comboy

*that we know of

Haha, here's some random AI generated content:

    At least 225 judges have ruled in more than 700 cases that the administration's mandatory immigration detention policy likely violates the right to due process[1] The Fifth Amendment's Due Process Clause generally requires those having federal funds cut off to receive notice and an opportunity for a hearing, which was not provided in many of DOGE's spending freezes[2]

(there's more but what's the point)

1. https://www.justsecurity.org/107087/tracker-litigation-legal...

2. https://www.cbpp.org/research/federal-budget/many-trump-admi...

Yeah, good tests are associated with cost. I'd like to see benchmarks on big messy codebases and how models perform on a clearly defined task that's easy to verify.

I was thinking that tokens spent in such case could also be an interesting measure, but some agent can do small useful refactoring. Although prompt could specify to do the minimal change required to achieve the goal.

Not really related, but does anybody know if somebody's tracking same models performance on some benchmarks over time? Sometimes I feel like I'm being A/B tested.

People are loading huge interpreted environments for stuff that can be done from the command line. Run computations on complex objects where it could be a single machine instruction etc. The trend has been around for a long time.

lol, yeah

> We’ve rewritten Claude Code’s terminal rendering system to reduce flickering by roughly 85%.

tells you all you need to know

and I keep running it remotely through tmux, that explains so many things

edit: if they are writing it in react anyway (sic!) maybe we could at least get a web interface, skipping mapping it to terminal output part ..

Wow /insights is genuinely useful, perhaps CLI should be pushing that as a tip, if one has enough sessions, instead of keep nagging me about the frontend developer skill which I already have installed

In general CLI could be more reliable and responsive though, it's a text based env yet sometimes feel like running windows 95 on 386dx

It seems clear from the insights that some model is marking failure cases when things went wrong and likely reporting home, so that should be extremely valuable to Anthropic

I think stability and reliability have vastly improved over the last years in general (not necessarily talking about gh specifically)

It's just that everybody is using 100 tools and dependencies which themselves depend on 50 others to be working.

OpenX is becoming a bit like that hindu symbol associated with well being..

Claude getting clawed.

> a state sponsored threat actor

your CPU, your OS, CPU and firmware on your motherboard chips, ethernet, wifi, HDDs (btw did you know your sim card has JVM?), your browser, all your networking equipment in between, BGP and all the root certs and I'm just scratching the surface

the ballpark is on anther planet

Fascinating how HN is torn about vibe coding still. Everybody pretty much agrees that it works for some use cases, yet there is a flamewar (I mean, cultured, HN-type one) every time. People seem to be more comfortable in a binary mindset.

> That is extremely stupid. What does that ban get you?

confidence in firing coders I presume..

Me too, the problem is that it's hard to come up with tools that are needed but not made yet, and we don't want to end up with https://malus.sh/index.html

Perhaps we should start making LLM- open source projects (clearly marked as such). Created by LLMs, open for LLM contributions, with some clearly defined protocols I'd be interesting where it would go. I imagine it could start as a project with a simple instruction file to include in your project to try to find abstractions which can be useful to others as a library and look for specific kind of libraries. Some people want to help others even if they are sharing effectively money+time rather than their skill.

Although I'm afraid big part of these LLM contributions may be people trying to build their portfolio. Some known project contributor sounds better than having some LLM generated code under your name.

Alright, my original comment was wrong (as was the parent). I still stand by my opinion that it is not practical though.

Eshittification (by Cory Doctorov) is a shitty book but it does explain how that dynamic works.

Applying them to only one side of the conversation doesn't seem practical.

That's your worldview. Crocker's rules is that you don't have to take receiver feelings into account you just communicate efficently.

Not sure if it's a common knowledge but I've learned not that long ago that you can do "/compact your instructions here", if you just say what you are working on or what to keep explicitly it's much less painful.

In general LLMs for some reason are really bad at designing prompts for themselves. I tested it heavily on some data where there was a clear optimization function and ability to evaluate the results, and I easily beat opus every time with my chaotic full of typos prompts vs its methodological ones when it is writing instructions for itself or for other LLMs.

Your sim card is an entire computer.

I highly recommend Aerospace[1], went through a few approaches, I cared about not completely compromising security either, it works really well if you come from something like i3

1. https://github.com/nikitabobko/AeroSpace

What is the $/Mtok that would make you choose your time vs savings of running stuff locally?

Just to be clear, it may sound like a snarky comment but I'm really curious from you or others how do you see it. I mean there are some batches long running tasks where ignoring electricity it's kind of free but usually local generation is slower (and worse quality) and we all kind of want some stuff to get done.

Or is it not about the cost at all, just about not pushing your data into the clouds.

Under this name or not I think it's happening regardless..

Limited precision of float numbers is deterministic. But there's whole parallelism and how things are wired together, your generation may end up on a different hardware etc.

And models I work with (claude,gemini etc) have the temperature parameter when you are using API.

I agree, but let's say you are looking for a library to solve your problem - you see one repo updated 2 weeks ago and the other one updated 5 years ago - which one do you choose?

Oh how I despise these suggestions. You sometimes look for a way to express something and you are on the verge of giving the world something truly original, but as soon as your brain sees the suggestion it goes "oh yeah that fits"

My broken english now officially bumps my comments up instead of down. Sweet.

Just post it to social media to generate the text description for free (multiple humans in parallel work better)

If that sounds fun, you gonna love the story of Joy Milne who can smell the Parkinson's disease.