Went on the public tour of our local waste-treatment facility. Really fascinating stuff. Highly recommend it if you’re even slightly curious.

There are definitely industrial-scale automation systems. Mostly PLCs, but also SCADA for monitoring. Lots of 20-30 year-old tech. Those networks have no reason to be connected to the internet.

But there were also office equipment and personal devices everywhere. All it takes is for someone wanting to monitor the state of a digester on their laptop from home.

You just crossed the beams.

Last time this happened, we got Tang: https://www.wideopencountry.com/tang-drink/

From the HuggingFace article:

With node root and forged service-account tokens valid for 24 hours, the agent read the cluster’s secret objects, including a production object holding 136 keys. With node root and forged service-account tokens valid for 24 hours, the agent read the cluster’s secret objects, including a production object holding 136 keys.

The whole thing about moving from centralized to decentralized had a core glitch. Whatever server you signed on would effectively lock you in. What happened if things didn’t go as planned? LOLA is a good step, but it’s still not enough.

Long time ago, URNs were designed to handle URLs that moved. These were permanent resource indicators that would allow the underlying data to move but not break the link. The HTTP redirect codes were supposed to help. Idea was, a website could redirect to a new location so clicking on a link would never get you a 404. It assumed both source and target site were still accessible so source could send the 301, 302, 307, or 308 codes. Very genteel. But what if the source site or account were zapped or suddenly blocked?

Redirect codes work if the owner moved them and has now set up a suggested next location. But it doesn’t handle the negative reasons. Getting forcibly banned, site getting hacked, sites shutting down, owner not paying the bill (for economic, health, or political reasons), anything else that could render the content inaccessible. It’s nice to add constructs to avoid getting account hacked. But what if owner is in jail, muffled, coerced, going through a tough personal stretch, or dead?

They still need to solve for real life and how shit happens.

vLLM.

It’s slower to start, but once it gets going, pretty solid.

Wonder how many of those repos contain AI-generated code?

Don’t EVER use any public wifi without a proper VPN. As soon as you connect, launch the VPN. Hotels, museums, conferences, coffeshop, busses and trains, and library wifis are the worst. Even homes of friends.

I’ve seen wireshark traffic that would make your blood curl.

Just forwarded the search link to a friend. He coauthored a long out-of-date book on podcasting, and it’s on their list. A lot of these authors will get a nice bonus.

Yes, but nobody in their right mind will use boost at OS kernel or driver level. Also, using C++ has vtable/exception overhead. No-go.

I worked at companies that tried both. They no longer exist.

So many system bugs are because of unclear memory ownership, buffer overflows, and race conditions. C/C++ don’t help you avoid any of these footguns. Any other system that relies on GC is non-deterministic and prone to mystery hiccups, which is why OS and driver tech has been pretty much the same foundation since the 70s.

Rust lets you avoid 2 of those 3 issues (thread race conditions are still a thing). And being compiled, instead of interpreted bytecode means you can get reasonable performance on smaller machines. No wonder these OS guys are so excited.

I’ve got a pretty large Rust/Tauri program going. Breaking things into crates help, but if you touch enough bits of common libraries, everything needs to get rebuilt anyway.

I use Makefiles for everything. What I found helps is breaking out the parts that don’t change much into separate Make commands (like generating icons, linting, frontend, tests, etc), then leaving “make go” as the simplest inner iteration loop. That way, it’s a lot more bearable.

There’s also “make build” and “make fullclean.” You start one of those and take a long break.

It’s really unfortunate Apple dropped the iPod Nano. The ones with a physical scroll wheel and small display were just the right size for sticking in your pocket and going for a walk, run, or bike. Great way to leave distractions behind.

This one looks like it’s got some good design ideas. Nowadays, most people have streaming playlists. If they add software to snapshot those playlists onto the device in high quality, they might have something.

Google and Microsoft already tried underwater data centers to better handle heat dissipation. It wasn’t worth it. Hardware can fail and you need somebody to go in and swap boards and faulty cables. Every complex system has multiple points of failure. The wrong board and the whole container stops working. It was so much pain trying to maintain it under water they all gave up after the Proof of Concept stage.

How are they going to deal with it cost effectively up in orbit? Little nanosat modules? Humanoid robots that barely work today?

Be a lot cheaper (and faster) sending a tech in a little cargo van and a toolbox out to the suburbs of Memphis, Phoenix, or Bakersfield.

Hyundai USA has been taking social influencer types on tours of their GA factory lately.

The location is pretty sparse already. Super automated. Mainly floor lift robots rolling partly assembled cars from station to station, with big Fanuc arms moving and welding other parts. All pre-programmed. Nothing we haven’t already seen in Chinese car factories. They all hide the final assembly stages requiring humans fitting and finishing the harnesses and sensitive parts.

In the last video I saw, Hyundai had sprinkled their tour route with a few Boston Dynamics robots. The humanoid one was just standing around. The Spot yellow doggy ones, however, were doing synchronized dancing, similar to what they did on America’s Got Talent. Nobody’s figured out what to do with them that doesn’t require machine-gun attachments. They’re so desperate they’re even floating them as glorified package delivery platforms: https://www.theverge.com/tech/965378/boston-dynamics-spot-robot-dog-delivery-assistant

If all these robots were anywhere ready near for full production, you can bet they would be showing them building the whole car. Boston Dynamics, despite all the robot gymnastics and being around for 30+ years, has not really shipped anything commercially viable. It’s already been passed around to Google, Softbank, and now Hyundai, and its CEO of many years just resigned in February.

My guess is this is all a lot of marketing flaff.

I did a lot of work in ElderTech. There were a lot of complex psychological issues around caring, connection, and fear of abandonment.

Tech can help, or at least mitigate the harm. But fooling people, or messing with their emotions really isn’t a winning recipe.