battery, a camera module, lidar, GPS, compass, gyroscope, accelerometer, modem, beam forming GHz antenna package. Motors, props, shaped charge explosive.
Remove the last three and add a screen, what do you have?
pbmonster
1mo ago
How is such an obvious target not immediately blown to pieces via airpower?
The Houthis are operating in a much smaller area than the Somali pirates did before them. The oceans are big, the horizon is small. Finding a pirate ship via aircraft hours after a call for help (which is how long it takes until airpower arrives if your navy isn't dominating the seas and your allies are far and few between) is difficult, especially if you try avoiding blowing up random fishing boats.
And both Somalis and Houthis are extremely low tech. Doesn't take much to give a pirate ship thousands of miles of range.
pbmonster
1mo ago
For a less insane aproach just have soldiers with RPGs and Snipers with shoot on sight orders stationed on every ship.
I'm not convinced. The escalation is to easy for the pirates. Current Somali militias could do it if they cared. Slightly bigger ship, 10 tons of hand mixed concrete to form a ghetto pillbox, soviet artillery piece. Done.
They can't board as easily anymore, but they can threaten to blow up the cargo ship's bridge.
Escalating for the cargo ships is not as easy. They already have to keep most of their hardware and mercenaries in international waters on their floating armory ships, they really don't want to install naval guns every time the mercenaries board a cargo vessel.
Yeah, there's several theories like that. It's also possible he counted including wisdom teeth - because women are statistically more likely to never have (some of) their wisdom teeth break through.
The point is, he (and his followers across the centuries) evidently never just counted the teeth (or the tooth gaps) of people.
More options
Context Copy link