matherial

You have 20 kids in a classroom. Fifteen of them want to know what's going to be on the state-mandated test. Five, if you're lucky, may have some interest in the subject you're teaching.

So, do you design the class for the five gifted kids and risk failing the other 15, or do you get everyone over the finish line through rote repetition until they can all recite the names of the Founding Fathers for the test, knowing full well that you can't make them care about history (or, in some cases, even keep them off drugs and street crime)?

Sure, there are exceptional teachers who can keep the entire classroom engaged, but it's so rare that we make films about them. In most cases, in public schools, it's a choice that needs to be made.

> I get your point, and the book does say that. It's just... For a teenager, it rarely lands.

Basic education is weird because it has two conflicting goals. One is to get everyone to the same basic level needed to function in a society. And that's where all the incentives are - that's what you and the teachers are graded on. But another goal is to give you a taste of what's out there, so that some of the more talented kids can decide they like literature or math or physics or whatever.

That second part really requires the teacher and the student to care, but the first part conspires to make that outcome unlikely. My memories from public school are that everything I'm interested in today, I'm interested in despite the efforts of the school system to make the topics appear painful and dull.

So on some level, I don't think the problem is with the literature itself, or with the student's age. It's just that the setting often sucks and you kinda start with the assumption that if the teachers are making you read a book or learn some dates, it won't be a fun experience. And so it isn't.

I wish the author lived by the same principle, seeing how the text on the webpage is pretty evidently AI-generated.

A common way of thinking at Google and other big tech companies is that you're not building for your current users. Your current users will hate it if you move or change the color of a single button in Gmail. Also cue the "faster horse" aphorism. You're building for the next billion users instead.

But of course, the problem with Google is that they already have basically every person on the planet as their customer, and their ideas for how the future should look like aren't exactly transformative. The Gmail UX in 2026 is not radically different / better than in 2016. The same goes for Chrome - there's a lot of changes under the hood, but the user experience is essentially the same. Android phones have better cameras and displays compared to ten years ago, but the software functionality is once again very similar.

So I guess they're in the same spot as Microsoft, trying to innovate in a mildly user-hostile way, looking for an audience that doesn't exist?

Zoning is a per-city and per-county thing. Traditionally, they usually occupied some sort of commercial / light industrial zones, the same as office parks and warehouses. But hyperscale stuff often requires one-off variances / rezoning.

There's no other account. The period is disregarded. When you sign up for foo.bar@, you effectively get foobar@ and all the other variations (foob.ar@, etc). No one else can sign up for these. Quirky but not terribly interesting.

No. I was on a free plan for years and found out by accident. It's interesting that you set "nefarious" as the bar. I'd simply call it "sleazy".

I'd just set my wallpaper to Hasselhoff as soon as I found out about the tradition and be done with it.

More seriously, the wallpaper stuff is OK, but I never found some of the more disruptive office pranks on the list to be particularly endearing (e.g., switching keycaps, taking the ball out of a mouse). It's work, and you sometimes need to do something urgently, and it's annoying to troubleshoot or clean up after someone wrapped your entire cubicle in shrink wrap, tied your shoelaces together, or something like that. If it's more fun to you than to the recipient, don't do it?

I did some office pranks, but tried to keep them nice. A tall stack of TPS reports on a manager's desk, sure.

Because your institution's PR department and the journal you're publishing in want the publicity of "scientist cures cancer". And you probably wrote "I'm curing cancer" in your grant application, team charter, or whatnot.

It's an AI-generated post on another ephemeral AI-generated blog, now a multiple-times-a-day occurrence on HN. We're taking issue with what a chatbot thinks about "leadership". And some people will probably show up and say it shouldn't matter who wrote it, but it obviously does. There's just some undeniable comedy in this.

How so? The article is talking about a fictitious person that's listed by name only on a sanctions list despite the government knowing the person doesn't exist. There is no possibility of the real Sean Byrne presenting a government number that is a negative match against the list because the fake Sean Byrne has no known ID to match against. The essence of the entry is just "here's a name to avoid".

The real Sean Byrne can produce a government-issued passport number, but that's clearly not enough for Apple.

