HNHacker News
TopNewBestAskShowJobs

Anoian

97 karma · joined March 8, 2026

submissionscomments
Anoian··on Codex Micro
Yeah because open offices really need people talking to their computers next to the meetings they always have.
Anoian··on European "age verification" "app" forcing everyone to use Android or iOS
This highly depends on your world view. If you don't care about people or society, then yes you can be right. I, for one, want to be part of society.
Anoian··on Notes on Software Quality
The same applies to software though.

Most people with bad vision want higher contrast for text, while most people with dyslexia want text with lower contrast.

Anoian··on Side-Project: Free icebreaker question web app
Made this in my free time, I am not making any money with this, this is just for fun.
Anoian··on The great blogging collapse: What happened to 100 successful blogs?
What is your blog? I love reading blogs that exist just for the sake of it.
Anoian··on Most arguments are about ego, not ideas
I find it to be very effective to make the people I talk to come up with solutions instead of handing them to them.

Instead of telling them "We should do x, because its better for y", I ask questions even though I have the answer in mind: "Hmmm... I was thinking about y, how could we solve that?" and the answer is often "x" or something else that sometimes even impresses me.

Anoian··on Instagram is incorporating users' photos in ads for Meta Glasses
Meta glasses are basically for people who do not care about privacy so this ad promotes that brand perfectly.

Its like designing bad email spam to only get the gullible leads and not the smart people who will smell foul play from miles away.

Anoian··on Feds Killed Polestar and Spared Volvo
The audacity of the USA to do this to the entire world but banning anything foreign that does it to them.
Anoian··on Anthropic says Alibaba illicitly extracted Claude AI model capabilities
Haven't we all been parasitic towards china for the last half century though? They were our source of cheap labor and they got out of being the ones being used by playing everybody and stealing knowledge.

This all feels like everybody is playing everybody dirty.

Anoian··on Anthropic says Alibaba illicitly extracted Claude AI model capabilities
In Kindergarten there are three children, one has made a fun toy (copyrighted material), another took the toy and made a toy replication machine out of it without the first kids permission. The third kid replicated the toy creation machine by looking at it. The second kid is now accusing the third kid of theft.

I wouldn’t shout too loud if I were the second kid.

They’ve shouted before about the dangers of their favorite toy and the teacher took it away from them, lets see what happens if they shout too much this time.

Anoian··on Never Give Them Your Face
Although these are indicators, real people also use these sometimes.
Anoian··on GPT-5.5 hallucinates 3x more than MIT-licensed GLM-5.2
I don't understand how this did not reach frontpage. Great post!
Anoian··on Stop Using JWTs
How does an identity server store auth information with the user?
Anoian··on SpaceX to buy Cursor for $60B
Tesla no idea, but for Grok:

It probably now gets a prime position or default selection in Cursor, which is somewhat widely used.

They aquired not just a company but a market for their models.

Anoian··on Statement on US government directive to suspend access to Fable 5 and Mythos 5
I think the implications of this to companies like Mistral are also really big. Imagine Mistral developing a model with the same capabilities like Mythos. The government might order that all american infrastructure must be shut down for mistral.
Anoian··on AI agent bankrupted their operator while trying to scan DN42
This is absolutely beautiful
Anoian··on Building an HTML-first site doubled our users overnight
I am a tough guy, I only cried for twenty minutes after reading this.
Anoian··on LLMs are eroding my software engineering career and I don't know what to do
Too much code means more complexity for humans and LLMs to comprehend. More complexity means harder to change and more prone to bugs.

This is not just right, no matter on which side of the argument you are.

Anoian··on Changing how we develop Ladybird
I think GitHub should fix this.

They should make the barrier to create a pull request higher, they should detect slop pull requests, they should hand maintainers tools to deal with this bullshit.

Sadly they do not have the right motivations to create all of this. Because all of this would mean less usage of AI.

If I were GitLab or Forgejo, I would be working exactly on this, to get the projects who are struggling the most with this problem on my platform.

Anoian··on AI Engineers aren't safe from being replaced by AI
Once we have AI, that can infinitely improve itself in all directions humanly imaginable, then we have solved AGI, by which time we will have other things to worry about.
Anoian··on The newest Instagram “exploit” is the goofiest I've seen
That solution is like putting glitter on poop.

The entire process is bad, doing the measures you described would have helped, but it's still negligent to allow AI to do password resets to arbitrary emails without auth, even if the account is of low value.

Anoian··on I Will Never Use AI to Code
This resonates a lot with how I felt about it, too. I came to the conclusion that if you cannot fight it, join it and search for other things that can be fun. I wrote a blog post about it which I will link, but to give you a summary: People don't pay you for having fun, they pay you for solving problems and creating something that sells, and AI often makes me faster in achieving that. So the fight is futile. What remains valuable are the outcomes, so one should start owning the outcomes instead of the tool to achieve them. Find joy in solving real problems for real people, find joy in becoming one of the best at using AI and seeing usage of it itself as a craft one can be proud of, and find joy in producing quality output with AI. If AI is bad at producing quality outputs, wouldn't it be a lot of fun to figure out how to change that?

If you want to read my thoughts on the matter in more detail, here is my blog post: https://fabianletsch.de/blog/ai-made-me-hate-my-job/

Anoian··on Three-Em Dash
Sloppy: "–"

Sloppier "—"

Sloppiest ⸻

Anoian··on Red Squares – GitHub outages as contributions
That would actually be a product one might buy
Anoian··on Red Squares – GitHub outages as contributions
Why cannot both sides be right?

Meaning that the load is crazy high but also that the tech stack and everything else is bad?

Anoian··on [dead]
Text can have all of the following conditions:

human written, valuable, not slop human written, not valuable, not slop human written, valuable, slop human written, not valuable, slop ai written, valuable, not slop ai written, not valuable, not slop ai written, valuable, slop ai written, not valuable, slop

I don't think that valuable and slop are opposites, nor that ai written means slop.

Anoian··on Cybersecurity looks like proof of work now
The best way to secure your code is to have less of it. Thats why i try to cut dependencies in my own projects like my life depends on it. This way i also don't have to update things to stay secure.
← PreviousPage 2 of 2