g9yuayon

> Once it has finally released, it usually remains stagnant in terms of having its knowledge updated....meaning that models will not be able to service users requesting assistance with new technologies, thus disincentivising their use.

I find such argument weak. We can say the same thing about a book, like "Once The Art of Computer Program is finally published, it usually remains stagnant in terms of having its knowledge updated, thus disincentivizing people to learn new algorithms".

> Solve all the problems at the end of the chapters in the textbook

It really depends on the textbook, isn't it? I find it impossible to solve all the problems in CLRS, for example. Our professor assigned one of the problems about universal hashing, and it took me hours to get the key insights to find the correct proof. I can't imagine how one can solve all the problems given so many competing priorities, except for a few truly talented.

> Pushback: Is math acceleration equitable?

The fact that the author felt necessary to even discuss this shows how sick the US culture has become. Was it equitable that Newton could invent Calculus before he was 26? Was it equitable that Poincaré could manage to work only 4 hours a day yet still be the most successful mathematician in the world? Was it equitable that someone could get the idea of limit in one pass yet some kids struggled to understand even what percentage is? Was it equitable someone could beat the shit out of everyone in her class by merely engaging in class, while someone could flounder their math class despite taking 20 hours of tutoring every week?

Since when the US elites have not been able to recognize that we should nurture talent instead of suppressing it?

I'm not sure the title captures the essence of a school, at least for STEM subject. For instance, why is doing maths in schools not "doing"? We solved the same math problems that the real mathematicians had to solve hundreds of years ago, no? The way we tested the Avogadro constant was not that different from a real chemist did a hundred years ago, no? We didn't do the modern ones mostly because the students had not the advanced prereq yet. I think it's more accurate to categorize what we do in school as controlled doing. Yes, we abstract every object as a point mass or a rigid body when learning Newton Mechanics, but that's only because the real-world is too complex for a high-schooler to handle. Yes, we implemented only a single-threaded SkipList using standard C library when taking the first course on algorithms, but that's only because most of the students wouldn't have enough knowledge to go beyond that.

Is it just me who feels that Anthropic has been innovating faster than ChatGPT in the past year?

An elementary student in China probably won't learn more than 3000 characters, yet they can read advanced novels. I suspect the difference is that a native speaker feels more comfortable skipping or guess unknown words. I certainly felt more discomfort when encountering new words when reading even a popular fiction in English, while having no problem guessing the meaning of a new word in my native language.

> The importance of an SRS system like Anki cannot be overstated.

I'm not sure if I agree with that, as no native speakers need to have an SRS to learn their native languages. No doubt that SRS will allow us to remember words, yet few can really acquire those words intuitively. When starting to learn English in school, we used some kind of SRS system to memorize words and phrases and sentences, and man, the result was abysmal. We spent 10 years learning English (3 years of middle school, 3 years of high school, and 4 years of college), trying to memorize new words every day, passing TOEFL and later GRE through intense SRS, yet few students could understand TV shows, read fictions, or communicate with English speakers. And the learning was arduous, to say the least.

In contrast, I was lucky enough that my mom gave me a set of graded readers compiled by National Geographic, and simply asked me to read them through. And then Sidney Sheldon's books, Friends, etc and etc. So basically I immersed myself into the language, never having to do SRS, and I could easily pass TOEFL and the GRE Verbal years before graduating college. As a bonus, I started to enjoy TV shows and movies early on, and was able to socialize with my classmates and professors without even trying. I also used the method to learn Spanish and Japanese, and the results are similar. No SRS needed but consistent exposure to the languages. In less than two years, I can read books like The Alchemist, If Tomorrow Comes, and Project Hail Mary. Another interesting contrast is that I couldn't understand much conversation in those languages, precisely because I spent most of my spare time in reading.

Do US schools and media educate the public that taking drugs is dangerous, ruinous, self destructive, and shameful. I understand that the US in general favors personal freedom, but shouldn't there be some universal values, like don't take drugs?

Funny that some also argue that it is why the left, no matter how extreme they are, often win in the end. The left keeps fighting, always trying to change the status quo. If a legislation attempt fails this year, they will just try again the next year. Year over year, whoever keeps fighting wins. In contrast, conservatives just want to be left alone, therefore the conservatives lose often.

> from Barack Obama doing cocaine to Donald Trump with a pregnant woman who (vaguely) resembles Kamala Harris to Trump and Harris pointing guns. With US elections approaching and X already under scrutiny from regulators in Europe, it’s a recipe for a new fight over the risks of generative AI.