I think it's more of a problem of all these "hyperscale" platforms where the cost of not being zealous enough is long litigation and devastating fines, while the cost of losing a single customer (or a thousand) is basically nil. This leads to all kinds of opaque, customer-hostile outcomes like this, also if you trip some filters not related to sanctions / mistaken identities. There's a recurring theme of HN posts along the lines of "an automated process at Google cut my business off, HN plz help".

Do you also get upset when people buy Lego sets, model airplanes, or jigsaw puzzles? It's the same thing.

And FWIW, it's the exactly same reason most people were buying DIY electronics kits in '70s, '80s, and '90s.

> This part has been beyond baffling to me. The last thing I want

Most people don't think about it when buying a car, or they have no practical way to evaluate it. So it is one of these things you probably need to fix with legislation, except the legislators love the idea too because sensor-rich, always-online cars give them more tools to police the society.

I don't really know what to do with that, short of going neo-Luddite. People often see the excesses of ad tech as a failure of capitalism, but in a sense, we're seeing a tech-driven failure mode for free, democratic societies. In a world where your online presence is tied to your identity and always under the watchful eye of a large language model, and where you can't move from A to B without leaving a digital trail, it's not gonna be fun if you become a thorn in the government's side.

Right, which is still few steps removed from the actual genocide. IG Farben was a direct participant, to the point where a lot of their execs were tried and convicted at Nuremberg.

But I'm not sure the focus on corporations is all that interesting here. The Allies eventually let most of the senior Nazi leadership walk free, and few in Germany faced any lasting consequences for what they did during the war (even IG Farben execs were released before too long). I think the idea was that they needed to govern the nation they have, and mass-imprisonment wasn't a viable strategy. It worked out in a pragmatic sense, but it was hardly justice for the millions of civilians exterminated by the regime.

> Waiting a year to get from intuition to theorems is a perfect way to ruin math.

It's interesting that you chose to make that point in a thread about calculus specifically. It had pretty shaky foundations for most of its history, and even today, there's a pretty significant disconnect between the mechanics of epsilon-delta and the meaning we assign to the result.

> Math is not supposed to be easy/simple, but it supposed to be a great way to understand systems based thinking.

Math is a means to an end. Making the tool easy to use is a desirable property. I've heard "it's not supposed to be easy" applied to many disciplines, from film photography to software engineering, and I think it's mostly gatekeeping.

> However, although G is undecidable, it’s clearly true.

That's... not really true; it's surprising to see it in Quanta, of all places.

Godel's (separate) completeness theorem says that in first-order logic, anything that's semantically true in all possible scenarios can be syntactically proved. So, if G is "clearly true", that ought to make it provable.

The theorems don't contradict each other because in FOL, G is not guaranteed to be true. Its truth is independent of the machinery Godel put in place.

It's not something you really need to get into an introductory text, but it actually makes the whole outcome easier to grasp, and leads to many more counterintuitive results, such as Skolem's paradox.

What are you trying to say? That Facebook should be excused because others also behave badly?

Otherwise, we can dislike both things and it adds nothing to say "but why aren't your talking about the other thing instead". And for what it's worth, there is a pretty clear difference between the journalistic standards of FB bait and of the legacy media.

You don't need to accept at random. I'm selective - I only connect with people I worked with or know well from the industry - and it's still 99% AI slop.

Part of the problem is the tech industry. There's plenty of people trying to get rich quick in tech and LinkedIn is essentially made for them. I don't see it as much in other fields where most people don't imagine themselves building unicorn AI startups with VC money and becoming obscenely rich.

It is a way to keep up with opinions about industry news, but at least for tech, I think this is all downstream of what's happening on "short-form" social media. It breaks on Twitter and then you get LLM-powered thought leadership on LinkedIn for the next two weeks.

How so? They have the same radios, the laptop has more human input devices (touchpad, trackpad, keyboard, pen), quantitively better almost everything (more RAM, much better CPU / GPU, much higher pixel count, faster and higher-capacity SSD, better audio, etc)... laptop cameras are the only thing that's obviously worse.

What an odd website. It explains everything in incredibly childish terms ("The display you’re reading this on is an absurd grid of exactly [wrong number] tiny, multicoloured light bulbs, known as pixels."), has lots of needless word-salad LLMisms ("Electricity made light effortless. The torch in your phone made darkness optional."), and all in all, feels like a wasted concept because it doesn't really say anything that isn't patently obvious: a phone has a screen and a camera and a CPU... okay? Not exactly mind-blowing.

