foltik

Because someone understood and verified it.

If it’s written by AI it’s gonna be the same overcomplicated overfit slop in one way or another. Performance pressure doesn’t fix that.

That's quite a creative definition of positive.

So a man “illogically and hysterically” punching a wall because he had a bad day is positive because it's biological?

If you really learned anything you should post your own thoughts, not whatever this lowercased Claudish is.

I was referring to this claim from before the Jepsen report, which they have of course now quietly removed:

> ParticleDB’s home page advertised that ParticleDB was “formally proven” with a TLA+ proof. However, during our collaboration no TLA+ specification was available.

Ok sure, so an LLM told you all the bugs are fixed and it’s faster than Redis, just like how it told you before that it was formally verified?

Please try to resist the urge of being a meat proxy.

“No.”

Afraid to show the slop?

It would be silly for a cached GET request to have to wait in line behind a bunch of POST requests blocked in serialized database transactions. A cache hit is basically free so it’s not starving any other requests.

Surely you realize most people would disagree?

There’s a _very_ good reason. Humans have a much richer mental mental model of the codebase, design decisions, external factors, etc.

Of course Humans can make errors when translating that to English. But if you’ve read any amount of Claudish it should be obvious that interpolating the code to documentation with an LLM typically produces a significantly worse result in almost every way. Except for superficially “sounding” correct.

You’re engaging in bad faith, and parroting the same false equivalence between security and user freedom that Apple uses to keep a chokehold on their software distribution monopoly.

Having some way to unlock an iPhone doesn’t at all imply it must be done in a way that reduces security, or trusts the average user with upholding it.

Use your imagination. They could require you to have an apple developer account. Or visit an apple store. Or even limit it to a special developer unlocked iPhone flavor.

You’re missing the point. Hackers (in the tinkerer sense) just want to program their own devices, with full access to the capabilities of the underlying hardware. Apple could obviously allow that if they wanted to, in a convenient way that doesn’t increase the attack surface for malware.

Uhh, what? Clearly nobody is arguing that "arbitrary remote execution is good." Apple could easily satisfy everyone by simply allowing users to unlock their own bootloader.

Consider that their point is not just about this specific use case.

Apple spends a huge amount of effort maintaining two different operating systems for what is pretty much the same machine under the hood. Don’t be naive and try to rationalize their bad behavior. They clearly just want to maintain their lucrative software distribution monopoly.

And this is of the typical quality of what you’d get.

80% incoherent narration of random numbers, all down to the 1’s place so it looks super duper precise and impressive if you squint and don’t think too hard, like the “author” seems to have done.

I use them every day, in systems programming. They produce a solution, but it’s often several orders of magnitude away from an optimal one in various dimensions. That’s still quite useful, but for some problems that doesn’t cut it.

To reach a more optimal solution, significant human steering is reqiured for every tiny decision. It’s the same as how early models needed hand holding on the basics, playing out again in the finer details. Human experts need to inject “eureka moments” from the outside.

And no feedback loop that I’ve tried has been able to conjure an understanding of these finer details that, as far as I can tell, just aren’t represented in the weights.

Modeling the basics of software engineering was easy, since things like writing code that passes some tests is easy to RL. But for anything not easily RL-able, we’ve basically ran out of human generated tokens to distill from.

Without some breakthrough in generalization, I remain unconvinced that we’ll be able to model the finer details of software engineering by looping the existing training corpus back into itself. Garbage in, garbage out.

Maybe you think there’s no reason for you to be writing code by hand anymore, but it doesn’t follow that the entire skill is obsolete.

You must not be looking very closely yourself then.

Thanks for clarifying, Claude.

Complying in a way that follows the letter of the law, while clearly contradicting the intent.

For example, finding a loophole to keep it difficult unsubscribe. Or making it easy to unsubscribe, but adding an egregious cancellation fee to all subscriptions.

Another good example is when NYC passed minimum pay laws for delivery drivers, and the big delivery companies immediately all colluded to hide the normal tip screen behind a bunch of submenus (until eventually forced to change it back).

Is it really unclear to you what the intent is here? What specifically about the rule do you think is ambiguous?

How do you think Cloudflare gets this bot traffic metric?

Making it more specific only invites malicious compliance. If businesses TRULY make a good faith effort, they’ve got nothing to worry about.

It feels like everything has endless fractal depth to it, and art is what we call it the act of compressing that down to something simple.

But you need some understanding of the underlying complexity to appreciate compression. We’ve all seen and tasted stuff, so a beautiful painting or a great pizza are easy to recognize.

Programming is tricky because the complexity is hidden. To laypeople it’s confusing, “why is a simple feature so complicated??” Even to other programmers it’s often non obvious. Someone could have spent days factoring the solution to a difficult problem down to 20 lines, but unless you've also wrestled with that particular problem, it looks like code anyone could have written. But really that’s the beauty of it. Programming is a big space to explore that’s hard to fully comprehend.

Writing comments with an LLM helps posting efficiency, but for most HN accounts, the bottleneck remains having something worth saying, not the typing.

It does during reasoning. You could even put it in a loop and force it to reflect at every step. Or spawn a bunch of review agents.

And yet you still just get slop.

Exactly. In my experience, even giving VERY specific design guidelines and aesthetic criteria, these models always produce subpar overcomplicated code (and writing). Unless excruciatingly spoonfed at every step.

Ciechanowski’s works of art are in a completely different league than this AI slop.

It’s all just surface level complexity with no intention behind it. A clumsy approximation at best.

So we should judge each piece of evidence in a vacuum?

Facebook has broken its promises about user data many times before. I don't buy anything they say about Muse, and nobody in their right mind should. Unless they’re getting paid to, of course.

Aside, if “a skill file organizing purchase history in markdown format” is the only thing you took away from that thread, you might want to take a closer look.