groby_b

They are also collectively rational, as a response to an ecosystem that's spun out of control and habitually consumes rats nests of dependencies.

Early participation and beta programs are outsourcing careful engineering via making everybody else guinea pigs. If we want to sling around accusations of free-riding (really?!), you're slacking on testing and free-riding on your early users.

And yet, aviation safety keeps improving.

I didn't miss that point. I'm saying it's blown out of proportion, and that diminishes the value of the actually important content.

I really wish we'd stop arguing about AI with an "some automation failed, so all automation is bad" approach.

Yes, AF447 crashed due to lack of training for a specific situation. And yet, air travel is safer than ever.

Yes, that Tesla drove into a wall, and yet robotaxis exist, work well, and are significantly safer than human drivers.

Yes, there are a lot of "witchcraft" approaches to working with AI, but there are also significant accelerations coming out of the field that have nothing to do with AI.

Yes, AI occasionally makes very stupid mistakes - but ones any competent engineer would have guardrails in place against.

And so a lot of the piece spends time arguing strawmen propped up by anecdotes. And that detracts from the deeply necessary discussion kicked off in the second part, on labor shock, capital concentration, and fever dreams of AI.

The problem of AI isn't that it's useless and will disrupt the world. It's that it's already extremely useful - and that's the thing that'll lead to disrupting the world.

That's a bold assertion without any proof.

It also means you're so helpless as a developer that you could never debug another person's code, because how would you recognize the errors, you haven't made them yourself.

HTTP 451

"We recognise you are attempting to access this website from a country belonging to the European Economic Area (EEA) including the EU which enforces the General Data Protection Regulation (GDPR) and therefore cannot grant you access at this time. For any issues, e-mail us at info@franklinnews.org or call us at (847) 497-5230."

This is extremely funny given it's an article about privacy concerns :)

I see we're once again missing the existence of indirect impact. There's a reason organizations look at revenue/engineer overall instead of trying to attribute it directly to specific teams.

I guess his students get to relearn that on their own.

Also, any post talking about building software and then contains the suggestion that "cost per unit" is an efficiency metric needs to come to the red courtesy phone, Taylorism would like to have a chat about times gone by.

"The principal security problem of von Neumann architecture is that there is no architectural boundary between data and control paths"

We've chosen to travel that road a long time ago, because the price of admission seemed worth it.

> It's not a weird sentiment to want to know what benefits a gift is providing.

"I bought you tickets for your favorite artist for your birthday. I expect a detailed trip report" :)

Yes, you're right, personal gifts aren't donations, but then maybe we should stop calling donations gifts, too. Gifts are given without any expectations attached. Donations do and should have expectations.

Very much not.

Its origin is as a mathematical notation for algorithms. It was used to publish research reports and (IIRC) a book or two.

You're confusing "possible to read" with "accessible to people unwilling to invest any effort understanding"

Without showing the prompts and responses, it's yet another meaningless AI benchmark.

Many of those numbers do not really match what I've seen in the wild, and without clear illustration why you arrived at the number it's not a helpful number.

The issue isn't even just the prescription.

It's that your health care system the doctor is in builds a few extra hurdles. I've talked to my (non-tele) doctor about GLP-1. I've tried losing weight before, with her, there's a long history.

To get approval, between the hospital my doc is in and the insurance, I need to:

1) Have a BMI of >30. Since it's only 29.5, I get to stuff my face if I want to lose weight.

2) Have six sessions with a nutritionist. Which are massively useless, their advice is roughly equivalent to reading Cosmopolitan. I know because I had prior conversations, and they're documented. But still, gotta do it again.

3) Do six months on Weight Watchers. Which is one massive scam leading you right to disordered eating. Also, I've tried for years to lose weight via diet changes, documented and talked through with my doc.

4) Before I can get tirzepatide, I have to get semaglutide for three months to see if it works. Never mind there's study over study over study showing it's slightly less effective and has massively more side effects.

Or I can just cough up the cash directly and buy from Eli Lilly, if somebody signs that receipt.

I'm fortunate enough I could afford that, so I did. (After a second consultation with my family's doctor back home - both they and my doctor agreed it was appropriate, so it's not just a case of "wanting is enough")

And after six months, my weight was in a much better region, lipid panels were much improved, other related biomarkers looked better as well - exactly as numerous studies and my doctors said one could expect.

So, as long as I cough up enough money, sure, I can bypass all the hoops. My health didn't enter the equation, just screw the poors (whose treatment for worse outcomes because they couldn't get access will cost a whole lot more than GLP-1 would've cost).

So, fuck the "prescription hurdle" and the medical system in the US with a hot white glowing iron rod right up the ass.

As for "these are not drugs that anyone can safely take without guidance", that's not really true either.

They're neither hard to take - "inject one vial once a week into the flabby part" isn't rocket science - nor does it cause massive health risks by itself. (And the hazard ratios for diabetes 2 and cardio events are so spectacularly low that they dwarf the other risks)

Yes, talking to a doc is a good idea. No, the current gatekeeping is in no way necessary.

The article is written using AI, so unless you verified the complaints, the safe default assumption is that they don't exist.

> If you followed the law word for word the authors of e.g. curl could be charged under this law.

They really couldn't. BVerfG (Germany's constitutional court) has clearly said that dual use tools have a presumption of not being tools to break the law. It's been very clear that mens rea matters. And that a narrow reading of the law is the only constitutional reading.

The problem here is taking "word for word" as "by dictionary meaning", which is never how laws are read.

