I guess anyone who uses ShitLocker is shit out of LUKS.

I run k3s on a single node and it’s not really that much more overhead than Docker Compose if you understand k8s. I mostly have a deployment.yaml, service.yaml, ingress.yaml, and network-policy.yaml for each service that I’ve copy / pasted and updated. Here are some of the benefits over Docker Compose for my setup:

• Has a built-in Traefik reverse proxy / ingress controller so I can access my services by domain name instead of by port, like http://jellyfin.lan/, http://forgejo.lan/ (using local dns on my OpenWRT router)

• I use the Calico CNI so I can have network policies for each service to allow them to access only what they need. If a service doesn’t need internet access, it doesn’t get it.

• I use Bitnami Sealed Secrets to store my secrets in YAML files that can be safely stored in git

• ConfigMaps make it easy to manage configuration files

• Easier to have separate YAML files for each service while sharing a network between them. Services connect to each other like http://forgejo.forgejo.svc.cluster.local/

Of course, if you’re looking to load balance across multiple machines, k3s makes even more sense.

We are planning to join the Plasma release schedule starting with Plasma 6.7 (in June), which will make it possible for distributions to ship it in their repositories.

I’m imagining it will only be available in rolling release distros for a while.

I assume Valve would prefer everyone use Steam.

I can see laying off the dead weight dipshits who are holding everyone else back. My company does layoffs every year, and while I think it’s shitty overall, my team did rejoice when a particular middle manager got the axe.

The article has specific advice about how to make sure the layoffs are a wise decision before cutting the jobs. In the movie, Office Space, they had consultants interviewing everyone and determining who is adding value or not. I guess those were the good ol’ days and now we’ve gotten to the point where the only consideration is whether this will make the numbers look good this quarter.

Maybe make regulations to limit the degree to which any car sold can spy on citizens? Nah, let’s limit competition so U.S. companies can keep making huge profits from inferior cars that still spy on everyone with no need to up their game.

I have a Linux HTPC on Mint where I have the Pegasus frontend with Jellyfin, VacuumTube, and a bunch of games. The main usability issue is the need to exit each app differently to get back to Pegasus. There’s no notion of using a home button to go back to the home screen and pause the current process, which is what I’d be hoping to get with Bigscreen. Otherwise, Pegasus is fine for launching apps with a remote, and Jellyfin and VacuumTube work fine with a remote as well.

Yeah, LLMs are useful tools, though not the silver bullet the hype proclaims them to be. The tech bros tightly controlling LLMs and chasing insane profits with their closed models, data centers, and subscriptions are the main problem. Open models like Qwen 3.7 27B that are approaching frontier capabilities while running on consumer hardware is really the only thing that gives me any hope for the future of LLMs.

IMO, there’s no expectation of privacy on a company device because it’s their device you’re using for the express purpose of doing work for them. However, a device like the Meta Quest that you bought and paid for, and yet the company insists on locking down and spying on you like they still own it. It’s mine, not yours; fuck off!

AI is not doing the jobs; the remaining employees are.

Interesting read. Here’s a link without the paywall: https://archive.ph/20260507223322/https://www.bloomberg.com/news/articles/2026-04-27/why-china-s-deepseek-qwen-and-moonshot-are-a-worry-for-us-ai-rivals

Right? Qwen 3.6 27B is fantastic, for example. Data centers need to stop hoarding all the hardware so we can afford to run models like that.

Spanking the code monkey.

I think the key point is that you’re not outsourcing critical thinking to LLMs, but are instead using it as a tool to do grunt work that you could’ve done yourself, but the LLM can pump out faster. This means constantly being critical of everything it does, asking questions, asking for links to credible sources, asking it to provide info to help evaluate the pros and cons of multiple approaches, with you making the decisions and learning along the way. Overall, any work a LLM produces that will have your name on it should be work you entirely understand and agree with. For coding, I find agent markdown files to be especially helpful to make sure the LLM follows my desired practices without me constantly making it refactor.

Largely, my assumption at this point is that LLMs may not always be around, so I definitely don’t want to be left holding the bag with a bunch of slop I can’t manage on my own. I think I’ll feel better when I can run open weight models on my own hardware that are fully competitive with cloud models. With models like Qwen 3.6 27B, it seems we are getting closer to that.

Are all these data centers really going to be running at full capacity with open models like Qwen 3.6 27B that have performance approaching frontier, but can run on consumer hardware? Sure, it’s slow as of now, though there are tweaks to optimize it, and how long until we see open models that run reasonably fast and give frontier models a run for their money? My company MacBook can run models like this, so will there be a point where companies stop paying hundreds per user per month for cloud AI and have devs run open models on the laptops they already have? I definitely won’t be surprised if that’s the case.