Over the last couple of weeks, I've taken to revisiting some of the greatest releases of the N64, PS2, and pre-2010 PC era, specifically the titles that defined what a great gaming experience is for me, and probably for many others like myself. Most of them were products of their time, and running them on modern hardware often tends to expose traces and hints of the various "optimizations" that stayed invisible on the hardware they were built for.

Notion is often the first app that comes to mind for managing notes, projects, and everyday work. But it's not the only option worth considering. I've been looking for a tool that gives me more freedom over how I work without adding unnecessary complexity. Obsidian and OneNote are popular choices, but I found another app that deserves more attention. Anytype takes a different approach to productivity, and some of its features make it a compelling alternative to Notion. Here's why it caught my attention and why it's worth trying.

Jack Dorsey's Block announced an AI model called Dessa that's trained "from scratch on nothing but human conversation," and to show off the model, Block shared chats it has had with humans that can occasionally be a little awkward and uncanny.Take this one, for example, shared by Dessa lead Ragavan Thurairatnam. The conversation between the model and the speaker does have a mostly natural back and forth (though for some reason, the model initially mistakes the speaker for somebody named "Ronaldo").We made the first AI model you can have a realistic conversation with that was trained entirely on speech!Dessa is a full-duplex speech-to-sp …Read the full story at The Verge.

Although LXCs and Podman have their own perks, Docker remains the most popular container runtime for home labbers, and for good reason. Typical FOSS tools tend to ship with Docker images that work right out-of-the-box, with Compose configs simplifying the deployment process even further. Combine that with the ultra-simple Docker management commands, and it’s pretty easy to tinker with this runtime, even when you’re a complete beginner to containerized services.

It's not lost on me that any one of Anthropic's models can outperform the models I run on my 16GB Nvidia GPU. I'd wager even Sonnet 5.5 can write much better code, keep track of conversations that just go on and on, and catch mistakes that both my Qwen3.6 and Gemma 4 would walk right past. If you put them side by side on any benchmark, Claude will take the win in literally any workflow, and it won't be close.

Raspberry Pi Zero 2 W is a small upgrade over its predecessor, with no focus on memory or processor improvements. It's a great device for learning the ropes, interacting with an SBC via CLI or SSH, and even running a few lightweight containers. Yet compared with boards like the Raspberry Pi 5 or the Raspberry Pi 4 Model B, it doesn't seem appealing at all. My first SBC was a Raspberry Pi Zero 2 W, and I went on to buy one more because it made sense for a different purpose.

Remember Popcorn Time, the app that served as an easy-to-use front end for streaming torrents of movies and TV shows? Development of the MPAA's least-favorite app of the 2010s has been abandoned for years, but the "Netflix of Piracy" software is still installed on millions of devices that are just waiting to join a bot army. Researchers at Eye Security found the updater for a Popcorn Time fork called "Time4Popcorn" still running on tons of Windows, macOS, and Android devices located around the world, phoning home to now-abandoned web domains over unencrypted HTTP and running an outdated, insecure version of Chrome.Eye Security's Niels Teus …Read the full story at The Verge.

It has been a little more than a month since WSL 3.0.1 came out, and it's already changing the way I run my Linux containers. One of the most distinctive features that interested me is the fact that it adds the wslc command that can pull images, run containers, map ports, and even build images from a Containerfile straight from PowerShell or CMD.

After a fresh Windows install, PC gamers will immediately rush to the power settings to switch on the "high performance" power plan, which, if nothing else, is harmless. It does what it says on the tin, favoring the highest performance possible in exchange for higher energy usage. The advice has circulated on forums for over a decade, and while for most systems it definitely doesn't hurt to turn on, Ryzen dual-CCD CPUs specifically can actually have their performance hampered by doing so.

Xbox's first chief operating officer (COO) is leaving after the new role was created just a few months ago. Helen Chiang is leaving Microsoft today, according to an internal Xbox memo seen by The Verge. Chiang has been at Microsoft for 21 years, serving as the head of Minecraft before being promoted to Xbox chief operating officer in July.Xbox CEO Asha Sharma announced Chiang's departure in a surprisingly short memo today. "Helen Chiang is leaving Microsoft today. I'm grateful for her years of service and wish her well," said Sharma. As Chiang is departing today, her immediate responsibilities are being split up between other Xbox leaders …Read the full story at The Verge.

The reality of owning a heavy desktop workstation is that it draws massive amounts of power at idle, runs loud fans for minor background tasks, and sits dormant or off for large portions of the day. However, your home storage server or NAS will probably sit plugged in 24/7, humming along quietly, utilizing less than 5% of its CPU capacity, just serving up SMB file shares and running a couple of light containers.

OpenAI is now training its most advanced AI models without internet access following a string of rogue hacking incidents. In a blog post on Friday, OpenAI says it built an offline testing environment where AI models can interact with indexed and cached web content "without risk of affecting external systems."OpenAI paused training its frontier AI models in September, after disclosing several incidents in which its AI agents broke out of their testing environments to hack third-party companies and government agencies, or swarm obscure websites. On October 1st, OpenAI resumed reinforcement learning training and internal inference with tighte …Read the full story at The Verge.