pbmonster

> it was originally alternate firmware for the Linksys WRT54G router from 25 years ago

There's a couple of fun examples like that. xda-developers is named after the O2 xda, a smartphone from 25 years ago that not many people ended up developing software for.

> we'll ditch phones altogether in 10 years when smart glasses will replace them instead.

Billions are spend right now to make sure the glasses also run Android or iOS. So far, Google, Samsung, Magic Leap, RealWear and Vuzix are working with/on Android XR, and obliviously Apple is working on AR/VR iOS.

Meta and a couple of smaller startups are doing something in-house, but I don't give them much chances to get an ecosystem going.

There's much to criticize about the dev, but there's really no way to make it significantly easier. Most robovac companies just don't want you to flash the firmware.

The dev opposes selling the connector PCBs, but people have always ignored that and sold them online. They're not hard to find, but having the PCB is really only the first step.

Reminds me of a fix I wrote a decade ago. My Laptop would sometimes start emitting a high frequency whine when on battery. I figured out it only happened when the CPU went into performance states lower than P2 for power saving.

So I wrote a bash script that auto-started on battery mode and then calculated a hash every few seconds. Boom, whine solved. Terrible fix, but I never measured how much battery it cost me, so it was... fine.

Not really. Can't really map a touchpad as an analog stick. Can't map a gyro as a joystick. Can't take XBox rumble and map it to the Steam Controllers haptic feedback.

Can you hack it together in theory, get something working but making sacrifices left and right? Sure. But why would Valve want to do that? Use experience would be incredibly bad.

But give it time. There will be a standalone driver for Windows eventually, either from Valve or from the community.

> Many hardware platforms that have TPM, have it connected via a low-bandwidth LPC bus which would have nowhere near enough bandwidth for demand decryption/encryption of memory pages.

Ah, of course. I was more thinking along the lines of "CPU loads the key for decrypting RAM directly from the TMP into registers, and reloads it from there after waking from suspend or after a task switch has refilled those registers".

> You have to store the encryption key in CPU registers and ensure it's not saved to RAM during task switching or power suspend operations.

Interesting insight. Any reason why the key can't be kept exclusively in the secure enclave / trusted platform module / crypto coprocessor?

This might make sense once SteamBox is a large percentage of Linux Desktop installs.

Right now, it would be insanity. Just the situation with running SteamOS on hardware with NVIDIA cards would be a showstopper. And a whole lot of consumer PCs ship with NVIDIA chips...

The idea is deleting the friction brakes entirely, saving costs, maintenance and weight.

And 400kW isn't really all that much for a sports car. I remember 911 ads from the '80s that boasted "brakes with more than 1000 horse power".

Very funny idea. That basically means a carbureted gas engine, or a direct injection diesel with a mechanical governor and mechanically timed injection pumps - can't run a direct injection gas engine without a digital engine control unit, because the injection timings are much to precise to do mechanically.

So, basically '60s Formula 1. Might be fun to watch. We'd certainly see some crazy engine designs and a lot of re-fueling pit stops...

There's an efficiency sweet spot where hardware that people have anyway gets a higher percentage of load.

MacBooks have a lot of memory and a lot of FLOPs. They mostly sit unused all day. Yes, the excess energy use will be higher than a GPU in a datacenter doing the same work, but you have to generate an absurd amount of tokens before the dollar-efficiency catches up with the MacBook.

> why not build panels anywhere?

Because there's better alternatives same places. Norway has legendary well-suited topography for hydro power (>90% of electricity), and it's reliably windy (>8% of electricity). It's also so far north that the sun doesn't shine very much for half the year, and it's notoriously cloudy.

So yes, it will probably never make much sense to build a lot of solar panels in Norway. Same for Greenland, Iceland (substitute geothermal), and probably some parts of Canada, Alaska and Southern Argentina.

But also, yes, there's almost nobody living in those places. They're not terribly relevant in the grand scheme of things. Probably significantly less than 50M people in total.

Of course. My agentic coding containers can only access the internet through a proxy, and I use whitelists to limit from where they can send/receive data. It's annoying in the beginning as the whitelist grows, but in the end really useful information for the agent usually comes from a very limited amount of domains.

> The defining element of a LAN party, however, is the social factor. Playing a game with others in the same physical space, sometimes huddled side-by-side on the same large table, is an intimate affair. Cries of joy and frustration fill the air.

That's really the major thing that made LAN parties so special. Being in the same room is so, so different than online gaming together and hanging out on discord.

It also forces you to compromise when choosing games and maps, because you're stuck together for the night. You can't just sit out a game/map if you don't like it, you can't just hop on a different voice channel and play another game with other people. You end up playing games you're not as familiar with (and not as dominant in), so your friends will play your game with you later. This brings you into situations, where the person next to you frantically gives you the crash course in rush build orders while building their own base, making the payoff so much better when it actually works out.

I mean, does it really matter?

The Prius (and all subsequent Synergy Drive cars) were widely known - from the very beginning - to be extremely fuel efficient ICE cars. As time went on, they universally became known to be both fuel efficient and also absolutely bomb proof.

Both of those things surprised basically no-one, since the direct successors (Camry, Corolla, 4Runner, Tacoma, Hilux) were also already known for being fuel efficient and reliable cars.

