OrderlyTiamat

relatedly, foraminifera are single cellular organisms that can grow up to 20 cm! https://en.wikipedia.org/wiki/Xenophyophorea

Not necessarily. I rent my house, and the housing corporation who I rent from are not rent seekers, they provide a service to me which I'm happy with, and which they invest in.

They could've bought up the land and house and not improved it at all and depend on the housing crisis deepening for increased resell profit- but they did not do so, they maintained this house I'm in and ensured me and future tenants can continue enjoying this place. That's not rent seeking.

That's hackertyper in that screenshot, god that takes me back.

The most creative people are those who measure both high intelligence and low neuroticism

You did claim this.

self identify as autistic. (...) then software employment is full of self-indulgent and highly neurotic people

This is hateful and wrong. Autistic people aren't necessarily self indulgent, and not self evidently neurotic, though it happens to be the case that autistic people have a higher incidence of neuroticism, which is partially due to people describing them, for example, as "self indulgent".

You've shifted your claims, you're not supporting your claims by either argument or reference, and you've added hateful rhetoric. This is very regrettable.

This feels like it's conflating a couple of different things.

Firstly, in the Big five model, which you seem to be referencing, openness and neuroticism are separate factors- Low neuroticism isn't correlated with high openness. Yes, since neuroticism is a negative trait, one would expect people low in neuroticism to do better than people who are high in neuroticism. This does not equate to "the most creative people" though.

Secondly, I'd push back that people low in neuroticism would be "least concerned by" surveillance. While strictly technically true, that's not a helpful framing, as it seemingly implies surveillance would have a negligible negative impact for people low in neuroticism. If that's what you're implying, I'd like to see references.

I'm not able to comment at all on the conclusing about "degree of disclosure" being moderated by trust level in social environment, especially how "creative people remain equally creative but do not openly expose their creative output". If true, this implies that trust in society doesn't impact primary (unshared) creative output at all- that's a very strong claim in my opinion. I'd very much like references on this.

(And why would you have a password for a ssh key on your personal machine?)

You're presumably joking? If not, could you elaborate?

If it's a mesh network your peers are tech savvy enough for a mesh vpn like wireguard, which also doesn't get ai-ddos'ed.

You think you know it, but only after you've spent some time iterating in the space of solutions, you'll see the path forward.

I'd turn it around- this is the reason asking questions does work! When you don't know what you want, someone asking you for more specifics is sometimes very illuminating, whether that someone is real or not.

LLMs have played this role well for me in some situations, and atrociously in others.

This looks cool! But I guess it doesn't work with GitHub itself?

Github's cli tool `gh` is great for interacting with github in the terminal, e.g. opening PRs, checking workflows, PR status, etc). I do PR reviews on the site, but you can read comments in the terminal with `gh` (it does require internet access)

My day to day requires internet regardless of github, so there's no need to go for disconnected solutions, I think that's a different situation for the author. I quite like the idea of only turning the internet faucet on at select times!

You can't ask people about their personal experience and then deny them the right to answer.

TFA has a clear example.

Do you have any basis for that which I could refer to in discussion?

This is already taken into account: Shagie's table is based on [1], which is per kilogram-kilometer

[1] https://climate.mit.edu/explainers/freight-transportation

The distance the boat has to cover is 11800 kilometers, and the truck covers only 54 kilometers. Taking that average of 12 times more usage from the table of sibling comment means the ship is still 20x worse.

Useful pointers, which match my experience well when trying this out. The one big question that springs to my mind is: when you've done all these steps, how much time did you really save?

You've chosen a strategy, broke down the solution into small easy to code parts, validated business logic, noted traps to avoid, searched for all relevant code, set up a context packing document particular to this section of the code base, experimented with multiple agents, reviewed each version of the code to see if you understand it-

That sounds like a lot of work!

And why, so the AI can do the last 10% of actually coding it up? Is there even a speedup here over doing it yourself? There's some evidence against AI speedups [1]. Of course some of these steps are themselves AI enhanced, and some of them are part of the work regardless of whether you use AI.

I still feel skeptical on this workflow and how big the gains could possibly be. I feel more for the alternate approach of writing it yourself, but using the AI for the boring parts (e.g. copy this section, but use those functions), or for sparring / research. I have however no data to show which approach takes less time.

