danielbln

Opus 5 writes like its off it's meds, opus 5.5 is a lot less weird that way.

Suffer? Where's that rugged individualism when you need it.

Please don't, not everyone appreciates small talk.

You state this with such conviction, do you have some sources for this? As I wrote in my other comment, that 2014 study is often cited here, but it did not replicate well.

Care to elaborate on this take?

The model clearly criticises the concentration of power of the church and manifests it with clusters of churches. Art!

Any moment now, guys, seriously, this time for realsies!

Long distance runner here, feels amazing from start to finish. Ok, amazing maybe only during runners high, but definitely good all the way through. Running sucks until it doesn't.

I don't get this thread. Just tell the agent to use conventional commit messages and to keep it nice and tight. Are you all just raw dogging agent output with no alignment/conventions?

I carefully bottle feed every function by hand, and let it out to pasture at least twice a day.

The same. Think in systems, interfaces, integrations, the problem at hand. Zoom out, leave the grunt work and yak shaving to the machine, and keep your eyes on the price. If you only get satisfaction when you hand set functions in a particular way, then the only thing to say is that time is moving on.

Same here. Loved hacking on things from all angles, still do. The only thing lost to me now is the yak shaving, and good riddance. I want to think in systems, solutions, integration, and first and foremost the higher level problem, not nitty gritty implementation details.

Ever since Fable and Sol level LLMs/agents came online, I have ceased to look at the code. I recently launched a project for a client that integrated many many systems, has considerable traffic with low latency requirements and everything just.. works. I've never had to do this little firefighting when launching a project of that complexity and scale. It's been liberating and I was able to focus mostly on polish and QoL improvements of the stack and systems. What a time to be alive for a technical builder.

Well, this one folds, so you get a much larger screen surface that still fits in your pocket. And $200 won't get you that from any vendor. And while the low end phones have gotten amazingly good, a flag ship will still have a much better camera, better screen, better battery etc. If that's worth the price is a decision everyone has to make for themselves.

Sidenote, I've not had a single one of my smartphones crack a screen since 2008 when I got my first smartphone, take that for what it's worth.

Na ja, "keiner" ist vielleicht etwas kurz gegriffen bei der nach Englisch am häufigst gesprochenen Sprache.

If capability increase continues as it has, then an incident that cannot be resolved by AI will stump humans no matter the practice.

I like the plane example from the article,but I think in reality it will be like code. 1.5 years ago engineers would routinely say that they still write code by hand here or there to keep their skills sharp, and that's just not something you hear much if at all.

If an SRE is faced with a situation an AI can't solve, then said SRE will use the AI systems to triage further, point it to different places and so on.

This works for SREs with pre-AI experience and intuition, possibly less so with new recruits coming in post-AI. I don't know what the solution to this is, maybe practice drills is it, but I have a hunch the entire field will be subsumed, same as many other engineering fields.

There is only so much need for taste and judgement, before even that has been incorporated into the models.

I may or may not have told my agent to run the "open" command on a text file.

Implementation does matter, my point is that different implementations can lead to the same result.

And why compare to the brain? Mostly because of complexity. I can't interface with a bacteria in any meaningful way, but I can interface with an LLM to a significant degree.

There are strong indications that brains are prediction engines. Tokens are an implementation detail of LLMs and irrelevant to the discussion.

Not more than a bunch of neurons have conciousness. But put them together and wire them up just right..

We are seeing the same. We are moving significantly faster and ship more complex stuff, and there are still bugs as they were in the before times, but nothing catastrophic, no gremlins in the machine. Our main worry is about process and communication, not the actual coding really.

Why can't an emergent capability arise in service of a predictive objective?

Where is the reasoning in synaptic transmission?

Maybe speak for yourself, or if you insist on that tone, don't even do that.

I've been raw dogging Claude Code for the better part of last year as well, but I'm under no illusion that it's safe to do so, I just cannot approve seven trillion approval prompts and stay sane. I've since. switched to auto mode, however.

An LLM is more akin to a staffed pro kitchen than a microwave.

I'm in Central Europe and there aren't many manual transmissions around anymore where I am. But if you paint with a brush of two entire continents, you might paint a little too broadly.

That article is surprisingly prescient.

I would wager not a lot. There are some real hard constraints in space, from power consumption, to weight to thermal output (lack of convection is a real PITA for thermal shedding), and the list goes on.

If my laptop breaks I'm not going back to pen and paper, I'm buying a new/different laptop.

Same here, iterm2 with CC in fullscreen mode and I haven't seen a flicker or other TUI display bug in months.