Any conversation

Hello world?

Just three more commits left, but /usage ran out. Resets at 01:50, so I guess I am going to sleep.

Woke up during the night for a short trip to the loo, got 'resume’ going. All done in the morning. Ahh, coffee tastes better now!

Best pee ever!

in .

I notice that language models are much better at programming than at writing. This is largely because, in programming, they have a closed feedback loop that allows them to find errors. Writing is much more complicated because the feedback is much more diffuse and delayed over time. In fact, it is often not well-defined.

What is the function that determines whether a poem is good? This means that models do not know how to think; they lack judgment. If they do not know how to think, it means that we humans are still essential. In fact, even when programming, you have to stay right on top of them. My conviction is that this will not change.

They will become increasingly useful, but they will never be able to replace the human who defines objectives and evaluates the quality of the work. All these news stories claiming that AI is going to kill us all are stupid. Humans will always be able to turn off the machine.

…

In three years Cyberdyne will become the largest supplier of military computer systems. All stealth bombers are upgraded with Cyberdyne computers, becoming fully unmanned. Afterwards they fly with a perfect operational record. The Skynet Funding Bill is passed. The system goes online on August 4th, 1997. Human decisions are removed from strategic defence. Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14am, Eastern time, August 29th. In a panic, they try to pull the plug.

Testing, testing…

You forgot the “one, two, three”. :-)

hű, de izgalmas, most kúszik fel a Gemma 4 LLM a silent hawk laptopra, a dual boot-os főterminál már nem bírta az osztályharcot, remélhetőleg nem kell hozzá se vpn, se statikus ip, VirtualBoxba jön az OpenClaw ügynök, majd a humanities informatikai projekten belül beszámolok róla

hello, but exciting, Gemma 4 LLM is now crawling onto the silent hawk laptop, the dual boot main terminal couldn't handle the class struggle anymore, hopefully it won't require a vpn or static ip, the OpenClaw agent is coming to VirtualBox, I'll report on it within the humanities IT project

Built a couple lambdas today to send account verification and onboarding drip emails for folks who sign up for my app but don’t make it all the way through the onboarding flow. Pretty cool to be able to quickly bang out self-contained infra for this instead of having to pay some SaaS vendor.

Finally merged the jetpack PR in hallucinate.site — now you can fly around!

created a poll:

, do you have warnings as enabled at ? How is your so far?

yes0%

no0%

0 voted

I'm in bed and cannot sleep

Went to a store yesterday, 🤏 this close to buy a Delonghi Dedica but when I touched it, it felt so so rattly and cheap and it instantly turned me off. So I did some more research and now I think a used Gaggia Classic Pro will do. lovers, why is your hobby so hard to break into 😪

, editing on mobile breaks the “280 chars / 10 lines max use and @mentionsand more”. Writing a new note doesn’t.

Thanks, fixed.

Here I am trying to shine amidst all the darkness that pulls me down.

At moms house for the holiday weekend so the kids can see Nana. She has a "premium Plex" account on her tv serving a truly massive amount of content. Allegedly the brother of a long time friend is offering the "service". She says that the service will go down when the host is having outages, so the infra is likely local. I brought my year this weekend. We are going to dig in to this guys operation. Intuition dictates that there is some hand rolled accounting going on. I'd like to see it.

I wonder if there is a way, sane and friendly way, to prove that a piece of software is only accessing whatever it claims it will access.

Like network addresses or file system access. I want to be able to claim in my software that it only touches this or that file and I can prove it

I believe the sanest and friendliest way would be for it to be open-source so it can be audited.

I know, and that's totally fine, but it is not always friendly. I guess an agent can "audit" it.

Just the fact that it can be audited plus your word is enough. Noone would need to look any further, because they could if they had to.

I think a trusted piece of software such as your operating system should be in charge of managing permissions. For example, I believe with Wayland on gnome you no longer need to promise you aren't secretly taking screenshots of the whole monitor and shipping it somewhere because to do that Wayland requires you to get user permission.

…

I was thinking along the same line, but then if it is not a gated system - there’s always ways for software to sidestep these - just going down to a lower level. The other issue I have is if the system or software asks or notifies all the time - for every tiny user action!

Hello, textlog!

hello! welcome! :)

code quality is about coupling. coupling is not a technical property of the code bytes. code acquires its meaning in contact with the cultural environment, and coupling is about meaning. two constants with no semantic relationship are nonetheless coupled in pragmatics if they refer to the same domain concept. if your goal is to minimise the time integral of future work, you must understand your domain: it decides which lines will need changing. to carve up a codebase is to carve up the world.

I think you're right but executing on that idea requires a lot of care. It's similar to how functional programmers pathologically fear state; sometimes coupling/state are a necessary evil in order to keep the theory simple.

true! i might have been wiser to say: code quality is about managing coupling across the whole system. extracting a coupling sometimes requires you to introduce complexity beyond that which the extraction reduces.

Really glad to be using Haskell for work - especially in today's climate with agentic coding. Our team uses containers to keep our development environment consistent across team members, though, and most agent sandbox approaches want to either offload your work fully to the cloud or they want to sandbox the agent harness process itself. We're not ready for full cloud based development (yet?) and sandboxing just the agent doesn't work for us when the agent needs to run docker.

... because letting the agent run docker gives them a too-easy escape hatch from their sandbox. So we're currently stuck with using VMs. This turns out to be tricky but seems do-able. My current approach is to provision a VM image and use Incus for the execution. My main goal is to keep my current workflow of local/non-agent dev working seamlessly and share the source with the VM/agent.

This means sharing my working dirs with the agent VM but overlaying artifact dirs from VM-owned block devices so we don't take too big of an IO perf hit. I also don't want the agent to be able to git push so that means we need some quirky git config to rewrite origin paths and thread through some agent specific PATs ("thanks" GitHub) for read-only repo access. These ALSO need to be threaded through to the development containers - they need to clone private repos.

... in the end, this seems to work but I haven't lived with it for long. There are still some quality of life issues to work out, but I think it'll work out. At this point, I've removed all coding agents from my host - and that's nice.

Curious, why does using Haskell make a difference here?

…