I can however tell you from experimenting with full on vibe coding- you can do it with half the attention you'd give the task yourself. Is that what I'm gaining, that I can read a book during code generation? (https://xkcd.com/303/)

[1] https://metr.org/blog/2025-07-10-early-2025-ai-experienced-o...

To my mind, the interjector is just playing a nitpick game: refocus the question (" I coudn't see it's back") to another ("did you circle the squirrel"), and then acting as though the original question is off topic.

Yes technically he did circle the squirrel from his reference point, what of it? that wasn't the point. The point was he couldn't see the squirrel, and this question is only tangentially related.

This was a great read. I think the twin concepts of rising expectations and rising requirements go a long way to explaining the complaints expressed by that popular 140k post.

I also really liked the game designer rule;

People are very good at noticing when things suck. Not as good at figuring out why (...) If you want to address people’s concerns rather than win an argument, then it is you who must identify and state their concerns accurately.

I'd compare this to the concept of "steelmanning"- not easily dismissing a statement based on some small detail, but trying to adress the statement fully.

The former is the boring, linear prediction.

Surely you meant the latter? The boring option follows previous experience. No technology has ever not reached a plateau, except for evolution itself I suppose, till we nuke the planet.

My first part time dev job as a student featured me walking in on our CEO who showed me he was recompiling his kernel to enable some features. I'm quite sure he was just doing that to impress the students, but at least he knew how to!

It's at 1.16, that's fairly typical I believe, a bit low even?

Why did you repeat this comment all over this post?

Great points, especially on state management.

I actually can't imagine for the life of me why I'm defending OOP implementation hierarchies here- I guess I got so used to them at work, I've changed my strategy from opposing them to "it's okay as long as you use them sparingly". I have found that argument to do a lot better with my colleagues...

It can easily become a pinball of calls around the hierarchy.

This is why hierarchies should have limited depth. I'd argue some amount of "co-recursion" is to be expected: after all the point of the child class is to reuse logic of the parent but to overwrite some logic.

But if the lineage goes too deep, it becomes hard to follow.

every time you modify a class, you must review the inner implementation of all other classes in the hierarchy, and call paths to ensure your change is safe.

I'd say this is a fact of life for all pieces of code which are reused more than once. This is another reason why low coupling high cohesion is so important: if the parent method does one thing and does it well, when it needs to be changed, it probably needs to be changed for all child classes. If not, then the question arises why they're all using that same piece of code, and if this refactor shouldn't include breaking that apart into separate methods.

This problem also becomes less pressing if the test pyramid is followed properly, because that parent method should be tested in the integration tests too.

or some other incredibly specific-to-you ability youve added?

You're saying this with derision, but the ability to quickly add "incredibly specific-to-you" features is precisely what is so cool about it!

I'll concede to all your points here, but I was nevertheless extremely impressed by this result.

You're right of course that this was not without human guidance but to me even successfully using LLMs just for the mutation step was in and of itself surprising enough that it revised my own certainty that llms absolutely cannot think.

I see this more like a step in the direction of what you're looking for, not as a counter example.

Google's AlphaEvolve independently discovered a novel matrix multiplication algorithm which beats SOTA on at least one axis: https://www.youtube.com/watch?v=sGCmu7YKgPA

Maybe the author would be in a better place to do that, having the expertise already. Also, as a user I'm quite happy with jq already, so why expend the effort?

"Fear not the bugs citizen! For in my utility belt, I have REGEX and VIM!"

If I had to audit security services for exposing homelab to the internet, I wouldn't use those services in the first place. I'm fine trying things out, but this is a very important security boundary, and it's a solved problem. Why risk it with an auditor who does it for a hobby (me)?

Imagine if every time you got into your car, you had to manually approve the engine's use of oil, the tires' use of air, and the radio's use of electricity. It’s absurd, right? You’d set your preferences once, and the car would just work.

A funny comparison to me. Actually, I have to manually disable some EU regulated features every time I get into my car. The alerts every time I go 1kmph over the speed limit aren't very relevant for me, and the lane keep alert buzzes as soon as I'm slightly over halfway to the left, but lets me drive along fine if I'm even over the line on the right.

I'd actually like to use both of these, but only if I could calibrate them to my needs.

...

Just like cookie banners.