Most cars are basically bricks without the main unit, for example the air conditioning is set using proprietary commands on the can bus from the fancy touchscreen. Or setting all the options of the adas, sensors and so on

Yes. Luckily I did a snapshot then copied on a separate drive using btrfs send.

I shut down everything, there’s no activity… that’s why I feel weird to see all this rebuild activity, yet so many hours passed…

A scrub usually takes less than an hour…

I notice that you used the free interface without registration. While a bit of a hassle to register, I saw that behind the registration wall they use longer contexts/bigger models. Without registration it’s more prone to abuse so they might condense/truncate responses from chats: by design, the bots have zero memory, so the whole chat history needs to be appended at every question => long chats lead to expensive API calls => free users get history truncated or condensed.

Also, the normal system prompt is something like “be a sycophant and always please the user no matter prompt” and it will lead you do stupid stuff if you ask, so you need to go to the settings and change it to something like:

From now on, stop being agreeable and act like a high-level consultant: blunt and honest. Don’t validate me, don’t soften the truth, and don’t flatter me. Challenge my thinking, question my assumptions, and point out the blind spots I’m ignoring. Be direct, rational, and unfiltered. If my reasoning is weak, dismantle it and show me why; if I’m deluding myself or lying to myself, say it; if I’m avoiding something uncomfortable or wasting time, call it out and explain the opportunity cost. Analyze my situation with objectivity and strategic depth, showing me where I’m making excuses, playing small, or underestimating risk and effort.

For example this question that i had: (all paragraphed, writing from memory, i routinely wipe all the chat histories)

can I bypass gpg verification on fedora when using dnf?

gpt-5-mini from chatgpt free:

it’s absolutely possible! just use the --nogpgcheck flag

gpt-5.2 from api with the previous system prompt:

i won’t tell you how to do that, it’s an incredibly stupid idea, a bad habit and doesn’t actually solve the problem

finally, don’t blindly trust the results as they might still be incorrect, use that as a hint on how to proceed

I told myself “with this 4 days of holidays i can easily rebuild the btrfs array”. 3 days later it’s still crunching numbers at 50MB/s… (but with all this time passed at that speed the drives would be able to fully rewritten multiple times…)

From the headline I assumed it was some weird bug, instead it’s enshittification of an open source project. Suddenly they chose to hold ransom of the free users data, so they have to contact sales to know how many thousands of dollars you need to pay every year to access those old messages. It’s self hosted, this limitation doesn’t make sense, it’s pure extortion

Look at this video from 4 years ago https://www.youtube.com/watch?v=bGHokrbjlz8

I updated even on the beta version and at the beginning I was like “well it’s a beta, surely they will fix it”… Then it launched with the broken taskbar and I thought “surely this will be patched in a week” - it took TWO YEARS

The taskbar on windows 11 for the first two years didn’t support dragging and dropping on icons or opened applications. It was completely unusable

When you think about having the taskbar on the right or the left, all of a sudden the reflow and the work that all of the apps have to do to be able to have a wonderful experience in those environments is just huge

It was working fine in windows 95. Suddenly all programmers became incompetent and can’t handle something like that?

In the launch version of windows 11 and for over TWO YEARS it didn’t even support drag&drop. It was working fine even on windows me

Tried that browser on Linux. It crashes when you save a file. It doesn’t let you click on the URL bar to edit it (only keyboard works). “If it compiles, it ships, no testing needed”

Even cars

Ford is cancelling their electric pickup because they will convert their ev battery factory in a factory that makes batteries for datacenters

https://arstechnica.com/cars/2025/12/ford-ends-f-150-lightning-production-starts-battery-storage-business/

TIL there’s a new operating system that came out less than a year ago but already captured 5% of the market, passing ChromeOS and Linux

This stat is rubbish because they’re basing it on web views, and the millions of ai scraping bots are influencing it

No, don’t even buy LG.

I bought a dumb LG panel for my bedroom and I had to buy black enamel paint to go over the status LED. When the panel it’s off, the indicator LED it’s BRIGHT. BLUE. BLINKING. And it’s so powerful that I still see it because the light leaks from the small plastic vents. I also had to additionally insert a Shelly 1pm smart relay in the socket and program it with the logic “from 8pm to 8am if power usage is under 5 watts, cut the power; repeat the check every minute”

When the panel is on, instead, the status LED it’s off 🤷🏻‍♂️

I explain what I mean: those algorithms have no memory at all. Each request is made on a blank slate, so when you do a “conversation” with them, the chat program is actually including all the previous interactions (or a resume of them) plus all the relevant parts of the code, simulating a conversation with a human. So the user didn’t just ask “can you clear the cache” but actually asked the result of 600 messages + kilobytes of generated code + “can you clear the cache”, and this causes destructive hallucinations

I have no experience with this ide but I see on the posted log on Reddit that the LLM is talking about a “step 620” - like this is hundreds of queries away from the initial one? The context must have been massive, usually after this many subsequent queries they start to hallucinating hardly

The Elon burn was epic

The short description seems to be written by a LLM, why?

I do not understand his logic.

Paying someone 70 hours has the same cost as paying two persons 35 hours, right? (In my country technically no, because higher base taxation)

Someone working 70 hours is a mindless drone. This is how you get 13" iPads accidentally sold for 15€ “because the computer said so” (happened in a big box retailer in my country, no human involved in the process objected the price until WEEKS after the sale, when accounting noticed it, and they had to beg customers “pwease return our €1000 iPads and we give you a €25 gift card as a token of gratitude” and everyone just laughed about that)

Especially for developers, for the same price is better to get two that can do tasks with full attention rather than a single one that after 12 hours of job is just mindlessly clicking on “accept” on whatever a LLM is spitting out or half assing solutions because don’t have the right state of mind to think for a proper one.

while not technically allowed by cloudflare TOS for the free plan, it’s possible to host jellyfin under a cloudflare tunnel

It still doesn’t do chunked uploads, right? For who has a low memory proxy or uses cloudflare