2026-04-30 10:00:20
There is a persistent belief in the ‘AI’ community that large language models (LLMs) have the ability to learn and self-improve by tweaking the weights in their vector space. Although there’s scant evidence that tweaking a probability vector space is anything like the learning process in biological brains, we nevertheless get sold the idea that artificial general intelligence (AGI) is just around the corner if we do just enough tweaking.
Instead of emerging super intelligence, the most likely outcome is what is called model collapse, with a recent paper by [Hector Zenil] going over the details on why self-training/learning in LLMs and similar systems is a fool’s errand. For those who just want the brief summary with all the memes, [Metin] wrote a blog post covering the basics.
In the end an LLM as well as a diffusion model (DM) is a statistical model of input data using which a statistically likely output can be generated (inferred) based on an input query. It follows intuitively that by using said output to adjust the model with, the model will over time converge on a kind of statistical singularity rather than some ‘AI singularity’ event. This is also why these models need to be constantly trained with external, human-generated data in order to prevent such a collapse.
In the paper by [Hector] a mathematical model is created to demonstrate that an LLM, DM or similar statistical model undergoes degenerative dynamics whenever said external input is reduced. Although in the paper a mechanism is suggested to counter the entropy decay within the model, the ultimate point is that a statistical model cannot improve itself without continuous external anchoring.
The idea of LLMs being at all intelligent in any sense has been a contentious one, with the concept of language models being equated with ‘AI’ dating back to the 20th century, including as fun home computer projects. Much of the problem probably lies in humans projecting intelligent behavior onto these statistical models, turning LLMs into ‘counterfeit humans’, not helped by how closely generated text can resemble something written by a human, even if completely confabulated.
Thanks to [deshipu] for the tip.
2026-04-30 07:00:37
An often overlooked section in the datasheets for popular humidity sensors like the BME280 and DHT22 is the ‘non-condensing humidity’ bit, which puts an important constraint on which environments you can use this sensor in. This was the painful lesson that [Mellow Labs] recently had to learn when multiple of such sensors had kicked the bucket after being used in a nicely steamed-up bathroom. Fortunately, it introduced him to sensors that are rated for use in condensing humidity environments, such as the SHT40 that’s demonstrated in the video.
This particular sensor is made by Sensirion, and as we can see in the datasheet it features a built-in heater that allows it to keep working even in a condensing environment. This heater has three heating levels which are controlled via the I2C interface, though duration is limited to one second in order to prevent overheating the sensor.
Of note is that you cannot take measurements while the heater is operating, and its use obviously increases power draw significantly. This then mostly leaves when to turn on the heater as an exercise to the engineer, with [Mellow Labs] opting to start the heater when relative humidity hit 70% as a conservative choice.
In the comments to the video other options for suitable sensors were pitched, including the Bosch BME690 which is similarly rated for condensing environments. All of which condenses down to the importance of reading the datasheet for any part that you intend to use in possibly demanding environments.
2026-04-30 04:00:46
You may not know what a ADM-3, a TV910, or a H1420 are, but you probably have at least heard of a VT-100. They are all terminals from around the same time, but the DEC VT-100 is the terminal that practically everything today at least somewhat emulates. Even though a real VT-100 is rare, since it defined what have become ANSI escape sequences, most computers you’ve used in the last few decades speak some variation of the VT-100’s language. [Nikhil] wanted to see if you could use a VT-100 for real work today.
While the VT-100 wasn’t a general-purpose computer, it did have an 8080 inside. It only had about 3K of RAM, which was enough to act as a serial terminal. A USB serial port and a terminal with modern Linux, how hard could it be?
As it turns out there were a few issues. MacOS assumes terminals can take data at 9600 baud with no handshaking, apparently. It also means that any application that assumes redrawing the whole terminal is fast will be sorry for that choice.
Of course, there are commands modern VT-100-like terminals accept that the original didn’t. However, as you’ll see in the post, all of these things you can either live with or solve.
It is easy to make your own VT-100 replica. While the VT-100 may seem simple today, it was a marvel compared to even older terminals.
2026-04-30 02:30:10
This week Jonathan chats with Andrei, Mahir, and Praneeth, live on location at Texas Instruments! The team at TI has been working hard to provide really good Open Source support for Sitara processors, including upstreaming support to the mainline Linux kernel. We talk about the CI pipeline for these devices, the challenges of doing Open Source at a big company, and more. Check it out!
Did you know you can watch the live recording of the show right on our YouTube Channel? Have someone you’d like us to interview? Let us know, or have the guest contact us! Take a look at the schedule here.
Direct Download in DRM-free MP3.
If you’d rather read along, here’s the transcript for this week’s episode.
Theme music: “Newer Wave” Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0 License
2026-04-30 01:00:21
For the average rider, inner tubes have been one of the most enduring and unchanging parts of bicycle design over the decades. They’re made of rubber, they have a Schrader or Presta valve, and they generally do an okay job at cushioning the ride.