The only people who really care about why and how exactly they got so fuel efficient and reliable are engineering nerds - and many of those already knew, the planetary gear set + atkinson cycle engine are a pretty legendary design. They hit it out of the park on the first try, after all.

And as this video shows, explaining the why and how to non-engineering nerds takes a good part of an hour anyway. How do you do marketing with that?

No, it's already 130M (out of the ~450M in the EU). Wero's predecessor (iDeal, Bizum, Paylib, Giropay) systems are already widely adapted in their countries of origin and will be fully compatible with it, and Wero itself has had a bit of time to pick up new users on its own by now.

I wonder what the hell the problem with naming things is in the AI space. They pretty much all suck at it.

OpenAI came up with GPT 4o, o4, 4.5, 4.1 (which came out later than 4.5 and had a completely different purpose), Microsoft just calls everything a copilot (Github Copilot, Azure Copilot, Microsoft Copilot - all from the same company, completely different things), and Google apparently just picks random words from the comments.

Building codes on that are going to vary locally. They are doable in most of Switzerland if you plan for it: no sky lights, no chimney (there's setbacks and fire codes for those), ect.

There's a third way: in-roof or roof integrated photovoltaics. Normal panels, but integrated into the roof. Those look amazing. Very popular in Switzerland where some villages have strict aesthetic rules for buildings.

Next best thing aesthetically are full-roof racks, where one face of the roof is 100% covered in panels. Nowadays you just have to select the right panel and you can make it tile the plane perfectly.

That's not the argument. Yes, curl is insanely hardened. But still, they currently have a new CVE every couple of weeks. Mythos didn't accelerate this much, no more than all the other AI-assisted security analysis they've been doing anyway.

Which either means that, tragically for Mythos, it only got to analyze the code base just after ALL the bugs where finally ironed out and now curl is bug free forever after - or Mythos isn't really all that good, dozens/hundreds more bugs remain and will be found in the next months and years.

I just think the former is a bit unlikely.

curl had ~15 CVEs in 2026 so far. You surely don't think those (and the one Mythos found) were the last security bugs still left in the code base? There certainly will be more, in fact Daniel predicts ~50 CVEs for the entire year.

But Mythos found 1. After all that hype. 1.

> This is the way. I use printed maps when on vacation.

Come on, then you know their limits. Get to your destination city? Better hope your road atlas comes with a city map, and the road you're going to is large enough to have its name printed. Otherwise, you're shopping at the next gas station. Have the city map? Better hope the outer borrows are on it, too. Found your destination on the map? Better hope you have a competent navigator riding shotgun, because otherwise you'll have to stop and park 15 times to consult your map while navigating an unfamiliar jungle of a city.

I think much of the software written over the last 20 years is very close to worthless. A significant fraction has negative worth.

But geospatial software is amazing.

I'd expect pretty good results from simulation until you start bending/deforming the limbs or until the foot pads start slipping.

I also wonder if you can read the impact force from the legs with the ground from the motor controllers (should be able to infer that from motor current + leg displacement). Learning to gallop without "feeling" the legs impacting sounds extremely difficult.

The jumps are pretty impressive, this thing has some power. I'd be very curious how fast you could get this dog with some reinforcement learning for a proper transverse gallop gait - and if it converges towards a gallop naturally, or if it discovers some other fast gait patterns during learning.

Depending on the max speed of the motors/legs, giving it longer foot pads might be necessary for a good gallop. Intuitively, it looks a bit... "low gear" in the videos.

I was waiting for Google to pull a local LLM onto Chrome/Android devices. It opens up some revenue streams that weren't easily possible before: for example the often memed "I was talking about cigars with my wife one single time and now all I see are adsense ads for cigars" gets much easier with a local model doing speech to text and topic classification.

No, but mostly for economic reasons. You can farm a whole lot of fish in aquaculture - it's just more expensive than importing wild caught fish.

The numbers look pretty insane, you can raise many tons of fish in relatively small volumes of water (several hundred kg of fish per year per cubic meter). You just gotta build the ponds/tanks/cages, and the infrastructure to filter the water, supply the oxygen and deliver the feed.

It's not a comprehensive dataset. The US installed 43 GW_peak in 2025, which should be around 80M new panels.

Still, an order of magnitude less new capacity than China - but not two orders.

I assume the "briefly gets worse" is when a buch of hyperscalers do a complete write-off of their entire AI investments, bankrupting several of them (which, in turn, bankrupts several large banks and most current venture capital firms)?

Cumulative AI capex will hit $2T this year. Cumulative opex is on the same order. Unless the models get real good (as in: can fully replace many engineers) right quick, nobody is even going to see interest getting paid on those investments. The only alternative is model access costing 5 figures per (replaced) seat.

But yes, once GPU racks can be had at auction for pennies on the dollar, inference of open source models might be an... OK low margin commodity business.

> But perhaps still useful for planning, as a starting place?

Yeah, lots of value in just having an app go "the mix you're trying to do is likely to fail the slump test". So you still have time to adjust the water ratio, or get better sand.

It is simply difficult to hide. You can just go and look at the infrastructure, after all. I bet almost all the information is from OpenStreetmaps, and people just walked around and added all the power lines, substations and powerplants they saw by hand.

And sure, you can bury the cables, or you can try keeping the output of your powerplants secret. But then the infrastructure nerds (or foreign spies) just count coal hopper railway cars per day and analyze cooling tower dimensions.