rokmonster
Lives under a rok.
No bio...
User ID: 1473
rokmonster
Lives under a rok.
1yr ago
On the other hand, how many of the current crop of AI researchers were directly motivated by Eliezer, and how many followed independent paths? As computational power and GPUs improved (be it for gaming, for servers, or for bitcoin), gradient descent becoming practical was an inevitability. Once gradient descent became practical, researchers start pivoting to it, and the only barrier (that we know of now) is the availability of datasets and hardware. The snowball was doomed to start rolling with Hinton's publication of back-propagation in 1986.
rokmonster
Lives under a rok.
1yr ago
Martian colonies have an asteroid dropped on them, and whatever pathetic escape craft we make in the next 20 years get swatted before they reach the orbit of Saturn.
In 20 years the AGI apocalypse will not be nearly as romantic as that. It is much more likely to look like a random bank/hospital sending you a collections notice for a home loan/medical treatment you definitely didn't agree to, bringing you to court over it, and putting you up against the equivalent of a $100M legal team. The AI-controlled Conglomerate wins in court and you spend the rest of your life subsistence farming as a side gig while all your official income is redireted to the AI Conglomerate.
For extra fun, if you are married, social media and increasing economic struggle poison your relationship with your spouse and both of you apply for the services of AI Legal. The hotshot AI Legal representatives fight acrimoniously, revealing every dark secret of both you and your spouse, and successfully breaking apart your marriage in divorce settlement. Honestly, you don't remember why you ever loved your ex-spouse, or why your children ever loved you, and you totally understand your real-world friends distancing themselves from the fiasco. Besides, you don't have time for that anymore. Half your salary is interest on the payment plan for AI Legal.
As a smart and independently wealthy researcher, you look into training your own competing, perhaps open-source AI model to fight back against the Machine, but AI Conglomerate has monopolized access to compute at every level of the supply chain, from high-purity silicon to cloud computing services. In despair, you turn to old web and your old haunt The Motte, where you find solace in culture war interspersed with the occasional similar story of despair. Little do you know that every single post is authored by AI Conglomerate to manipulate your emotions from despair into a more productive anger. Two months later you will sign up to work for a fully-owned subsidiary of AI Conglomerate and continue working to pay off your debts, all while maximizing "shareholder" output.
rokmonster
Lives under a rok.
1yr ago
No. In the specific incident that comes to mind we had an new student try to take credit for the results of a 3rd year PhD candidate after fixing/running the nearly-successful experiment while the older student was at a conference. Thankfully the PI saw through it. I'm sure it goes the other way too, though.
You're onto something here. Where I did my degree, the following was pretty much understood by all the students after their first few years:
-
The purpose of research grants is to get research done for the funder more cheaply than is possible in other sectors of the economy.
-
The purpose of Professors is to get funding and write grant proposals. This means anticipating what research will be trendy and making a lot of friends among the people who staff grant proposal review committees.
-
The purpose of the older graduate students is to do the research, write papers, and write grant reports, while mentoring the younger students.
-
The purpose of the younger students is to study and learn, while assisting the older students on writing grant reports and doing experiments. Oh, and to teach undergraduate classes.
-
Graduate students needing additional mentorship must actively seek it.
(We didn't have post-docs or research staff, but they basically allow scaling of the grant-writing work and supervisory work of professors.)
This was a decent system for graduate students who were self-driven and capable. It had many different failure modes, however: It rewarded professors for just enough surface level knowledge to come up with cool sounding projects that were in reality infeasible. It was hell for students who were given the new projects, because they had no mentors in their specialty, and had no idea that things were infeasible. Older students could be abusive or predatory, and unscrupulous younger students could wait until an older student had worked out nearly all the kinks in an experiment and then swoop in to take credit for the results. Professors had a bias for sudents running simple but creative experiments over meticulous work that was actually necessary long term for good engineering.
