HNHacker News
TopNewBestAskShowJobs

PunchyHamster

4,416 karma · joined May 9, 2025

submissionscomments
PunchyHamster··on Typesafe AI raises $870M at $7.5B
"now"? It always was. Pre-AI we still had ton of great apps "losing" just because the couldn't be brought before the eyes of people that would like them
PunchyHamster··on Typesafe AI raises $870M at $7.5B
Because answering the why reduces investor hype train
PunchyHamster··on I tried to move my notes out of Emacs. I failed. Again
Ignorance seems to correlate 1:1 with hype for LLMs
PunchyHamster··on Programming Isn't Special
sounds like art people clueless about engineering again

And also we do worry, but not in "replacement" way but just "using AI to create bad code with no proper oversight". It's absolutely enabling clueless to develop crap apps

PunchyHamster··on OTel-Native by Design – Building Products That Export to Any Observability Stack
I wish the format wasn't so ass, it's so wasteful and annoying to modify
PunchyHamster··on Why isn't the industry freaking out about DeepSeek 4.1 Flash?
They are. That's what the push for regulations is
PunchyHamster··on US man given prison sentence for bot-farming music streams
The whole entire problem is the system that is used to divide the income from songs. If it it was "the share of your subscription (or whatever ad-supported account is worth) gets divided to the songs you listen", most of the problem would be gone, at the very least for money from subscription accounts.

Instead, it is all dumped in one big pool and distributed by total listens.

Which means, if you listen to more songs, you have more "voting power" on who gets the money. Absolutely asinine system.

PunchyHamster··on US man given prison sentence for bot-farming music streams
Because essentially they are counting it wrong.

They basically put all the revenue into one pot, then divide it by the songs play count.

Which means that account that streams 10 000 plays, have more "voting power" on the revenue that one that just listens "normally"

PunchyHamster··on OpenAI annualised revenues $20B less than previously signalled
I think that's mostly so govt blocks their competition as much as possible
PunchyHamster··on Meta and Microsoft take steps to reduce employee usage of Claude AI
It's funny, the thing that makes effective prompt also makes effective documentation/communication.

It's bizzare to see people that made clown issues (not enough detail etc.) suddenly start writing detailed prompts just because it is AI that will do the task and not the human on the other side.

PunchyHamster··on Jev-Driven SRE Diagnosis: What Worked and What Failed
My experience is that SRE-esque work you experience the full range of model, from "wow, it would take us ages to even get to this theory for why something failed" to "well the priority field in this protocol goes from lowest=most important, but LLM wrote code as if it was highest=most important, so if someone actually pushed it to production the company wouldn't have working internet access any more".

And a lot of "wishing up a feature", I was using it to explore solutions for a given problem in too I didn't knew 100% and it pretty much came to same solution I wanted to do but... the capabilities were not there in the tool so it just started making up probable config clauses, and of course, it didn't work.

Even on simpler stuff there were traps, for example in middle of debug session I asked it to modify Gitlab config to add request duration logging, so it added correct config format to a flag that didn't exist (option was there, just under different name), because it didn't bother to read the docs (since then I generally link it the docs first so it doesn't try to remember and get it wrong).

All of that is both very dangerous, and also easily fixed by just having competent operator there. And as a tool it's great, as replacement it is just AI bros delusion

PunchyHamster··on JetBrains reports revenue growth, net financial loss for 2025
> Well, I used to pay more attention to the code, and then I found that if I spent more tokens having the AI review its work than actually creating new features, I was able to get more, better work done.

Of course it did. Because reviewing code thoroughly is harder than writing it.

Writing only consists of "making sure it does the task", reviewer have to understand code first, check if it does the task, then check whether it doesn't affect something else negatively (and it can be meta things like "doesn't adhere to project style", even if technical side is correct)

PunchyHamster··on JetBrains reported a net financial loss first time in its tracked history
please tell me what you work on so I can pre-emptively ban it
PunchyHamster··on JetBrains reports revenue growth, net financial loss for 2025
https://junie.jetbrains.com/ probably.

No idea why they thought they can compete with frontier AI labs on that, it's entirely waste of money

They also tried to push AI feature set as a separate subscription which is just insane in current tooling market.

