johnfn

I don't really get the Bun thing. Bun is running Claude Code which is probably the single most actively used development app there is. You say this was a bad use of LLMs, but it's been in production for a while and I haven't heard of any evidence that Claude Code has increased a significantly larger quantity of errors, segfaults, etc, than before.

That also isn't saying "don't say this is an LLM". If the guidelines didn't want us to say someone is an LLM, it would explicitly say "don't say someone is an LLM". It wouldn't hint at it in an indirect way.

FWIW I do find it somewhat useful to have someone point out "this is an LLM" - I don't always have my AI detector on and I appreciate it when other people do. And when someone says particular text was LLM generated I need to go back and think about whether I believe it a bit harder.

I don’t see anything in there about not commenting on whether the author is using an LLM?

The thing that annoys me most about this obviously AI-generated article is: "The quotes in orange boxes are real and checkable." No they aren't - you AI generated them, just like you generated the rest of the page.

You mean the obviously AI generated README? Did you even read it yourself?

To be blunt I can't take this product seriously when they don't even run benchmarks. Your prompts make Claude better? Cool: prove it. Methods to evaluate LLM performance exist, they're called evals/benchmarks, and every company that is serious about AI runs them when they release a new version. (Of course benchmarks have their own issues, but squabbling over which benchmark is best and what issues there are is step 2 in being a Serious AI Company and step 1 is running them at all!) The fact that the only proof they have that 6 is better than five is a hacky table in a screenshot from Fable is, honestly, concerning.

And Jest was itself a huge step up from what came before (Jasmine, Mocha...)

It's a linter, a code formatter, a tester, and a bundler. What exists in your "boring" stack that's more boring than that?

It's disclosed in the ToS like it's disclosed in the ToS of basically every service that exists today.

e.g. "Claude Code connects from users’ machines to Anthropic to log operational metrics such as latency, reliability, and usage patterns. "

This entire thread has lost its collective mind. Tracking time zone has to be up there with IP address and referer in the list of "trivial things every company in the entire world collects about you", and these things are entirely trackable without your consent or even knowledge. You're gonna have to turn off the Internet.

Do you honestly think that no service out there collects basic analytics?

> Why would they brag about something like this? It's like they know people want to use models to perform cybersecurity tasks yet knowingly deny them the ability.

What exactly do you want Anthropic to say here? "This model, the one we are about to give to the entire world for cheap, is really good at hacking"? Saying Sonnet is terrible at cybersecurity is the most reasonable thing they can say, out of a lot of bad options.

That's just one benchmark, though. Tab to the next one and Sonnet 5 performs better as effort goes up just as you'd expect. I imagine the suggestion is that performance vs effort tradeoff is task dependent.

You completely nailed it.

It is better to live in the United States in poverty today than it is to live in virtually any nation at almost any point in history. The weathliest king in the 1500s didn't have access to vaccines, clean water, anesthesia... all things easy to find even in poverty today. Crime is lower, healthcare is better, nutrition is better, diseases are fewer and further between. Project that out and I would absolutely prefer to live in poverty in your "AI dystopia" than living like a normal person today.

Honestly your argument sounds detached from reality.

What exactly do you imagine AI doing in this future? Making more money out of thin air somehow? How is it “enriching the lives of a select few” but also somehow in your future I simultaneously provide no economic value, where is that enrichment coming from exactly?

Your view of a post-AI world admits both a fantastical, amazing AI that can do work so well it puts all knowledge workers out of a job for a century, but also a crippled and limited AI that is incapable of solving even our most basic needs.

It can't be both ways. It will either be bad enough that most of us will still have jobs, or good enough that most of our needs will be solved.

To unpack your point a little bit: what does it mean for "me" to be obsolete? Let's take that at face value and engage with it. I consider myself to be a decent SWE, maybe top 5% or so. There are somewhere around 40M SWEs in the world. If I am obsolete, that means AI is doing the work of ~38M people. I don't exactly understand what you mean by obsolete. In this future where AI is doing the work of 38M people, is anyone watching over the AI? How many people? Does it scale with the 40M figure, and if so, does it scale roughly O(n) or O(1)?

I think you basically have to accept there will be one of two options -- either it scales at something like O(n), in which case you have some number of highly skilled professional which can use AI to do the work of 100x of a SWE today (that would be a pretty cool job). Or it scales with something like O(1). I think in the second world the only logical conclusion is that AI is so incredibly capable that the world is profoundly and utterly transformed. If AI can do the work of any SWE without oversight, it can basically do anything -- which would mean that any need you have could be satisfied. If you are hungry, food could be grown. If you are thirsty, water could be procured. Or you would need to propose some crazy future where AGI is achieved but for some reason trivial things like this can't be solved.

I think the thing that makes me the most nervous is some enemy actor going rogue with AI and killing us all, but somehow we've lived for decades with enemy actors having access to nuclear and bioweapons and somehow we are all still here.

The worst people on the planet already have access to guns, bioweapons and refined uranium. Somehow I am still alive and mostly fine.

When did I say anything like “it will free me up from so much labor!”? This is not even remotely close to what I believe. You are responding to a caricature of me you have in your head.

It is astounding to me that people are not excited for the future. I live in a moment in time where I can ask AI to do something and it can think about it rationally, work on it for an hour, and basically do what I want. This is incredible and amazing.

> Simple example: Node is the only serious OSS software I know of that has no way to document its config (in the config file itself). It's moronic! The Node people just adopted JSON without a thought, and then refused to consider any alternatives (even "JSON with comments").

Tangential but this also drives me absolutely nuts. If I have to see `"//": "some comment"` one more time I'm gonna lose it.

You might remember the io.js fork of Node.js back in 2014. Node was stagnating, a bunch of people forked it into io.js, which eventually got merged back into Node and got it back on track. Or, going further back, CoffeeScript, a "fork" of JS that had its best ideas adopted back into ES5.

A small scrappy team can prove out a good idea because failure is not a catastrophic risk to them. In short, forks are part of a healthy ecosystem.

Your response doesn't do a lot to convince me that "debugger" is anything more than a dev tool you like. It is not the only way to see what a program is doing while it's doing it -- there are plenty of other effective ways. I have encountered many bugs where a debugger wouldn't have been a help at all.

To be totally honest, having such an absolutist view makes you look inexperienced -- and accusing people who disagree of "malpractice" is absurd.

It is a little crazy to accuse people not using the dev tools you like of malpractice.

I feel like on HN there is an endless cycle:

- Vibes are too subjective, I want an actual A/B test!

- An A/B test is too limited, I want a benchmark! (You are here.)

- Those benchmarks never seem to be reliable, I just go on vibes.

The blog post is entirely AI generated.

I remember, when I was younger, printing out MapQuest bike instructions, following them, and getting lost trying to find the next street -- and then feeling incredible when I finally found it. When GPS became so commonplace you could use it on any phone, I realized that the feeling of suddenly realizing where you were was gone forever. Of course you can only feel euphoric about finding a street if you felt at least equally despondent that you couldn't find it, so it's probably a worthwhile tradeoff. I feel somewhat similarly about Spotify. Yes, some things are lost, but you can't honestly tell me you think it was better before, right?

You seem to be misinterpreting my comment, which can be summarized as "be kind to others and assume best intent".

This is a sharply negative interpretation of behavior of people who may be acting genuinely, if without social grace. I think few people shocked that you don't know bash are displaying that surprise as a way to keep you in the out-group - I think they are surprised.

I would argue that the real in-group/out-group behavior is excluding people who aren't naturally adept at being social.