henning

[flagged]

> For the macOS side of things, they'll use a translation layer since Apple don't support Vulkan directly (they made their own API with Metal)

Where does it say that? Why not use MoltenVK?

They'll just do what Anthropic does: let it Ralph Wiggum a pile of broken shit, and then say "wewwwww, doing pwogwamming is vewwy hawd, UwU >_<" when it won't build and fails at basic use cases that would be easy to test automatically

I clicked comments to say the same thing. Can recommend

The Beatles released "She Loves You" much more than 24 years ago. You don't get to redistribute their music however you want. Me pointing out that AI bros disregard basic common sense and the law as part of pursuing their objectives doesn't weaken anything, it reinforces how they should be held accountable.

Everything in AI is built on copyright infringement, so redistributing Blizzard assets while slapping an MIT license on everything is par for the course.

LLM programming is very easy. First you have to prompt it to not mistakes. Then you have to tell it to go fast. Software engineering is over bro, all humans will be replaced in 6 days bro

> They can write code better than you or I can

Speak for yourself. They produce shit code and have terrible judgment. Otherwise we wouldn't need to babysit them so much.

If you think contract law can be accurately represented in a 170 LOC Markdown file, you are incredibly stupid. If you think the current frontier models do not hallucinate non-existent court cases with case IDs like 1234567, check YouTube for all the idiot lawyers and pro se defendants showing their whole ass. They are using ChatGPT with one of the current models.

Their chart only goes up to 70.

This will totally work since we have an unlimited amount of rare earth elements we can just ship off into space never to see again. Infinite raw materials + infinite power equals infinite AI!!!

Have you ever used an LLM with Zig? It will generate syntactically invalid code. Zig breaks so often and LLMs have such an eternally old knowledge cutoff that they only know old ass broken versions.

The same goes for TLA+ and all the other obscure things people think would be great to use with LLMs, and they would, if there was as much training data as there was for JavaScript and Python.

This is one of those instances where bullshit takes more effort to debunk than it does to create.

We already went over how Stack Overflow was in decline before LLMs.

SaaS is not about build vs. buy, it's about having someone else babysit it for you. Before LLMs, if you wanted shitty software for cheap, you could try hiring a cheap freelancer on Fiverr or something. Paying for LLM tokens instead of giving it to someone in a developing country doesn't really change anything. PagerDuty's value isn't that it has an API that will call someone if there's an error, you could write a proof of concept of that by hand in any web framework in a day. The point is that PagerDuty is up even if your service isn't. You're paying for maintenance and whatever SLA you negotiate.

Steve Yegge's detachment from reality is sad to watch.

Q: Why did the self-driving car cross the road?

A: It thought it saw a child on the other side.

At least they're shipping a million lines of code per month per engineer. That's what counts.

it is so annoying when Jonathan Blow says batshit unhinged shit about "the collapse of civilization" and he is kind of right.

JSX is a convenient notation for structuring HTML that is created with React and similar frameworks. You are imposing a markup hierarchy on attributes that are not inherently hierarchical while giving examples of deceptive weight loss scam slop.

In the past I would say you should be ashamed of yourself but now I don't bother.

The entry for Theranos does not mention the massive fraud that occurred. All of this bullshit is AI-generated, isn't it?

People might want to use RSS to check on your site for updates whether you write it by hand or use some kind of CMS to generate it for you.

WordPress was a technical mess before their founder had a psychotic break and their company posted features advocating for business owners to put bait-and-switch AI slop on their websites.

Stability is not that important or else research would show machines give better results than free weights, and they don't.

Yep, lots of different ways to get jacked. That means if you couldn't care less about strength, you can do pretty much any decent exercise that targets the muscle(s) you want to grow in a very wide rep range. Most people want a combination of both size and strength, so you can just do some sets of 5-10 if you aren't already. If you want to have a strong deadlift or squat or whatever, you should train that movement. Not as complicated as fitness social media people want to make it seem: train for what you want.

I like good type systems, too, but they won't save you from bugs that are better addressed by fuzz testing, fault injection testing and adversarial mindset shifts.

And as things fell apart / Nobody paid much attention

Some implementations seem vectorization-friendly like the C one that uses a bit-twiddling trick to avoid the `x = -x` line that the Odin implementation and others have.

When you put these programs into Godbolt to see what's going on with them, so much of the code is just the I/O part that it's annoying to analyze

He explicitly endorsed JD Vance in a past stream. Buying his games directly fuels authoritarianism.

Unless you feed a spec to the LLM, and it nitpicks compiled TLA+ output generated by your PlusCal input, gaslights you into saying the code you just ran and pasted the output of is invalid, then generates invalid TLA+ output in response. Which is exactly what happened when I tried coding with Gemini via formal verification.

This looks very similar to the SDL3 GPU API and other RHI libraries that have been created at first glance.

Can't imagine a worse angle for regaining trust than doubling down on AI slop.

UUIDs make enumeration attacks harder and also prevent situations where seeing a high valid ID value lets you estimate how much money a private company is earning if they charge based on the object the ID is associated with. If you can sample enough object ID values and see when the IDs were created, you could reverse engineer their ARR chart and see whether they're growing or not which many companies want to avoid.

Ah, yes. If the thing that is false is true, all kinds of interesting things happen! For example, if I became the queen of France, I could make people do silly dances! That is an interesting hypothesis that could play out in my imaginary world!

SaaS maintenance isn't about upgrading packages, it's about accountability and a point of contact when something breaks along with SLAs and contractual obligations. It isn't because building a kanban board app is hard. Someone else deals with provisioning, alerts, compliance, etc. and they are a real human who cannot hallucinate that the issue has been fixed when it hasn't. Depending on the contract and how it is breached, you can potentially take them to court and sue them to recover money lost as a result of their malpractice. None of that applies to a neural network that misreads the alert, does something completely wrong, then concludes the issue is fixed the way the latest models constantly do when I use them.