100% accurate. I can tell you put a lot of research into it. Thank you for your work.

Yeah, the “6 seconds” part is just a clickbait. The “near-fully automated” part is interesting.

If someone is able to self host then of course it’s a better solution but self hosting requires knowledge, time and money and is not for everyone. EU tech is better than US tech so if someone is not able to self host moving to European providers is the next best solution and should be encouraged.

You’re too cynical. Yes, EU is not immune to lobbying but it’s way better at regulating companies than US is. Those alternatives are clearly better now so not switching because “they may not be as good in the future” is not a valid reason to keep using US tech.

What I’m saying is that we don’t see anything beyond personal projects. I’m sure a lot of AI generated code ends up on github but do we see acceleration in public projects? The open source projects I see and use move just as fast. I don’t see explosion of useful tools or all the issues being resolved faster. Open source projects never say “we have seen increase in contribution and bug fixes”, they always say that they just see useless pull requests. If AI is so great at coding where is all the great code? Right now it’s like generating lots of gibberish text and saying that AI is great at writing books.

Nice, I’m using leptos on front end so I don’t have type issues :)

I’m trying to migrate github workflow for my Tauri app to gitlab using claude code. I have some leftover subscription from my previous job that will expire very soon. I would also like to generate leptos client binding for all the plugins but there would still be a lot of testing involved. I don’t think I will have time for that.

What tool for Tauri?

And I’m not saying no code is done with AI. I’m sure some people use it successfully. What I’m saying is that people that achieve that are the ones that would be able to do it without AI, understand its limitations and use it as just another tool. Not the ones that treat it as religion, generate bunch of code without sleeping and preach about how great it is.

I like neovim for personal projects in Rust, Lua and JS but for collaborative work in Java it’s not really usable for me. Database access, merging big PRs, unit testing tools, debugging, integrations with Spring… I always saw too many feature gaps to even try.

You are 100% right, people tend to make up crazy theories while trying to make Google more evil than it is. My main issues with Google are:

• using Youtube viewing data to push braindead clickbait content at everyone to maximize engagement. It’s been proven many time that watching Youtube radicalizes people and it’s a huge problem affecting entire generations
• Locking down platforms like Android: first paying companies to put their apps and services on phones and now locking down AOSP and side loading. It’s just a dick move to force everyone to use their platform and sell more ads on it. They are trying to do the same with web
• supporting fascists financially and selling tech to war criminals, obviously, but that’s unrelated to data

I don’t care that much for other products because I don’t use them, I have all the ads blocked and it doesn’t really affect me.

I don’t the issue with Google is that they are careless with user data. It’s extremely valuable so of course they protect it, even from their own employees. That doesn’t mean they don’t collect, analyze and monetize shitload of data on everyone. Their business model is the problem.

When Peter first got me hooked on Claude, I did not sleep. I spent two months excessively prompting the thing and wasting tokens. I ended up building and building and creating a ton of tools I did not end up using much.

So where is all this code? Where are all the great tools everyone is building? So far all I’ve seen are some bad web games, parody projects and mentions of mysterious tools for personal use. If it’s so easy to code now why we don’t see explosion of projects everyone needs? Just use AI to make Servo based browser, make PostmarketOS run flawlessly on modern phones and give me open source apps for all the shit IoT products sold everywhere.

It’s a rhetorical question. We all know why we don’t see it.

I’m working on a app now and I’m spending half of my time on thinking about the UI/UX, trying different options, selecting/designing icons, perfecting minor details in user flow. Other half is mostly trying to make the code clean and maintainable and learning new tools. AI agent would probably make a prototype in a couple of hours and then I could dump it on github and forget about it.

That being said I would like to use AI for couple of project I had in mind for a long time. Translate some github actions into gitlab CI/CD so I can migrate my projects, translate some JavaScript APIs into Rust (WebAssembly). It’s boring work and AI should be able to handle it. It would be nice if some of the vibe coders did things like that but my guess is they don’t really touch real projects people use (or don’t understand them enough to contribute anything useful, just flood them with shit quality PRs), just generate bunch of slop that no one cares about.

You know you don’t have to reinstall the OS just to change desktop environment, right?

The same as your “we”. I assume you have some group ready to go.

You should target their internet connection. Fiber cables are easier to damage and harder to protect. It’s also easier to fix so longer, organized campaign would be needed but I think it would be possible to avoid capture for longer.

Obviously he just picks the one he hates the most.

Meh. So realistic attack would be that you know someone you want to track has one of those 17 models (which is hard to tell by just looking at the headphones) and never paired it with Android and he carries them everywhere. You force-pair and now you can track them. It’s pretty silly as a random attack because why would you track a random person. It’s silly to use it to record conversations because from 15 m there are easier methods to do it. I would say the risk that this will be used to actually track/record someone is low.

Windows: for clueless people!

No, from what I’m reading they just pump money into it senselessly. VR/AR devices are $1B a year market and they pumped like $18B into it in a year. They just don’t care because they make so much money they can burn it.

You’re right, a lot of it probably went to the devs but from what I’m reading the cash burn is so bad because it’s extremely chaotic in there, lot’s of bad decisions, canceled projects, lot’s of directors swapping places. No idea what part of it went to low ranking devs and what went to consultants and building useless prototypes.

Yeah, but why ask Gemini about it’s priorities? I can just lie about it.