So the journalists are saying that they are the moral judge of what's right and what's wrong and they can decide which information should be suppressed? Coming from China and knowing such much history about how communist countries treated their people, I'm deeply suspicious of these journalists. Just look at China, look at Cambodia, look at Cuba, Look at Romania, and of course look at Germany. When didn't it start with elites advocating for suppressing speeches for the sake of high moral ground?

When I grew up in China, students in a school were divided into fixed classes. Those classes formed great communities, as we spent hours every day for at least three years and some for 6 years. Each class had a head teacher, who fostered the sense of community too. No one would mock people for geeking out. No one would mock people for not being good at sports. No one would mock those who struggled at academics. At least not openly. We loved each other and still do. Our bond was so strong that we had regular reunions every few years, and most of my classmates would make it. We had multiple couples who were high-school sweat hearts, even though dating in high school was a taboo in China then. The concepts like nerds, like queen bees, like sports jockeys, like that those who can get drugs and drinks are popular... They were all new and parts of the culture shock to me when I moved to the US.

When I started to study English many years ago, my mom gave me a book series titled America Today or something like that. The book was life changing as it made learning English so much easier than the Chinese way at that time and it taught me a lot about the US. The book 3 devoted many chapters to a day in someone's life, one of which is A Day in the Life of a Farmer. At that time, the US was the beacon of the modern civilization to Chinese people, a superpower that was far ahead of everything we did. And despite that the work of a farmer was still tough. They still needed to get up at 4:00am or early to take care of their livestocks. They still had to take care of their crops before the sun fully rose. They still had to handle many issues about their land, their equipment, and their business. They still had to do tons of intense manual labor. Despite all the toughness, though, the chapter also conveyed the idea that farming can be gratifying and fulfilling. I certainly appreciate this kind of optimism and appreciation, and the life an American farmer has a lasting impression until this day.

I haven't used QQ for years. Now that I look at them, somehow they look less interesting than what I had in memory. Maybe I'm subconsciously comparing them to WeChat's or many of the user-created variants.

> I don't think there's ever been a good "looking skeptical" smiley anywhere, though.

There used to be one, a serious face that strokes its goatee. That emoji can be used to express doubts or "that's interesting" in a joking way. Not sure why that emoji disappeared.

Speaking of group chats, I recalled an interesting story how MSN Messenger lost Chinese market. Back in 2003ish, MSN Messenger was THE shit in China. QQ, the competitor of MSN Messenger by Tencent, was losing ground rapidly, especially to white-collar users in big cities. To Microsoft, MSN Messenger probably was a toy, but to Tencent it was life and death. So QQ launched group chat. In addition, QQ Group Chat allowed users to drop any images, including GIFs into a chat. And man, that feature killed. Hundreds of millions of users got hooked. It was just so much fun to be creative and spread memes using this features. In the meantime, what was MSN Messenger's response? Absolutely nothing. No, it's worse than nothing. It was insulting. If I dropped a gif to a chat, MSN Messenger would always convert the gif into a 8x8 icon, and sometimes turned that gif into a static image. BTW, I'm not sure why, but big techs in the US somehow couldn't get Chinese market. Case in point, the emojis in all the chat apps, be it Apple's or Meta's or Google's or whatever, are just ugly and not expressive, when compared to WeChat's, let alone QQ's. Of course, as a Chinese user I'm biased, but that's the point: very few Chinese natives like the emojis offered by western countries (and I guess vice versa).

And then, well, there's no "and then". The rest was history. MSN Messenger quickly became a niche and lost its user base. It made Chinese companies realized that they had a chance to out compete behemoths like Microsoft, by moving fast, by moving decisively, and by being creative to fit the psyche of Chinese market.

Well in this case Chinese is like Spanish and Japanese, a verb can also be used as a gerund.

The All-In podcast mentioned research on the concentration of micro plastic particles in human body. I was wondering if banning single-use plastics can somehow reduce that concentration as well.

Yup, 余光中 is one of such authors who urge people to write authentic instead of "westernized" Chinese. I wouldn't call it "blame", though. Just like in English, there's certain standard on good writing. In Chinese, 白话文 is really the way to go. That's why among the authors in the May 4th movement, 钱钟书´s modern Chinese is just so much more pleasant to read than Lu Xun's. If you don't believe me, try reading his translation of Nikolai Gogol's Dead Souls. Oh my...

