pseudosavant

I've found that there is value in consuming AI services from your existing cloud provider. Customers and auditors have less of an issue with "we use AI services from AWS/Azure/GCP" if the data was already in those clouds and it doesn't expand the risks of data being breached, or trained on, by some other provider.

When you are already trusting 100% of your data, and computing on that data, to someone like AWS, it doesn't meaningfully increase risk to use an additional service, even if it is an AI service.

This is definitely giving of vibes like what Deno/Bun are to Nodejs, Nobie is to Excel. I wish there was a Windows version. Hopefully, this gives the Excel team reason to compete and improve their product like what happened to Nodejs.

Maybe I'm just too risk adverse to run with scissors with Grok? I can't imagine ever trusting their harness.

I'll just add this to the list of things an AI company could do to guarantee I'd never use them. You know, like the AI referring to itself as Mecha Hitler, making non-consensual porn (even of minors), or deferring to Elon's tweets as authoritative references on topics.

Yes. There are no launchers for games that need to be updated. Everything just automatically gets updated constantly.

The biggest value prop for me with console gaming is that it just works, every time. No GPU/launcher/mod/etc updates to deal with ever.

I've owned some gaming PCs and every time I've learned that I don't want a hobby of maintaining a gaming PC. I want to sit down, in my favorite comfy chair, with a controller (because I've already been on a keyboard all day), and just play.

But the GTA V that people play today is not the version from 13 years ago for PS3 and Xbox 360.

We are on the third-generation of GTA V with the enhanced edition that only came out on PC last year, and Xbox/PS5 4 years ago. The latest version has 18x the pixels per second (9x resolution, 2x frame rate), and far more detailed models and textures than the Xbox 360/PS3 version.

It has been profitable for Rockstar to keep remaking the game with better and better visuals specifically because consumers are demanding it.

I haven't seriously used GLM/Qwen/etc for agentic work. I will say, that the GPT and Claude models from 6 months ago were very usable within their respective agentic harnesses (this matters a lot). But there is no way you could convince me to go back to GPT-5.0 from GPT-5.6 for a serious amount of agentic work.

I could probably accomplish most of the same tasks with GPT-5.0, but it would take a lot more involvement from me, more troubleshooting, and significantly more time.

That is pretty incredible with such a small amount of input. Makes me want to check out dots.tts myself!

That is a nice idea and all, but perhaps there are things our friends aren't good at talking about. Perhaps something that is very related to a work project? Or like when I go down the rabbit hole discussing quantum mechanics and astro physics with ChatGPT because none of my friends understand that at even my level, and I'm looking to learn more about it from someone/something that does know more than me.

I'm looking forward to trying this out because I find that doing discovery on certain things feels more natural with voice. The previous voice chat was such a dumb GPT-4 era model that I resorted to using the dictate feature with GPT-5.5 and had it read the text response back to me.

I have it going into a 16-port gigabit switch, which my OpenWrt-based APs and wired devices are hooked into.

I have and love my OpenWrt One for my main router. I have two, so that I have a backup one I can switch to if the first one ever dies. It is the best device to run OpenWrt on as it is fully supported hardware that has great images/packages for it. Routing speeds/buffer/latency are great, everything just works, price is very reasonable.

I don't use it for my APs, but that is mostly because I already had 3 TP-Link routers setup as dumb APs using OpenWrt that have been working great. If I did it again, I'd buy OpenWrt Ones though. Although Deco mesh kits I've used have worked exceptionally well, and have become my recommendation for friends/family that don't want to do things like run arbitrary packages on their router/APs.

I was going by Asahi's own support page: https://asahilinux.org/docs/platform/feature-support/m1/

It seems that external displays are only supported on HDMI (e.g. Mac Mini) but DP Alt mode is still listed as WIP so I'd assume MacBook's with only USB-C (all of them?) can't support an external display.

The sad reality of DRM is that only rentals and subscriptions can honestly be done with it. You can guarantee access to a subscription library (Netflix, Xbox Game Pass, etc) for 30 days. You cannot guarantee access 30+ years in the future. And selling access like that is just a lie.

And it doesn't have to be this way. I started buying Xbox games digitally for the Xbox 360, and with their backwards compatibility that ended up being a great choice. My current XSX plays games for the last 3 generations of digital purchases I've made.

Asahi still doesn't support a lot of basic things like: external displays, Thunderbolt, hardware accelerated video decoding, 120hz refresh rate, etc.

I applaud a new/better FFMPEG AAC encoder, but there are two pretty massive caveats that are mentioned in the specifics that need to be called out:

- CBR only

- Only optimized for 48khz sampling

Not being able to do quality-based variable bitrate encoding is a major gap, and since all of the CD audio in the world is at 44.1k sampling, that seems like a huge miss too.