It doesn't really even do anything to support its thesis that it's the "most intricate machine you've ever held". I don't know, I'm holding a Thinkpad laptop right now and I'm pretty sure it's a superset of what's inside a phone.

I guess this is where our true colors show. There's a significant contingent of HNers who always dunk on LLM text detectors and claim that they can't possibly work, that they ruin careers, etc. But now that a lab says "OK, we'll add a real watermark", the reactions are overwhelmingly that it's still somehow wrong.

Why do feel so entitled to being able to pass LLM-generated text as our own? I get that a lot of techies aren't good at writing. I also see a lot of tech hustlers who like to use LLMs to fake human connection and compassion - I've gotten LLM-generated recruiting emails that talked at length about how the recruiter "valued" my work. Just because we found a "cheat" button doesn't mean it's wrong for others to want to know.

Yes, LLMs are great. So is transparency. If you think an LLM writing is your new superpower, wear that badge with pride. It might mean you will lose some business from LLM haters and win some other business from like-minded customers. C'est la vie.

You have less latitude than in prose, but I think you're underestimating how much can be changed without causing breakage. For example, the most likely sequence might be "if (!foo) bar; else baz;", but you can also say "if (foo) baz; else bar"; substitute "foo == 0" / "foo != 0" for even more variety. Similarly, "foo = 1; <NL> bar = 2;" can be output as "bar = 2; <NL> foo = 1;".

Keep in mind that the LLM "sees" the previous (tweaked) output and picks what makes sense based on that. There are few situations where a perturbation like that would be unrecoverable, and I assume these situations also correspond to a huge probability difference between the most likely completion and the second most likely one - in which case, the watermarking algorithm can choose not to touch the token.

Minimal. I'm behind Cloudflare and 90% of the traffic is still scrapers. I don't think they're serious about the long tail.

I think the main thing Cloudflare is trying to do is block direct traffic from frontier labs and then start charging them for access. They might end up shooting themselves in the foot, as this simply empowers sketchy residential-proxy outfits to undercut Cloudflare and sell the data to labs for less.

That's not really true? Housing in depopulating / depopulated areas can be easily 5-10x cheaper than in premium metro regions.

In picturesque locations, there's upward price pressure because well-off urbanites sometimes buy vacation properties, but that's hardly a new thing.

All silicon capacitors are relatively expensive, they're specialty parts that generally go into expensive things (mmWave airport scanners, optical backhaul, etc).

No, not if you want 22 nF. That's a pretty substantial capacitance. It would need to be quite big on a PCB. That would mean significant inductance, and thus, bad performance at high frequencies.

This is, without exaggeration, probably the fiftieth blog post or long-form comment about how someone is using an LLM for "complex learning", and I'd just really like to see at least one of these to be accompanied by a statement saying what are the kinds of problems the author can now confidently solve that they couldn't before.

In my experience, LLMs are really good for taking up your time and making you feel like you're learning, in the same way that many of the popular educational videos on YouTube are fun to watch and don't really teach you anything.

If you ask an LLM to give you a 500-word summary of quantum physics, it'll give you an oversimplification that probably leans on a hodgepodge of pop-sci metaphors. And if you start drilling down, you risk drilling down on these ELI5 metaphors, which can get you farther away from truth.

You truly envision a future where every program is written in a custom programming language, for a custom operating system, for a single user who will now be in charge of understanding and maintaining it forever? That user being your grandma, your baker, your CEO?

What? No. Plants colonize new areas in complex succession sequences that require hundreds of years to play out. It might start with grasses, then shrubs, then dense stands of trees that die off and are replaced by stands of other trees... maybe there's a fire that needs to happen in between... and let me remind you that a "city park" is not exactly a natural formation anyway.

If you want a park in 30-50 years, you need to plant trees and design it with specific public-use objectives in mind.

It gets little use in the sense that only a small fraction of vulnerabilities are reported there, and there are very few non-advisory discussions (often by the same 2-3 people).

It does get use in the sense that every now and then, some vendor sends 50 emails that could've been one (most recently, some Apache Qpid thing). But I wouldn't call that part valuable.