All users wanted is good integration of multi-AI providers in their base offering. I think they are slowly shifting to that but that's a lot of money wasted already

PunchyHamster··on JetBrains reports revenue growth, net financial loss for 2025
They picked weird direction. "Here, pay ENTIRE EXTRA SUB to use the LLM we made that is worse than one you're already using".

Yeah, no, vast majority just want to pick a AI vendor and integrate it into their IDEA or CLion or whatever other editor they were already using. Pushing into entirely different subscription was dead on arrival.

They figured that out and started integrating it into main tooling (https://plugins.jetbrains.com/plugin/33314-air), but they definitely wasted a bunch of resources on something nobody really wanted.

PunchyHamster··on JetBrains reported a net financial loss first time in its tracked history
> Maybe people smarter than me know better but couldn't there be a middle ground where an IDE/Editor has an embedded engine (doesn't need to be a full-on LLM) that doesn't require external tool calls and token spend?

IDEA already have small LLM for one line code completion IIRC.

But the gain people want from LLM is generally "here, add this entire feature" or "here, go thru every dependency's changelog and update code to work with latest version". Those are not small LLM tasks

PunchyHamster··on Linux containers in 500 lines of code (2016)
bwrap needs additive mode; take everything (aside some basics like stdin/out/err acces) by default, then add permissions.
PunchyHamster··on Linux containers in 500 lines of code (2016)
Container escape is already vanishingly small amount of attacks.

Like, if app in wild gets hacked, you kinda already are screwed, even if the hack is contained to the box (whether VM or container) app runs in, you still get whatever app keys app used, and you still get whatever visibility to internal network the app had. https://xkcd.com/1200/ basically.

If your app server gets hacked, all user data leaks anyway. If you divide everything to microservices so they see minimum required amount of data, attacker can still see everything that goes thru it

PunchyHamster··on Linux containers in 500 lines of code (2016)
That just uses cgroups. So, no
PunchyHamster··on ChatGPT is adding real cartoonists' signatures to fake New Yorker cartoons
They can't be ethically sourced and good at the same time.

The current models intelligence depends on massive training dataset of essentially stolen data

PunchyHamster··on Opus 5.5 agents discover two room-temperature magnetic semiconductor candidates
Hey at least people accept your architectural decisions without questioning!
PunchyHamster··on The technology to eradicate mosquito-borne disease exists
we are already wiping out many species, wiping out one that's basically always bad both for humans and other animals is probably net positive
PunchyHamster··on The Tao of Backup
but why make it a primary storage in the first place. Sync to cloud as a backup, not other way around

My setup is Syncthing + restic for backups history. I have VM "in the cloud" that act as extra node/shit hits the fan plan too

PunchyHamster··on The Tao of Backup
Does Nix keeps old version indefinitely ?

Like, you can do the same with any distro and say Puppet or Ansible but only to within that distro packages, any 3rd party stuff (and people use more and more stuff that's 3rd party to the distro) might just have dead link for the download.

Also, plain restore from backup is still faster

PunchyHamster··on Improper redaction reveals Google Data Center water and electricity usage
It is worth remembering that it heavily depends on cooling system involved.

Traditional data centers have far smaller power density per rack and thus cooling requirements than AI datacenters; if you can use closed loop system without evaporation, water usage is very small. Difference is around order of magnitude

But if you want to fit maximum amount of compute in the space and need to dissipate it, evaporative based system will be far more effective space wise; especially if also for some reason you built it in hot place, rather than somewhere in the north

PunchyHamster··on The evolution of effective altruism
Calling it "most important movement" is like calling national socialism "most important movement", just because it made history doesn't make it good
PunchyHamster··on Surely you have ultra-wideband radios on your bins too?
PoE cameras solve all of the cameras problems. Just drag a cable, it's one time bother and if you're clever it won't be any uglier than camera itself
PunchyHamster··on Surely you have ultra-wideband radios on your bins too?
Yeah, when it comes to sensors wired power (or solar power I guess) is absolute minimum, it's extra effort that's always pays off.
PunchyHamster··on Surely you have ultra-wideband radios on your bins too?
I'd just do it with camera and some rudimentary image recognition, having bin present/absent (in process of being taken care of) signal should've been enough
PunchyHamster··on Federal judge calls Flock 'indiscriminate mass surveillance'
you kinda need the image to be 100% sure there wasn't some software glitch or inaccurate OCR
← PreviousPage 2 of 34Next →