Like you, I had a professor with only a surface-level knowledge of my research domain. I was often given bad advice and advice that wasted time. (The students figured out that our PI didn't read papers, but read abstracts and skimmed figures, which made for some funny misinterpretations of the literature.) The PI's feedback on student work was vague and hard to understand. However, when it came to overcoming stuck research projects my advisor was a genius. The experiment-breaking result became the new goal of the experiment, easily publishable. My advisor also eventually communicated an understanding of how to write a good research paper, after which all those vague comments suddenly made perfect sense. So the relationship turned out quite valuable.
The worst part was the social environment. In order to get the PhD students had to become first author on multiple papers, but the PI would assign multiple people to each research project, bringing in more people the longer it took. I'm not sure there was sabotage (I'm dumb enough to fuck things up myself, thank you), but there was definitely spying and theft of results between students. The students needed favor with the professor to buy equipment: seeking the favor of the professor resulted in schemes much like those of medieval courts. Reading The 48 Laws of Power during my PhD, the content of the book depicted the social environment of the lab quite accurately.
Overall, it was a fun time, but I would probably recommend a gap year after a masters degree instead of a PhD. Travel the world, get more life experience, suffer less stress, have more fun, and in the end you didn't spend four years becoming the world's foremost expert in some experiment that is only performed in one lab.
rokmonster
Lives under a rok.
1yr ago
My understanding is that the Starlink has three potentially profitable strategies, all of which depend on inter-satellite links to really be breakthroughs:
-
HFT without fiber latency. This is where clients will be able to pay eight to nine figure subscriptions per year, with major routes forming a high speed web between New York, London, Brussels, Singapore, and Tokyo.
-
Military/aviation. It's hard to put a dollar value on military contracts, but this is probably seven to eight figures per year total, since military already has their own communications web.
-
Worldwide consumer access with less than worldwide infrastructure. Note that once access (downlinks) are installed for the above contracts, the marginal cost to expand civilian access to the globe is almost nil. The cost of satellite launch may be spilt between customers across the entire broadcast range: Africa, Europe, Asia, all the ocean shipping and cruise ships, etc.
Seems like the extent of contamination is scientifically a similar problem to something between fallout tracking and rocket engines (combustion chemistry). It would be really fun to build a team to model this but realistically noone will pay for it when the victims are a few thousand poor rurals in Ohio, except the companies seeking to minimize liability. Naturally the residents will be skeptical of the accuracy of any modeling done by chemical companies.
Wow. That is a crazy ruling. That's basically holding that society must provide some form of shelter to everyone, either directly or via land-grants at the location of their choice, and it must be situated within city limits. I thought declarations like that were usually constitutional amendments or acts of congress, not court decisions.
As a heartless pragmatist, I would like to point out that the local prison is shelter, and usually has plenty of capacity. There is also a ton of room for innovation in public shelters/public housing: public office space is not used at night which could double as shelters, public parking space could be requisitioned for the contruction of shipping container capsule hotels, and cheap homes could be bought up and partitioned.
rokmonster
Lives under a rok.
1yr ago
The story of how and why they did this is long, and despite being matters of public record, not widely known, but that explains the difference.
Could you provide links to this story, or at least provide a hint where to start looking? It seems worth knowing more about, if only so other countries can avoid the American issues.
5 minutes is true for my experience of urban Europe and Asia. In both one can drive further to a big box store and do weekly shopping, but walkable grocery stores are near major walking commute routes and sell quantities of food that the single person can carry back to their home.
I usually buy fresh groceries daily 5 min from my house (but 10 seconds off my route) on my commute home and nonperishables 1.5 hours away by bus once a month.
rokmonster
Lives under a rok.