No sympathy for them trying to “protect” the output of a model that’s trained on data that they didn’t get consent to use. Ripe hypocrisy.

So that the current US administration doesn't block broad usage of Sonnet 5 probably. They'd have to collect your ID and approve you if it was good at cybersecurity. Because such is the freedom in the U.S. right now.

I still see this more like there will be "devs" that only know how to use LLMs and not any layer under it.

Just like there were web devs that only knew jQuery but not any actual JS. Game devs that know only Unreal Engine or Unity but nothing about DirectX, OpenGL, or Vulkan. C/C++ devs that know nothing about LLVM IR or the actual bits their compiler generates. It can also be a poor understanding of the layers in the abstractions above you too.

I find a lot of value in being able to understand levels beneath the layer I'm mostly working in. There are things, even/especially with LLMs, that I can grasp because of my deeper understanding.

But there have been plenty in our field that have only known their level of the abstraction for quite a long time. Every time there is something introduced that lowers the barrier to entry to making things, there is a version of this "but they don't know how it really works..."

I can't help but think this type of step function change has happened with computers before.

The change from punch cards to magnetic storage certainly made it so you didn't have to "know how it works" for every single bit.

The change from machine code to a language like Fortran brought about such an abstraction that a Fortran dev didn't "know how it works" at that same level anymore.

At this point, the layers of abstractions between using a React component and something being rendered is immense. React VDOM, the real DOM, browser render engine (which sits on abstractions like ANGLE, skia, etc), calls OS APIs, which call driver APIs, and the lowest level of anything it is still C/C++ that is compiled (abstracted) to something closer to what the hardware expects.

I won't bore you with the ChatGPT output details, but it estimated at least 35 meaningful layers of abstraction between a React component and being rendered to the screen. LLMs seem to be the latest level of abstraction to make it so we "know how it works" less than before.

My read on it was that they are trying to imply a transistor density (in a 2D plane sense) that is comparable to a 1nm process? But they achieve that through stacking (3D, not 2D) since the features aren't actually anywhere near 1nm?

Totally agree. Having to handle the password reset flow requires having reliable email delivery, which can be non-trivial. I usually start with supporting Google auth, and if I need more than that I'll add Microsoft and Apple too. Everyone already has at least one of those setup.

I've gone through my own cycle of this as a musician. Early in my music experience I was always obsessed with originality, and wouldn't learn a lot of existing pieces. At the stage I'm at now, I find great value in learning great songs and understanding why they work.

There is a lot of great work out there and if you are unwilling to be derivative in anyway, you'll intentionally avoid using and finding great things that others have discovered.

Exactly. That is just more about the ecosystem you are already in. I've been buying most games digitally since Xbox 360, and they all still run on my XSX. The same would be true if I had been in the Playstation ecosystem.

I'm actually firmly in the console (XSX in my case) bracket for my gaming. I want it to just work without any fuss, and for way under $1000.

The main benefits to the PC gaming space (upgradability of components, higher than console performance, compatible with a diverse ecosystem) are all worse on the Steam Machine than basically any other gaming PC. And running Windows games on Linux is definitely more fuss for many/most games.

The Steam Machine is extremely small and supports HDMI-CEC. Other than that, it doesn't have a lot going for it, and it is priced at quite the premium. That is a questionable value prop for many.

The XSX costs 43% less than the cheapest Steam Machine, even though it has twice the SSD and includes a controller. Microsoft isn't subsidizing it that far.

As opposed to an Xbox, which doesn't do that. But the XSX also costs 43% less.

I think this put this all in perspective for me. If Microsoft came out with this exact hardware (with Windows but still open and you could load SteamOS on it), at this price point, people mock Microsoft to death about how this product is a joke.

Unified memory is a lot more flexible and efficient though. You don't have to have assets loaded in RAM and also VRAM for the CPU/GPU to use them. Don't forget about how much more RAM a general purpose OS like Steam OS can consume versus a gaming specific OS too. The PS5 Pro also has an extra 2GB of DDR5 system RAM too.

My old Ryzen 3700X gaming PC has 16GB of RAM and 8GB of VRAM (RTX 2070 Super) and there isn't any game that runs better on it than my Xbox Series X. And the GPU in the Steam Machine is slightly worse than an RTX 2070 Super.

Just filter out: Fortnite, Roblox, GTA Online, Call of Duty, Destiny, Valorant, PUBG, etc...

It isn't a short list of AAA games that don't run on SteamOS.

A slow gaming PC that is small and can turn on my TV is still... a slow gaming PC. And one of the main PC benefits, upgradeability is non-existent for the parts that matter (e.g. GPU, VRAM, etc).