To be fair, there are still many jailbreaks for instruction tuned models. Surely less than before, but it could be because models got smarter (Claude often "reasons" that something is fishy even if it appears in the prompt or appears like a system prompt).

I remember those days and I'm glad to see another one who's been in the industry since the very beginning :) I was quite sad when davinci-003 was taken away; still remember the green-yellow-orange-red highlights the OpenAI console would make for log probabilities of tokens and all the cool things we could do with a pretrained model.

In my courses, I teach students about that era by using gemma 4 (pretrained) and showing how you can emulate a conversation even on something that wasn't post-trained for chat.

Yup, Bun is at the mercy of Anthropic, and we know the extents they go to protect their competitive advantage.

"we know the extents they go to protect their competitive advantage"

I don't. What do you mean?

Use bun when it's drop in replacement of npm, never use Bun API itself.

Like any other late stage capitalism corporation, curve must go up exponentially, otherwise shareholders will be depressed.

Kinda feels like a rug pull, similar to Bun.

Not even close as Bun is still being actively developed, albeit with less fervor.

speak for yourself, big release literally just yesterday.

Also "those who can't accommodate the storage" is funny.

What's that? You didn't pay Apple's 1200% markup on storage, just so you can have enough room for your actual work after the OS fills your disk with a bunch of bloat? What are you, poor?

The new one just works. Do try it.

>do try it

Thank you, I might. I never have, the launch was such a mess, I've never looked back and fastidiously turn it all off.

What's changed, for you to say it now works? And what do you successfully and regularly do with it?

> What's changed, for you to say it now works?

There's a hint in the submission you're replying under. Much of it runs on a 20gb local model now. Completely new system.

It damn well better, for using 14gb+ of storage.

Oh, things are about to get worse with the new macOS "privacy/security" measures. They are going to curb agentic workflows even more. I don't know how Apple just finds new ways to annoy developers, but we're in a minority after all. Of 200 million Mac users, probably just up to 1 million are developers, and the rest are normies who can't tell when they should authorize or cancel the pop-up.

I'd argue that a lot of developers can't determine if an LLM generated command is actually safe or not.

It’s not about privacy, it’s about kneecapping competitors, just like when they blocked the advertising ID but exempted themselves from this because “Apple is not a third-party, we’re a second-party”.

Apple is an advertising company and thus inherently untrustworthy.

Ah, if only that meant that there'll be less slop in the macOS code itself..

Apple hates developers

They love the ones that buy Apple hardware to develop apps for iDevices, pay the dev subscription and store fees for apps, or simply because they wanted a shiny UNIX and don't consider BSD/Linux OEMs worth their money.

By the same reasoning, why should I care about the output of someone else’s clanker if I can just get it from my own clanker?

But to answer your question, I do still care about software being maintained by someone who can make design decisions instead of just yielding control to bots who tend to produce mediocre designs.

These are all good questions.

Perhaps we can think of projects like these the same way we’d think of a good tweet - not valuable in itself but valuable because of the ideas or conversations it generates.

Maybe because they know more about the domain than you do, and it will take you a lot more back-and-forth with your clanker than it would for them with their clanker.

And maybe they don't, and that's the problem with one-day-old projects: we know for sure that the programmer didn't have time to learn anything from that particular project and we now have to trust them that they know what they are doing. Unfortunately this is wrong in 90+% of the time, so that's kind of a gamble to expect one particular project to tick this box.

This article assumes that this problem did not exist before AI. Especially in large companies like Google, the number of people who actually knew what they were doing was relatively small. The vast majority were just piggybacking on other people's work, which I absolutely hate. At least AI has made this obvious, and the difference between people is now their taste, which, for the majority of people, is bad news.

This! I work for a company that has several systems that are older than 20 years and no one really knows how they work. We are using AI to actually get insights in how they really works, to be able to rewrite and modernize the applications.

I can't read the source code since it's 8 million lines of code and written in a programming language I don't know and in a language I don't speak.

> I can't read the source code since it's 8 million lines of code and written in a programming language I don't know and in a language I don't speak.

^ This is the real world. Some comments talked about how maintainability is king and you just can't keep a mental model together of what the LLM produced. In real life software there is no single person with a mental model of how the system works end to end. In the most ideal scenario you have an architecture diagram, some readme's, and a runbook of how to use the system or get it running in a dev environment. Everything else is manually tracing through mountains of code of wildly varying quality.

coding harnesses are god sent tools when it comes to analysis and maintainability of existing code bases (including code they have produced).

Surprisingly, increasing user hostility is a filter of users' tech savviness and IQ. I doubt most HN dwellers will tolerate price increases or ads or anything like that. They can pirate shit, and they can block ads. They're better off compared to the rest of the population who is not tech-savvy enough or smart enough to bypass these user-hostilities. These companies basically extract revenue out of people who are relatively poor.

This seems backwards. The salaries around here are relatively high, so a marginal increase in monthly entertainment costs wouldn't be noticeable in the first place. People can get self-righteous if they want, but these are entertainment services, not essential goods.

Only speaking for myself, but these streaming services would need to increase costs A LOT to make me invest valuable time replacing them.

The "relatively high" salaries also mean that 20 minutes of ads in a 60 minute session "costs" more. It's not quite as simple as applying my hourly rate to the ads, but it's at least a decent place to start the analysis, and by that metric it doesn't take long before the ads are the dominant cost of the service. It gives me a lot of virtual money to deploy to the task of turning to entertainment that doesn't annoying vast quantities of my time to squeeze out fractions of a penny from me because they can do it in bulk.

"here"

It's a big world out there, whatever "here" means.

That last sentence makes me wonder, do you enjoy ads?

I went to a Super Bowl party a while back and the half time show glitched continuously. Random weird glitches for the rest of the time. => note to self: don't pay for this.

I could be wealthy beyond my wildest dreams and still be turned off by their obnoxious changes.

Someone should spend money on a marketing campaign teaching people how to be tech literate (while skirting the law, and making implications).

Someone like Nathan Fielder would have the balls for this.