They can confiscate your computers, too.

This person, of all people, would know this. He really thought he’d never get caught.

Look back as recently as last year and hardware wallets were held up as the recommended way to do self-custody all over.

Whenever I brought up the issue of losing coins due to hacks or bugs, I was told that hardware wallets had solved this problem.

The marketing department doesn't have to lift a finger in this market. They could quietly drop it for sale without a single announcement and it would be backordered for a year.

The number of people with hardware to run a model like that is very small.

It's going to be many years before hardware to run medium size local models is widely available. Running mid-size local models doesn't even register as a factor in pricing decisions right now.

It's such a small part of the market and will be for many years, until RAM and other hardware prices come down.

This project isn't meant to be vendor-quality documentation of the chipset. It's a reverse engineering project that succeeded and included enough documentation to build upon.

Comparing a freely shared reverse engineering proof of concept to vendor-provided 1st party documentation is a ridiculous high bar.

This is a cool project that was just shared hours ago. Criticizing it for not being 1st party vendor documentation quality is a really bad faith criticism.

A year ago it was different. I looked at a lot of these vibe-RE projects and a lot of them wouldn't even do what they claimed. Now if someone points a frontier model at some code and lets it consume tokens for a few days, the output is often valuable as a starting point for other projects.

The language makes sense if you've worked with radio chips or firmware.

Terms like XO (crystal oscillator) and pin mux (chip pin multiplexer, basically assigning pins to their role) are common for embedded work, not just radio work.

Usually in a project you get that language hidden in a separate implementation doc where casual readers won't get confused by it.

The plug in solar systems are small and usually mounted in ways that don’t capture as much energy as the same panels mounted on a roof. These systems aren’t going to get popular enough to make a big difference in anything.

I think it’s nice that they’re available for people to do some small scale offsets, but I think they’re more of a gateway to better solar installations. They also send people down the path of investigating battery storage add-ons, which might actually be a better long term solution for everyone.

If we can get to a point where we can buy plug-in batteries that are subsidized by requiring coordination with the grid, the power companies could get distributed, controlled battery storage to work with their demand curves. This can already happen with the bigger permanent install batteries.

Also known as a broken clock being right two times a day.

The fallacy is assuming ahead of time that they are actually right this time.

If you start assuming they’re right and also admit that their logic doesn’t support that conclusion, there isn’t even any argument left. You’re just implying that you believe one thing is true and anyone who agrees with you, regardless of the reason, is also correct.

I don’t find that persuasive in any way.

Having a universal rule is about the only way to ensure decisions aren’t being made for other reasons and that they comply with your privacy standards.

Source: I have had my genome sequence. I was not dissected.

When people use words like “chatbot” to talk about their AI use and can’t really name basic things like the model they were using, it reveals a wide chasm between the people who are deep into understanding how to use these things and those who poke at easy options every few months and then look for ways to confirm their priors instead of trying to understand how other people are getting value.

Now it’s comments like this lamenting that the amazing LLMs we’ve had since early 2026 haven’t entirely revolutionized the whole physical world and even global geopolitics yet.

> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using),

I guarantee a lot of software you use today has substantial AI development in recent releases.

Even the Linux kernel is getting a heavy input from AI and Linus is advocating for it. Unless you’ve frozen your entire software stack in 2024, you are using software with AI generated code.

> and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)

Saying nobody wants those problems solved is flat dishonest. The complaints from some mathematicians are that they and their peers didn’t get to put their names on the papers, not complaints that the problems were solved.

Some of those doors have gotten much harder to get through. The days of falling into a FAANG job by being willing to relocate to the right city and know enough to practice LeetCode for a few weeks have passed. It’s possible that the days of putting modest money into the S&P 500 and getting 8-9% year after year have also passed, but I don’t predict the future as well as I’d like.

If you have $3 million at 35 years old then all you have to do is invest it, keep your expenses manageable, and find a comfortable job that pays the bills for the rest of your career.

If you’re setting your sights on either very high net worths like $50 million or very early retirement targets like $5-10 million in your mid 30s, even high-normal IC career paths like working at FAANG aren’t going to get you there without some extreme luck. You either need to be grinding hard at TC maximized overall career paths or getting very lucky with equity at risky early stage startups.

It’s not always used when available. It can be a lot easier and safer while being fast enough to do it on the CPU. If the operating system has an image decoder path that will handle everything for you then it might be used (Apple hardware for example) but if you have to go out of your way to implement the API, common software will favor software decoding because nobody wants to maintain code testing branches for different hardware image decode platforms.

Even formats like H.264 are being dropped from some hardware decode engines because it’s so easy to do on CPUs now, as can be seen even on the Raspberry Pi 5. Many software stacks will skip hardware JPEG decode even when available because it’s extra maintenance and debugging overhead for such a little gain.

Progressive decode is a feature that is good in theory but rarely used in practice. If you’re concerning yourself with power usage and memory footprints, progressive decode goes in the opposite direction. Users also don’t like progressive decode as much as developers think they will as it’s often perceived as something being broken.

I think WebP made reasonable tradeoffs for the way images are actually used and delivered.

EQ can help a lot, but there are limits. The response curves you see are usually very smoothed and they’re only from one sample. The peaks and valleys and resonances can be narrow and can differ from unit to unit or even more from one manufacturing run to the next.

The response is also a function of the interface to your ear and leakage to the outside world. Some headphones are more sensitive than others.

I have a few high end headphones and I’ve tried a lot of the published EQs for all of them. In theory they should sound like they have the same frequency response but they still sound very different. Closer together when EQed, but you can tell they’re not the same.

I am an EQ proponent, but doing EQ on headphones doesn’t level the playing field so much that it only comes down to distortion.

The best case for EQ is in a sound controlled room for speakers when you can sit in the sweet spot and reflections and room effects are not a problem. Set up multiple sets of decent (distortion free) speakers in that environment with a moderate EQ and my listening skills aren’t good enough to identify differences most of the time.