1yr ago
This is the opposite of how grocery stores actually work in true urban areas: in the walkable urban area, the grocery store (and its selection) scales to fit the space available, so that it can afford to rent out a profitable location. My neighborhood grocery is less than than 30 ft wide and 90 ft deep on the ground floor of an old 3-floor building, but that building is halfway between a residential area and the subway and right next to a bus stop, so the foot traffic on the sidewalk is on the order of 10 people per minute. (The walk from subway to residential area is 10 minutes, tops.) They do their best to lure in customers by placing fresh fruit and sale items literally on the sidewalk. (I hear this would not work in America, because all the food would grow legs and walk away.)
Before a recent move, this grocery store was in an adjacent building with even less floor space (which building was torn down to install a 20-floor monstrosity). A tradeoff is made to between selection and bulk: the "snacks" aisle only has one, maybe two display items of each product, and the average American grocery run (with a grocery cart) would buy them out of their standard inventory. (They don't have grocery carts, the aisles are too narrow for them.) Fruits and vegetables are available fresh and in bulk, but you get what is seasonal or standard for local cooking. You pay a lot more than at a big box store like Costco, but that's the price of convenience and for having a store halfway between the subway and the residential neighborhood.
But I don't think the store would possible in America. The only parking is the loading area in the back, the display which lures people in would be subject to too much theft for the sole proprietor to make a profit, the aisles are too narrow for wheelchairs, and the entryway has a few stairs up from the sidewalk, which would fail any ADA requirements.
rokmonster
Lives under a rok.
1yr ago
How do Chinese children learn to read, if their every word is an ideogram? Do they all have abysmal literacy rates, like these "balanced literacy" children from the US?
First of all, there are only ~3000 commonly used Chinese characters, and they contain patterns which make them easier to memorize once you memorize a few hundred. At one character per day, you can learn all the characters in 12 years of schooling. Realistically, the characters are introduced much faster than one character per day and used much more frequently.
One might compare this to the number of phonetic exceptions in English: I remember 5-10 new words for the spelling test every week, from 1st grade to 9th grade.
Note however that if you are going to memorize English words without learning them phonetically, there are many more English words than Chinese characters. A few orders of magnitude more.
rokmonster
Lives under a rok.
1yr ago
Circumstantial evidence suggests that it was Spain:
https://www.cbsnews.com/news/poland-mystery-divers-gdansk-port-energy-oil-gas-infrastructure/
Kidding, but something tells me that Poland has looked into these divers and as a result has a pretty good idea who blew up the pipeline.
rokmonster
Lives under a rok.
1yr ago
Maybe affordable unit requirements mess with this? idk
"New residential development" in NIMBY cities (as opposed to rural America) usually means tearing down existing single-family homes on large lots and replacing them with high-density housing on small lots, which consists of multi-floor, multi-family apartments, at least 20% affordable or subsidized units, and no yards. While the developer makes a lot of money on these due to the high-density, the price paid per household is almost always lower than the area average, the area loses some of its greenery, and the average social class of the area falls.
I think getting rid of the affordable housing requirement would result in some developers focusing on large high-density, high-cost, high-quality condos near in-demand areas, but the affordable housing requirement puts a limit on unit sizes and quality overall, and makes NIMBY the equilibrium position of a neighborhood.
rokmonster
Lives under a rok.
1yr ago
Five years ago (pre-LLM) the Chinese were already been working on AI for automating court judgement on the theory that it would be more efficient and fair. Lawyers and law are one of the major areas in which next-generation LLMs have the potential to be very profitable.
It's amusing how his reasoning is similar to @pointsandcorsi's (although much more careful and conservative, especially with forecasts) and thus passes for academic content, yet pointsandcorsi's output merits a permaban on an obscure forum.
Huh? @pointsandcorsi was banned for a comment in a thread about James Webb, and it was a week-long ban, not a permaban (as evidenced by their most recent top-level post which started this thread):
https://www.themotte.org/post/240/culture-war-roundup-for-the-week/44716?context=8#context
Is there some other event that @pointsandcorsi discussed that I missed?
Oh, I see. They were banned for this post.
rokmonster
Lives under a rok.
