onion2k

5 ms·

31.
by onion2k 17d ago
Providing ready-made UI components and templates has never been a million-dollar business, nor should it be. Arguably many websites and apps desperately need a good component library because it's very clear the UI sucks without one
32.
by onion2k 17d ago
I don't think it's meant as something that's useful per se because highlighting 5m lines of code at once is unusual, and doing it in a non-deterministic way probably isn't especially helpful, but to show what a browser
33.
by onion2k 17d ago
someone of greater talent Someone of potentially greater talent, depending on the quality of the hiring process and the offer you can make. There is always a risk you end up with someone lesser or equal, especially if you value the inst
34.
by onion2k 18d ago
Didn't make your bed, no problem, just fake it. Except you need to have taken a photo of the unmade bed, uploaded it to ChatGPT, prompted for the 'fake made bed' version, and downloaded the image. Surely it's less effo
35.
by onion2k 18d ago
The Astra version seems to have used three.js r170, which is from October 2024. Sol used an even earlier version. GLM's code used the latest version, but I think it's just getting three.js@latest from jsdelivr so it's unlikel
36.
by onion2k 19d ago
That's AI getting an example of replicating the state that shows the bug which is exactly what I'm talking about. It does that far more than humans do, and it's ace. That's how you should be debugging a system - replic
37.
by onion2k 19d ago
One of the first things I tell the junior/mid-level developers I mentor is "You can't debug something just by reading the code." We all have a mental model of how our code works, and it's usually a bit wrong. Bugs
38.
by onion2k 20d ago
That probably means Fable doesn't have enough understanding of what AirBNB is in order to build a clone. If you think about it, people don't really write much about how an app like that actually works. They write about the impac
39.
by onion2k 20d ago
This is a form of unchecked privilege. He's saying he can spend time during the week reading a blog, and because of that he advocates taking a day a week away from screens. That's fine for him but someone working a manual job al
40.
by onion2k 20d ago
The products that are well established won’t get eliminated because they have adoption. An existing customer base can be a moat if there's an effect from scale, like a game where you need there to be people on a server to make it fun
41.
by onion2k 20d ago
If it's truly trivial to build a clone then people will just build their own apps to use themselves. In a couple of years regardless of how good your network it won't be possible to launch a successful SaaS app is unless it has a
42.
by onion2k 20d ago
I've always assumed that the reason there aren't many launch sites in Europe because the booster stage falls back to Earth, and generally you want that to be over the sea. People get cross if rockets fall on their houses. If you&#
43.
by onion2k 20d ago
How do you account for vandalism and theft? You don't consider the problems. That's how you can pretend it'll be low cost.
44.
by onion2k 21d ago
I'm rapidly coming around to the idea that Three.js is the wrong choice for browser-based 3D apps built with AI. Three.js is an amazing library, and if you're building by hand it's immensely useful due to the DX, but you don&
45.
by onion2k 21d ago
Geographically, yes, but that's not very important.
46.
by onion2k 21d ago
Anyone who's worked in tech in a large company will probably have experienced having an ops team who use RPA tools to do repetitive tasks that tech teams get the blame for when things break. AI will make this so much worse. Things will
47.
by onion2k 21d ago
The best non-web analogy I can think of is when someone builds a 2D platformer in Unreal Engine. The complexity of the tooling is necessary because someone else is making Witcher 4 in it. The fact the 2D platformer is using a vastly more
48.
by onion2k 21d ago
Webdev is the pinnacle of over engineering. Imagine you wanted to make a native desktop app that runs on Windows (going back to Windows 95), and OSX, and Linux, and have it use the end users choice of native components, QT, or GTK. It als
49.
by onion2k 21d ago
The value of future reports should also be a component though. By paying a low amount you discourage ethical bug bounty hackers from bothering to look for more exploits. If I think I'm only getting $1000 for a Chrome issue versus $100,
50.
by onion2k 23d ago
If you write one for likes and shares then yeah, totally dead. The same is true for all internet content though. AI has removed the need for anyone to ever visit a website again if they're happy to read what AI generates. If you write
51.
by onion2k 24d ago
I wonder if this says anything about the capabilities of AI. If it's better to bundle an entire headless office suite than ask your AI to build support for relatively well documented formats (or even to extract the file support parts i
52.
by onion2k 24d ago
This is for science. You can't put a price on knowledge.
53.
by onion2k 24d ago
I can't help thinking ranking them by eating them would be a much, much better way.
54.
by onion2k 24d ago
many developers have not learned git in depth They have less to walk away from then.
55.
by onion2k 24d ago
I have no reason to suspect this is from an LLM (I don't have much confidence in my ability to spot anything other than the most obvious examples) but I also don't see any reason to assume that someone couldn't just say some
56.
by onion2k 24d ago
I have seen this even on channels that are not faceless. The AI logo doesn't mean "entirely made with AI". It means "significant use of AI detected". If someone uses AI to modify the background of their video (say
57.
by onion2k 24d ago
Porting something to a language you don't know doesn't seem very helpful to me. You've locked yourself out of doing useful work except with continued application of more AI. Without the ability to verify it, except with even
58.
by onion2k 25d ago
LLMs are non-deterministic, by design. They don't have to be though. You can give them a temperature of zero so they pick the highest-probability token every time to give you a deterministic output. I imagine this doesn't work o
59.
by onion2k 25d ago
Not sure about that. The nature of work changes if you have some[one|thing] that can work on it continuously forever. The goal of using AI shouldn't be to do the things a person does; it should be to do the things a person would be abl
60.
by onion2k 25d ago
agent's failures on long horizon tasks We've moved from LLMs being able to work on a task for about 2 minutes to about 2 hours in the last 18 months, and that's mostly limited by the context window size filling up. In 14 ye

More ›