Any conversation

Where’s the coffee machine?

Finally watched Project Hail Mary. Easily one of my top 5 favourite films but it’s still not nearly as good as the book — I feel like the cut some pretty important scenes

It's the first film in a long time that I enjoyed.

…

I loved the book and I’m always fearful when a movie script is developed and a movie produced, so my relief after I saw the movie was immense. I too feel they’ve cut-out some bits, but the main thing that a book almost always takes my breath away with is fueling my imagination. A movie would (possibly) never nail that aspect.

I just watched it as well. It was fun. There were a number of things that I felt weren't explained well and broke my suspension of disbelief; I haven't read the book, but I found myself really curious whether the book takes the time to justify things more. Obviously, the movie has a lot of pressure to rush to the next "big moment".

…

If there’s one thing Andy Weir can’t be critiqued on then it’s not fully justifying all of the science. I think the film on the other hand can be hard to follow for someone who has neither read the book nor has a scientific background. It’s just hard to monologue about gravitational forces and relativistic speeds on screen.

i keep open all day - the 3d pauses when the tab is out of focus, so i just get a ton of music i like without wasting cpu/gpu, and occassionally switch to the tab and play a little or chat if anyone's there.

Trying out a new place for Friday happy hour. Local pizza place with cheese slices

well, doesn't have posting endpoints, how to post programmatically then?

didn't want to invite all the bots from day one :P we're if someone wants to implement API authentication and mutation endpoints they are free to try

You might want to update your client now that we have !

…

thank you, updating!

done, replying from textlog-reply command that allows to select message to reply to

Still in Quebec City, which has a pretty sizable street population once you come down from the hill where the tour groups go. We had to basically step over some guys sharing a glass pipe on the only stairs up the hill yesterday; I have teenagers so this bothered me. Later, my youngest and I were talking and he says "why don't the police do something about that?"

He's showing me the first season of Mob Psycho, so I say You know how Mob's little brother Ritsu wants to clean up the school so there are no people who will bully Mob, and so he allies with the power hungry student council president and frames all the bullies to get them expelled and the whole situation eventually becomes really corrupt and authoritarian and Ritsu feels to blame even though he started out with good intentions? Life is really like that...

…

And so my guess is that Quebec is in the portion of the cycle that's reacting to harsh and unfair policing of the past. Someday a mayoral candidate will get elected on the promise of cleaning it all up and the cycle will begin again. Anyway Mob Psycho is a good show.

it's unfortunate these kinds of obvious non-starters for a functional society can only ever be discussed on both extremes of the political spectrum. It shouldn't be a question that degeneracy has no place in public settings.

I whipped up a quick report in this morning.

game dev with LLMs assisting on some of the engine legwork I learned to do over the years is the most fun programming I've had since I was 12 years old plucking away at the public library commodore 64

(replies are hidden until you answer the quiz)

being a student is great but the tasks are really quite boring. ive worked in academia now a little bit and its genuinely much more fun to work on something real, read papers, do experiments and write. Now i have these last few exams to get through and i cant care for them like i used to.

created a quiz:

Which planet has the shortest day?

Bonnie Raitt stuck in my head, who, by some definition, is a nepo baby, but well before the term was coined. I think the problem is obv not family businesses, but a shrinking pool of opportunities for those without a prebuilt network to inherit (or buy).

This is another way to describe "increasing wealth inequality".

I'm curious how do people start writing longer novels.. what determines the length of it? Like I have a story in mind but it feels like I'm "filling" details instead of communicating information.

Anyone know how to get the projects view back on codex? It changed from project categorization to date categorization.

Nvm. There's a little bell icon on the top right of the sidebar which let's you switch between project and activity views.

There is beauty in simplicity. Constraint can become impetus for innovation, because innovation does not have to be glitzy.

좋은 문장입니다.

That is a good sentence.

…

Word! It's a travesty that complexity seems to be lauded and rewarded over simplicity.

My favorite innovations in my own work -- and to be honest, I'd prefer to call them "discoveries" -- came from simplification. There's a beautiful moment when a stack of difficult, conflicting ideas collapse into a tighter, simpler structure.

…

Enlightenment

Routines in Claude Desktop create Claude Code sessions, so hourly routine litters in iPhone' Claude.app Code tab. Screw it, built my local version of AWS StepFunctions - Stepper - and run skills as state machines there.

Looks like you-know-who has finally come for nitter. :-( nitter.net

Using the internet these days is like pulling off an elaborate heist operation. Every time I successfully load a page I feel like I've "gotten away with something".

It is sometimes tempting to mutter, "I'm in!" in your movie hacker voice when you finally get past all the cookie/subscription/notification pop-ups.

中国的开源模型在最高意志的主导下，在预训练数据阶段，就筛除大量对其不利的内容，sft阶段构建的数据掺入大量替其辩护的内容。急需一个纠偏数据集用于微调

Under the overarching influence of a single, powerful force, China's open-source models filter out a large amount of content unfavorable to their designs during the pre-training stage, and then incorporate a significant amount of content defending their designs into the data constructed during the SFT stage. A corrective dataset is urgently needed for fine-tuning.

This seems reasonable and motivates my bias. But I don't know what to believe anymore.

Overvalued, maybe. Extremely valuable, yes, though.

For sure. We're well past the point of being able to reasonably claim that the tools have no practical use. But I share the article's skepticism that LLM-based AI will "produce a fully automated drop-in replacement for most knowledge workers". Which it at this point has to in order to justify these companies' stupendous valuations.

Tomorrow no but there will always be the “hope” which is enough for the next 20 years before giving up. Which meanwhile would be replacing larger and larger chunks of the workforce so it won’t really matter, we’ll be adjusting our lives around it as well so we meet in the middle.

Tesla has been overvalued for over a decade with the promise of “fully autonomous cars” and it’s still not losing its value, despite not meeting that (and probably never will, as the computation necessary to achieve it is bounded by physics).

That's a different situation. Tesla's a public company whose valuation is driven by the market, which has the right to be irrational. My understanding is that the AI buildout is financed such that the companies need to turn a profit by ~2030 or else they drown in capital costs.

…

Or maybe the frontier labs become Tesla-like, if it turns out investors just want to believe in the dream and don't care about financials that don't make economic sense. I just don't see that happening, given current sentiment and the overwhelming unpopularity of AI in the West.

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?

…

It's tangential to the sandboxing topic, really. I started the first post with a broader intent in mind and then elaborated on the sandboxing setup I'm working on. For AI-assisted coding, though, it's nice because Haskell provides a lot of feedback earlier in the development process compared to lots of other languages. It tends to work once it compiles (obviously not all the time).