1yr ago
I'm annoyed at the reporting requirements too, but the mirror image of money laundering is tax evasion, and governments are very motivated to prevent tax evasion by any means possible, up to and including totalitarian monitoring of all money flows.
With respect to the specific requirements to report foreign accounts: the reporting requirement is clearly stated in tax instructions and up to a few years ago the IRS was remarkably lax about requiring people (with less than $50,000 in their accounts) to report on time. The form for reporting foreign accounts even included checkboxes where one could state one's "reason for reporting late": "I forgot" and "I didn't know I had to" were valid options.
Granted, I'm still a bit confused by the reporting requirements and process for large wire transfers.
rokmonster
Lives under a rok.
1yr ago
What's the point in forgiving someone who is dead? We're not going to forget mass murder, and we're not going to prosecute anyone for it either (now that they are all dead or prosecuted). So why forgive? What does it even mean to forgive someone you don't - and can't - know personally?
rokmonster
Lives under a rok.
1yr ago
There is another aspect to this. If you allow the victims a say in the punishment it increases the probability of victim intimidation for certain crimes, especially if the legal punishment is weak. The victim of a house invasion for example has to worry that requesting a higher sentence will result in their house being invaded again as soon as the perpetrator is out of jail. In Korea this was recently an issue as a 12 year old rape victim decided not to pursue charges against her home invading rapist, who was looking at 2 years in prison.
But code is just math, math is just code, and protein folding is just the intersection of the two!
rokmonster
Lives under a rok.
1yr ago
As a single point of data, a far-left millennial friend of mine (who probably thinks I'm a stealth conservative based on the conversations we used to have about feminism) blew up when I said I had started a diet and begun to lose weight: "Dieting doesn't work," "I've noticed you have some toxic ideas about weight," "I've heard you making insensitive comments before," etc. They then recommended I read a Health At Any Size activist's autobiography to engage with these ideas because she's "very eloquent."
The book was mostly personal and second-hand experiences of trauma due to people pointing out morbid obesity and its negative externalities. There was a whole chapter about doctors and the medical establishment being shaming and misguided. I scanned the bibliography for any academic papers (now on my reading list), but most of the references were to articles from the likes of Huffpost. Then I gave up on the book.
As regards the issue of obesity, I do think the problem is systemic: the US population has lost its ability to cook proper healthy food at home, has lost the last remnants of a culture which despised "gluttony," and has been brainwashed by Big Ag to think that eating more and more is normal, and that it's all genetic. Meanwhile Big Ag has hired flavor scientists to engineer hyperstimulus into food, hires lobbyists to keep politicians from addressing the problem in any meaningful way, and pays useful activist idiots to write books about Health At Any Size.
However, as an individual, I don't have many options other than to tune out the propaganda, establish my own system, and live my life of moderation. If that makes me "toxic" and "insensitive," ... fuck it. I'm not sacrificing my pursuit of excellence for some moral fashion.
rokmonster
Lives under a rok.
1yr ago
If you're actually raising a child, would you tell that child at some point in their life that you had had an abortion previously?
I am in the situation of being the baby that was supposed to be aborted, but contractions started and so I was born instead. It wasn't really a choice on my mother's part: she needed to undergo chemotherapy and they couldn't do it while she was pregnant.
I think I first understood the import of this story in late middle school, and it didn't have much of an effect on me: I was surprised, but my parents love me and our relationship didn't change. I probably went back to thinking about what teenage nerds think about. Now that I think back on that story it just makes me feel lucky. But I'm an outlier who leans toward stoicism, positivism, and "life is tough. Get over it". YMMV.
rokmonster
Lives under a rok.
1yr ago
Ok. If you are specifically talking about the people who rule nations, I can agree with you that cost externalization is rampant on the basis of the many nations that have been financially ruined by their wealth-extracting elites.