Joan Pinkham wrote a book titled The Translator's Guide to Chinglish[1]. It's a fun book to read and really teaches native Chinese speakers how to write good English. Interestingly, a lot of the lessons in that book can be applied to Chinese writing too, such as avoiding nounification, and removing duplication. For instance, just say "dislike" instead of "to have a dislike for". And instead of saying "give guidance to", just say "guide".

[1] https://www.amazon.com/Translators-Guide-Chinglish-Joan-Pink...

Note that modern Chinese is heavily influenced by modern English. During the May Fourth Movement[1], prominent authors like Lu Xun diligently explored how to write modern Chinese with "westernized" style. The experiment largely failed, but modern Chinese did get influenced a lot, to the point that multiple authors wrote books or articles pleading people not to write "westernized" Chinese. A typical example is nounification of verbs, something that traditional Chinese never had. In contrast, younger generations love to say Chinese equivalent to something like "do improvement" instead of "improve" (进行改进，instead of 改进, even though it is still considered bad writing style.

1. https://en.wikipedia.org/wiki/May_Fourth_Movement

> 268,481 words (16 hours 17 minutes) with a reading ease of 74.9 (fairly easy)

It's interesting when China introduced Ulysses years ago, publishers and reviewers often mentioned that how challenging the novel was. Often cited reasons included extensive use of stream of consciousness, long and complicated inner monologue, multiple languages and idioms used, and large vocabulary.

My experience as well. I tried it a couple of times more than 10 years ago, one is covering Nancy Lynch's Distributed Algorithms, and one is Maurice Herlihy's The Art of Multiprocessor Programming. We finished about 1/4 of each of the books before running out of momentum. People simply lost interest as the content became increasingly challenging. We talked about the reasons, and the overwhelming feedback was that people lost interest because they failed to see what they learned could be applied to the job in short term, so it looked to them that the effort of learning the hard stuff wouldn't pay off.

This sounds like the trend is good for the incumbents? That is, it's hard for the supply to catch up with the demand in broad scope, so people with solid experience will be marketable for a while.

> wrong things (leetcode-type challenges, unnecessarily specific knowledge..

I'm curious about the dynamics of this. If leetcoding or requiring some specific knowledge is too much for candidates, those employers will not be able to find enough talent, so they will be forced to lower their standards and the mismatch should be gone, right? Case in point, I'd venture to guess that no more than 10% of the tech workers truly understand AI or stats or machine learning or basic college-level maths, yet magically millions of ML engineers or data scientists have popped up in the past few years just because the market had such high demand.

> Structural unemployment happens when the skills of the work force don't match the needs of employers, so there is both unemployment and difficulty hiring.

Any study on why there would be this kind of mismatch in the tech industry? I'd imagine that tech industry is the most transparently competitive and least regulated sector so there should be a reasonable balance between the supply of skills and the demand. In addition, there are so much resource that teaches people all kinds of tech skills, so people should be able to move from one skill set to another if really needed.

I actually have a contrarian view: being able to do elementary math is not that important in the current stage. Yes, understanding elementary math is a cornerstone for an AI to become more intelligent, but also let's be honest: LLMs are far from being AGIs and does not have common sense nor general ability to deduce or induct. If we accept such limitation of LLM, then focusing the mathematical understanding of an LLM appears to be incredibly boring.

Do you have examples? The other commenter's example of generating correct unit test results is indeed interesting, and I was wondering if there are other cases.

Is being good math that important to the ChatGPT users, though? The ChatGPT's ability to do math is so limited that I'm not sure what math problems we have to ask ChatGPT to solve.

> PjM/PM time to organize initial build

This sounds what big companies or a disorganized company would need. For an efficient enough company, a project like this needs just one or two dedicated engineers.

In fact, I can't imagine why this project needs a PM at all. The database is used by engineers and is built by engineers. Engineers should be their own PMs. It's like we need a PM for a programming language, but no, the compiler writer must be the language designer and must use the the language. Those who do not use a product or do not have in-depth knowledge in the the domain should not be the PM of the product.

How is this compared to Obsidian? Currently Obsidian is by far the best Markdown editor I've used. The inline preview mode is amazing, and I appreciate how easy it is to embed math equations and diagrams. Speaking of this, shame on Google. The editor of Google Docs is so awful that I can't it is built by Google.

Can someone educate me why insurer should not know one's driving habits? I'd imagine that the risks calculated from one's driving habit will be more accurate than that derived from only past accidents, car color, user profile and etc.