IshKebab

It's definitely amazing being able to just make things with so little effort. I'm finishing so many unfinished side projects.

But I would prefer to have a job.

Not really. Even with left-pad there was still some skilled work required. At this point my wife could vibe code an app or a website and it would be actually quite good - not the slop of 6 months ago.

It's really not like that. Sorry to disappoint you but it's more like cars were invented, and someone says "have you tried the Ford Model T, it's much better than a steam carriage".

Or if you prefer "have you tried the new iPhone, it's much better than the HP iPAQ".

> Are you of the view that LLMs have more potential than the internet?

Yes, it's not even close. There are lots of entire fields that will be wiped out by AI. Financial advisors for example. Good luck if you're a translator or a proof reader.

The world is slow so I expect it to take some time, depending on the field.

Today I started vibe coding a replacement for a piece of software whose licensing cost is on the order of $100k/year/seat. It's going scarily well. I'm scared.

> Sure, math does require people in future, but it can't accommodate as many as it does today.

I actually think it probably can - after all mathematics isn't a commercially viable career and pretty much all mathematicians are scholars paid by universities and taxes. They're already fairly unnecessary and mostly kept around for interest and to have an educated population and so on. In a way they're uniquely positioned to survive the AI storm.

Not so for programmers who are almost all working for companies who would love to stop paying them given a chance.

I am glad. But don't tell me that era isn't over or that I can't be sad about it.

> The average age of the world's greatest civilizations has been 200 years.

A crane can't drive itself. AI is fundamentally different to all previous tools.

I don't know how that isn't obvious to all the "AI is just a tool" people.

You really don't need to be that large for it to make sense. A lot of these products are very rarely updated. Jira hasn't improved at all in the last 10 years as far as I can tell.

Maybe a 5 person company wouldn't do it. A 100 person one though? No brainer. And it only takes one to open source their solution.

Indeed, but only a tiny handful of companies have been successful with that business model, and they mostly make margin on things you'd have to pay for anyway like cloud services. No company is going to pay e.g. $15/user/month for Jira when they can get the same thing for free - better even. And yes it isn't an exaggeration to say you could vibe code something better than Jira. (Especially because that's a low bar!).

I'm kind of surprised nobody has done that yet actually. I guess Opus 5.5 has only been out for a very short time...

They did but IMO they had very low standards. I have very high standards and Opus 5.5 pretty much meets them (at least for Rust).

"People said it before and it wasn't true then" isn't really an argument about it being true now. There are plenty of things that happened where the zealots jumped the gun with the hype. Bitcoin for example. "People said bitcoin was going to succeed in 2005, 2006, 2007...". Or even the web. It was mega hyped in the dot com boom and then crashed and of course loaaads of stupid people thought that was the end of it.

I've only met one constant bullshitter in my life and unfortunately I worked with him so couldn't just leave. The thing that blew my mind was that nobody else seemed to care, at all. I get that it takes a significant amount of social capital to call out a bullshitter, especially at work... but still.. you don't have to believe them. Very weird.

Opus 5.5 + Claude Code (or probably any agent harness; I don't think they make that much difference). I wouldn't say software engineering is done quite yet but yeah it's definitely never going to be the same.

Someone at work vibe coded a replacement for a entire third party SaaS that we were using in a week. Their version is better.

It wasn't one-shotted of course but I don't think there's much difference in the outcome between one-shot and like 100-shot over a week. That company is still doomed.

Think of something like Trello. Who is going to pay for that when you can easily vibe code something superior (yes you really can now).

With Opus 5.5 you can vibe code a whole app without reading the code at all. Obviously you still have to tell it what to make.

With earlier models that doesn't really work. You really have to read the code and watch it like a hawk or it will make a big mess.

To put it another way, with lesser models you have to guide them at a code level. With Opus 5.5 you only have to guide it at an architectural / feature level.

> before falling into autocratic rule

Did that mean it was no longer a great civilization?

Also I would argue that a civilization that only lasts 100 years is not great almost by definition. Where's your data?

> The average age of the world's greatest civilizations has been 200 years.

Sounds like total bullshit. I briefly googled it and apart from AI also telling me that it was total bullshit, the first list I found split the Roman empire in two so it would neatly match the hypothesis!

Someone just thought of that and then found data to fit their neat idea.

Claude absolutely subtracts if you ask it to. I just regularly ask it to look for simplifications and YAGNIs and give me a ranked list of them, some of which I get it to do.

Why not? Code review is a crazy time sync. If I'm working on a project without code review (e.g. a personal tool) using Claude I can easily churn out 100 commits per day. And it isn't even slop these days.

It does make me wonder what on earth they are working on. Feels like most of their products could be basically finished already.

On the other hand Spotify somehow employs 3000 programmers, and that's just a music player.

They pretty much fixed that. Plus you can influence the writing style to some extent. I tell it to use full sentences and avoid assuming I already understand the code.

Another thing is that it's very good at creating diagrams (I think someone wrote a skill for it so maybe not by default). I usually find the diagrams are even more valuable than the text.

Honestly I was skeptical but you really can just point Claude at some code and ask it to make a markdown + SVG explainer and it will do a shockingly good job.

I'd still rather read human documentation, but even before AI almost nobody writes any - especially for internal code. And they certainly don't make great diagrams.

Ha fair point - probably not the best example! AI-proof too.

We still have horses. I wouldn't recommend a career as a farrier.

> Let me be careful about the plausibility: is it suspicious that such an obvious thing wasn’t discovered in 25 years?

Alright Claude no need to rub it in.

There's no way their slowness is going to make a difference in London. I would have thought the real problem is that they'll just get completely stuck at places where you have to be a bit aggressive to make any progress at all. They don't really have that in America.

That and cyclists/pedestrians can be kind of anywhere at any time. Again not a quite the same in America.

We'll see I guess.

Air is basically nitrogen. It can probably use both.

I would put good money on the flying demo one being a custom lightweight version.

To be honest it's starting to sound like Google were right and adding JPEG XL support was a mistake.

Ah interesting. It's also comparable in price though!

Uhm yeah probably. Are you suggesting that LLMs have been stagnant despite investment?

I don't understand your point. They're getting better so quickly it's hard to keep up.

Unreal Tournament also had a software renderer IIRC, with an interesting dithering effect.