rixed

Agreed. If it were my money, I'd encourage candidates to write their proposal themselves because any slop, from an LLM or otherwise, would not even be read in full. And I would encourage them to be more ambitious than in the pre-AI era, and require them to prove they can make good use of AI to reach their goal. It does not look like a good use of money to mandate human typed code everywhere. But I would also probably ask for a plan regarding use of resources in general, performance, QA, etc. I wonder how long before they change their mind.

The tidal wave of AI slop took even the dutch organisation by surprise

Not sure I understand your question, but in my case the agents live in their own monitoring dedicated boxes. The perimeter of those boxes are the identification.

What use case do you have in mind?

Wars quickly replace one another, how unfortunate!

Or we have just independently rediscovered why most regimes seek the justification of some external threats to strengthen their power over their population, 1984 style.

Super interesting read that's worth its own HN submission.

Letting an AI agent poke at the real network is effective but a little scary. Introducing a purpose-built packet language with guardrails.

(code and manual: https://github.com/rixed/robinet/blob/master/docs/crafty.md)

Could you give any hint about the title/authors/year of this report that would allow me to find it?

Any competition that does not select for a specific skill will select people who are good at competition. Psychopaths are good at competition.

What I would really like to know, is:

- why do meetings make attendees feel important?

- why do everyone pretend to dislike meeting, yet obviously very few actualy do

Isn't someone already working on a system were every employee has an AI companion that would represent them during meetings? The employee chat about the meeting's topic with her avatar beforehand until they are on the same page, then for all the avatars hold a meeting at the speed of light and come back with the overall decision that was reached. No more politics, and perfectly asynchronous. I'd love it.

Political ideologies are not scientific theories, they are just words layered on top of mental confusion. It seems vain to try to make sense of it.

Don't judge people by what they say but by what they do, goes the saying. Pay even less attention to what they call themselves.

Are we speaking of the same industry? The "ship fast and break things" one? The one that has traded everything from performances to security in exchange for "better developer experience" (aka "easy" aka "cheap") ??

Yes, for their sense of nuance.

Just as Claude is working in the background to fix any bug it can find in my ancient parser-combinator library in OCaml [0]. It's written in literate programing style, so it's not for everyone (I guess f you love the posted animated slide show, you won't like literate programming :)) but it has some cool features (and, according to Claude, some frightening bugs too): it can parse in stream mode, and it can be given an "error budget" to automatically fix syntax errors (be careful with this, as the complexity goes to the roof so quick I've never benn able to use this feature in any real scenario).

The main advantage of such a parser (and the reason why I'm reviving this one right now) is because they accept syntaxes that most other type of parsers would find ambiguous.

[0]: https://github.com/rixed/ocaml-parsercombinator

The modern version of the mythical man month: adding subagents to an already large context window makes it even larger. And harder to cache.

Will try to remember this next time I'm told that devs don't care about performance.

I'm not sure how knowledgeable folks can keep enjoying their life without burying their head in the sand.

And I though the BYD U9 jumping car was a gimmick...

If you compare with the early universe right after the big bang, the upward trend is even clearer!

The generation supporting the boomers?

Isn't that trivially true that all human problems are temporary?

Because that's simple, and quality doesn't matter?

Not my experience either. I find it both verbose and obscure, in its comments or explanations. It also oftentimes assumes that I know what I don't and that I ignore what I actually know.

The screenshot of commit history on their landing page mentions "MIT licence", yet many of their repo are actually AGPL.

I'm not into bitcoins nor physical wallets, but I'm guessing those are not supposed to attach any gsm network ever. Isn't that trivial to detect, even without opening a device, that something inside is attaching the gsm network? Those are pretty strong signals in a well defined tiny range of frequency, after all.

(Not so long ago you could hear them talk with the towers through regular home speakers.)

> Claude Code now ships free with builtin source control management

They can still do that without releasing the code, can't they?

What is not satisfying at all with this analogy for me, are the sticky notes: a lot of intelligence has to be pushed into the sticky notes mechanism: How to know that there is enough of that molecule in and outside the cell, and on which gene to stick that note?? It feels like in order to sketch a simple portrait of the mechanism of the genes then all the magic have moved into the sticky notes.

Or the obvious:

3. Use their military advantage while it lasts

Yet, it's the same media. Sure, "the same media" is already something. I remember a supposedly no-strings attached phone company that you could only cancel with a plain old snail mail (that was in EU by the way, for those of you who believe EU is the paragon of consumer protection).

In my experience, it's more about "local" vs "large-scale" thinking. An LLM will be better at finding a bug that's "visible on screen" (or on the context window), whereas some design decisions that are hard to even put in words can be super important in practice (like, in C, how you manage the lifetimes of various objects, how you organise your threads and locks...) That's not purely "word manipulation", and there are not many good explanations of that on the internet, so it's understandably harder to learn for an LLM. A bit like chess vs go. They will get there quickly I'm sure. After all, AI beats us at go, too.

Maybe it shows that France is a better place for your VPN exit node than Russia? :)