However, if you’re an above-average rider, or just obsessive about your gear, you might consider butyl rubber tubes rather old hat. Today, there are far fancier—and more expensive—options on the market if you’re looking to squeeze every drip of performance out of your bike.
Butyl rubber inner tubes have a lot of things going for them, which is why they’ve been the standard forever. Rubber holds air well, and is easy enough to repair in the event of a puncture. It’s also cheap. However, there are some ways in which the butyl inner tube holds a bicycle back. A thick rubber tube isn’t exactly light; even in a road bicycle application, a single tube can weigh 100 grams or more. They also add to the rolling resistance of a wheel and tire combination. In these regards, other materials have the potential to offer greater performance.
Latex is a material with many familiar uses, but it’s also recently become a popular alternative material for making inner tubes. It has the benefit of being very light, with a typical road bike latex tube saving 50 grams or more compared to the butyl rubber equivalent. The more flexible material also reduces rolling resistance by several watts at higher speeds, something which can make a real difference under competitive racing conditions. In a more qualitative sense, many riders also prefer the feel of riding on lighter latex tubes.
However, latex tubes also come with drawbacks. The ultra-thin, lightweight material can be susceptive to sudden failure from excessive heat, which can risk a crash in the worst cases. For this reason, the lightest latex tubes are often recommended for use on disc brake bikes only, due to the high temperatures that can be generated by rim brakes on a long descent. Latex tubes also lose air relatively quickly, and thus it’s recommended to pump up latex tubes to the required pressure ahead of every ride. They’re also difficult, but not impossible, to patch, and require some care to avoid damaging their thin walls during installation.
You might be familiar with thermoplastic polyurethane (TPU) for its use as a flexible 3D printing filament. As it turns out, it’s also a viable material for producing bicycle inner tubes. TPU tubes shave off weight and rolling resistance compared to butyl rubber, albeit not quite as much as the finest latex tubes out there. They do, however, hold air a lot better than latex, reducing the need to reset tyre pressure before each ride. Ride quality is also generally considered better than rolling on traditional butyl rubber tubes. TPU tubes also fold up incredibly small—a largely meaningless benefit in use, but really helpful if you’re trying to pack a spare or three to take on a ride.
Unfortunately, TPU tubes can be quite expensive to procure—often double the price of latex and three or four times that of a butyl rubber tube. The thinnest versions can similarly be at risk of heat failures when used with rim brakes, so it’s important to check before installation if your TPU tubes are rated for use with disc brakes only. Puncture repair can also be difficult, though there are some specialist patches on the market if you wish to attempt it.
It’s worth noting that there is another way to go, as well. It’s possible to buy wheel and tire setups that eliminate inner tubes entirely. These “tubeless” systems offer a major weight reduction, and tend to have lower rolling resistance than even the lightest, most flexible tube setups out there. They’re not really a development of tube technology, but moreso a divergence in wheel and tyre design. In any case, they are pricy, and can require some special equipment to install and maintain. To allow them to self-heal in the event of minor punctures, they’re also typically filled with sealant. In the event of more serious damage, it’s often still possible to install a tube to keep riding, but this is an incredibly messy process that will get sealant all over you.
If you’re a regular commuter cyclist, butyl rubber tubes will probably remain your go-to choice. They’re the cheapest to buy, the easiest to repair, and any benefits from lighter, more efficient tubes are largely wasted on a commute. However, if you’re an avid road cyclist looking for the best feel and efficiency, or especially if you’re getting serious about racing, then you really ought to consider leaving butyl behind for something better. Happy cycling!
2026-04-29 23:30:40
If you want to dabble in audio digital signal processing, you would probably think of grabbing a dedicated DSP chip. But thanks to [WeebLabs], you could just pick up a Pi Pico and use this full-featured DSP library.
The system supports plug-and-play USB audio interface that enumerates on Windows, Linux, macOS, and iOS. It can handle 16- or 24-bit inputs at up to 96 kHz. You can output up to four channels of 24-bit S/PDIF or I2S, or switch to an RP2350 to get eight channels. This lets you drive a DAC easily. There is also a direct output for a subwoofer that doesn’t require a DAC.
Each channel has a pre-amp, and a matrix mixer allows routing with different gains and phases for each input. An equalizer allows ten bands per channel. There are also modules to do volume leveling, loudness compensation, and headphone cross-feed.
The library uses both cores of the CPU and manages up to ten preset configurations. The Pico does get an overclock and uses a fixed-point representation. The Pico 2 (RP2350) doesn’t need overclocking and uses single-precision floating point.
Overall, this looks like a great base for any sort of soundcard-like project. We’ve seen DSP stunts on the Pico before. This might also make a nice base for other audio projects.