However, there are whole categories of private proprietors whose business models are explicitly extractive. Notice how most of the pre-Covid profit of large corporations was due to "financialization," which translates to roughly "taking out low-interest loans on our assets, guaranteeing short-term profits." You may also notice how most private equity deals are structured to merely externalize long-term costs onto subsidiaries (and other investors) while extracting short-term profits, or how whole sectors of the economy will lobby for special protection. What is a farm/CHIPS subsidy, if not externalizing the cost of business?
I'm skeptical the Kursk was about national pride at all, although I would believe you if you told me it was corruption or officer-level CYA. Nuclear submarines, their limitations, their strengths, their uses, and their construction are highly prized national secrets, to it stands to reason that the Russian Navy would be reticent to welcome foreign aid, let alone rescue subs or divers from NATO navies, which would no doubt be beaming video direct to Langley.
To put it another way, from the perspective of Russian Naval command, the secrets of the Kursk are arguably worth more than the lives of the crew (even before accounting for corruption and CYA), as those secrets protect all the other submarine crews. But telling that to the public in so many words is a great way to ruin future crew recruiting efforts.
rokmonster
Lives under a rok.
1yr ago
It was bungled, but I'm skeptical face played any part in that calculation. Doesn't the rationalization "the lives of 10 sailors is worth our nuclear submarine secrets" have enough explanatory power?
I'm moving to a job on a campus in the US. The question of what to do about social justice, political conversations, and social justice training requirements has been vexing me for a while. I just got my first email from someone who has pronouns in their email signature, with a link to the campus policy on pronoun use. (Tl;dr: staff are "encouraged," to use pronouns and "expected" to treat people in accordance with their claimed pronouns.)
Here are my options:
Poe's strategy: Agree and amplify. I use all pronouns as claimed. I believe we should racially segregate as much as possible because that would be good for making Black communities into safe spaces for Black bodies, but we should do this by forcibly unhousing white people, because anything else would be gentrification. I take full responsibility for the racism of all the people of my race, and think we should give full reparations to all Black bodies. This probably codes as high-class, but there is a large chance of being unable to keep up with the charade and a small chance of being cancelled as a result. There is also a chance of value drift and the mental risks inherent in living a lie.
Mainland Chinese strategy: I don't talk about politics or social issues at all. If asked, it's because I can't keep up with it. (This is mostly true!) Probably codes as low-class in the US, but I won't be cancelled for my opinions, because I don't have opinions... at least until BLMII (LGBTQIA+ boogaloo) comes around and everyone who doesn't fly the Rainbow-BLM flag is cancelled.
Mask strategy: I don't talk much, but when pushed I shrug and concur with moderately pro-SJ shibboleths that I still believe. When in private with a trusted interlocutor, it's mask off. This is what I currently do, but SJ isn't a significant factor at all in my current social environment, so I am able to spend more time mask-off than mask-on.
Earnest SJW strategy: This is the highest-class option, but I don't think I can pull it off. I don't know the language, I'm doomed to stumble, and don't want to break my reasoning capacity that much, and it goes directly against my values, interests, and tribe.
Earnest Mottizan strategy: True honesty. I oppose SJ (because it is in direct opposition with my values) and I'm not afraid to say so... in a friendly way with a smile on my face. I support equal rights, but not equality of outcomes, which I don't think is ever possible. I think most SJ is just an elite conspiracy to shift focus away from class issues, with the richest of the rich supporting it because they are wealthy enough to avoid its negative side effects... which hurt black people too. I think unfettered immigration is bad for blacks in America, and I don't think SJ really helps the people it seeks to help, instead infantilizing them and removing their agency. I think the biggest problem facing black america is lack of interest in education, and the biggest problem facing women in STEM is that STEM careers suck: the pay is for tools, and no smart woman would enter them when other careers are low stress and more lucrative. My experience living in a more conservative society has taught me that most SJW claims are false in traditional societies, etc. Etc.
So, I guess what I'm asking is: what's the safest strategy, what's the best for my career, and what's the best way to spread my values? For those of you in US academia, what approach do you follow, and what works or doesn't work?
More options
Context Copy link