It's still a problematic law (together with §202a/b) because it doesn't clearly carve out space for grey-hat activities (white-hat attacks with authorization really don't fall under it even with creative reading).

On the upside, Germany is considering fixing that. On the downside, it moves with the speed of classic German bureaucracy and is being "discussed" since 2024.

Then maybe it's not worth using Claude Code that much.

> I have little interest in helping people that don't want to make an effort.

What you were trying to say is "I can't be bothered to engage with people who are not taking my word as gospel".

That's not what I said.

I'm absolutely supportive of using blackboards & paper and pen over computers.

What I'm saying is that making unsubstantiated claims like "you'll hear the lecture again when you read it" is completely detrimental to making that point, because it's entirely unsubstantiated and doesn't make any relevant point you couldn't make in a well-supported way instead.

> It worked for me. Have you tried it?

I have. It doesn't work that way for me - but that hardly matters. More importantly, there's plenty of research around inner monologue and sensory replay also pointing out that this isn't true for most people.

> I attended a university where that wasn't a problem.

You were blessed :)

> But, frankly, I liked his in-person lectures using the blackboard and chalk better.

That's not something I'm arguing against :) I think they're a great teaching tool.

My objection is that making a specious but unsupported, and often easily anecdotally invalidated point to support a case that's actually got a ton of solid evidence in support is detrimental to making that case.

Physical note taking, and being actually present for a lecture are tremendously important. Laptops are hugely problematic for learning. And those points are important enough that we should make solid arguments in favor of them, not easily discredited ones. Because we also know that many students are very muched biased to discarding these points given half a chance.

There's currently a related post on the home page: "Good ideas do not need lots of lies in order to gain public acceptance (2008)"

You might value it differently, though.

The company also would be giving you money if they, e.g. , gave you sets of annual Disney tickets, or a family membership in a luxury gym.

That's the whole point - there's an unlimited amount of potential outcomes that your original hard limit doesn't cover. Of course your final decision can be targeted towards "net value to you", but fundamentally, you can only assess net value once you have the facts.

Your hard limit becomes "I'll quit if I don't like it". Which, really, you don't need a "hard limit" for.

Probably. Hey, you picked a risky thesis heading into large uncertainty, that's not free.

But for the folks who are dead convinced it's all just a scam and it will quickly collapse, that's a way to put their money where their mouth is.

> When studying, going over the notes, you'll hear the lecture again in your head.

That is a) a BS claim and b) wouldn't be a feature, on average, given the quality of college lectures.

It seems fairly clear that manual note taking help with learning, over using a computer, but overblown claims like this do more damage than good in convincing people to do the right thing.

6 km ([slide show](https://ntrs.nasa.gov/api/citations/20250009875/downloads/Op...) with data points and the worst slides government agencies were able to create)

Did they also shut down the bathrooms? You know, to focus the mind?

That is the worst possible take. The people launching the rocket and the people filming the launch are not actually the same people, nor do they take the same resources.

> You have a remarkably low threshold for "shit show."

I wish more people did. We certainly have an excess supply of shit shows these days.

You can post-IPO - depending on liquidity. I don't think that'll be an issue here.

And if the thesis of "it's going to look good for the first 15 days" holds, you can indeed be very profitable by e.g. buying ATM puts. (The problem being that markets don't like sticking to time tables just to accommodate your investment thesis ;)

So yes, you'll be able to take a bearish position fairly shortly after the IPO.

This is neat in theory, but unworkable in practice.

The hard line trigger almost always happens in a wider context. You often want the wider context, but it's impossible to fully define. Let's try a few examples from the article:

"Hard line: Annual raise is 0%" -> What if it comes with a much larger stock grant? Or additional paid time off? Or something else you value? Will you really just mechanistically quit?

"Hard line: A friend doesn’t repay an $X loan" -> really? What if, say, they lost their job? You want to make friendship contingent on your friends employed. There's a reason there's an adage saying "Don't loan money to a friend, you might need to decide whom you like more".

"Hard line: A government blatantly violates a constitutional amendment" -> Not even waiting for a court decision? What if it's Abraham Lincoln suspending Habeas Corpus? Or Wilson enforcing the espionage act, if that's more your political leaning? What if the government first repeals the amendment - will you be pro-slavery if it's made legal?

The world isn't a series of if-then statements. Even a hard line requires more than a single condition - at best, that condition reminds you to re-evaluate where you are.

> When you're building your business from $0 in revenue, you don't know what will work!

When you announce a post money valuation of $852 billion, you should probably be a bit better at figuring out what works, though. You're not a scrappy startup any more, even if you like cosplaying as one.

I mean, sure. But that's like equipping a sub with a screen door and claiming that in the grand scheme of things, it's a slightly different door with slightly different permeability characteristics.

Because the original post was noisy and lacked a concise summary of findings.

Or, more simply: Because folks wanted it enough to upvote it.

Pardon me if I misread, but wouldn't that be better served by a ready-made library (with, if you must AI, some futzing to account for call signature)?

What is the value add of having the AI rebuild code over and over, individually for each project using it?

If you didn't have the "random" buffers, you'd complain how slow it is. Syntax highlighting? Needs a boatload of caching to be efficient. Code search? Hey, you want a cached code index. Plugins? Gotta run your python code somewhere.

Run vi/nano/micro/joe - they're optimizing for memory to some extent. vi clocks in at under 8 MB. You're giving up a lot of "nice" things to get there.

Github's enterprise version "starts at" $21.99/seat, and requires you to "contact sales".

And I don't see any mention that that exempts you from being trained on. (Yes, the blog says you're still covered, but at that price I'd like to see a contract saying that)