Hacker News: ahofmann

New comment by ahofmann in "GitHub's Historic Uptime"

ahofmann — Wed, 01 Apr 2026 05:56:21 +0000

I loved that time. Websites, or "apps" that don't change every second time I want to use them, are great.

New comment by ahofmann in "Objections to systemd age-attestation changes go overboard"

ahofmann — Tue, 31 Mar 2026 20:14:21 +0000

I think the "overboard" part is that the developer was doxxed and received death threats.

New comment by ahofmann in "Show HN: Cerno – CAPTCHA that targets LLM reasoning, not human biology"

ahofmann — Tue, 31 Mar 2026 20:11:28 +0000

I managed once to complete the maze, but then it went into "checking" and a spinner forever.

New comment by ahofmann in "We broke 92% of SHA-256 – you should start to migrate from it"

ahofmann — Sat, 28 Mar 2026 11:16:46 +0000

Well does it? If I would be able to break SHA2 and make myself rich with it, I would need to be sure that nobody would find out, that Bitcoin can be manipulated. The second anyone is aware, I wouldn't be rich anymore.

New comment by ahofmann in "Local LLM App by Ente"

ahofmann — Wed, 25 Mar 2026 16:41:39 +0000

Sorry, English is not my first language and I tried to look clever.

New comment by ahofmann in "Ensu – Ente’s Local LLM app"

ahofmann — Wed, 25 Mar 2026 13:48:06 +0000

While I would have the same reaction, in this case I think it is a sane decision. Ente is cornering the privacy market and I think they're doing a great job. They have a lot to lose (trust) and it would be stupid if they did something shady with the data entered in the 2FA app.

New comment by ahofmann in "Ensu – Ente’s Local LLM app"

ahofmann — Wed, 25 Mar 2026 13:42:37 +0000

I have the same questions. After installing the app, it downloads 2.5 GB of data. I presume this is the model.

New comment by ahofmann in "Is it a pint?"

ahofmann — Mon, 23 Mar 2026 17:00:13 +0000

I remember watching the cornetto trilogy ( https://en.wikipedia.org/wiki/Three_Flavours_Cornetto ) and always wondering why on earth they fill up the glasses up to the brim. It is uncomfortable to transport this way and not spill beer.

New comment by ahofmann in "FSFE supporters affected: Payment provider Nexi cancelled us"

ahofmann — Mon, 23 Mar 2026 15:44:43 +0000

Just a follow up: I wrote nexi germany via some contact form, that I will avoid using their services because of that story. They called me back and told me, that they asked the fsfe for a test account only. They also made an internal investigation, if someone asked for passwords of real accounts, which is a clear no-go for them.

New comment by ahofmann in "Ubuntu 26.04 Ends 46 Years of Silent sudo Passwords"

ahofmann — Sat, 21 Mar 2026 09:02:31 +0000

Because you long forgot how confusing it was, that you can't see if your keystrokes are accepted by the machine. This is a change for people, that are new to Linux/Unix

New comment by ahofmann in "Ubuntu 26.04 Ends 46 Years of Silent sudo Passwords"

ahofmann — Sat, 21 Mar 2026 08:03:13 +0000

I also think it is a good decision. Nevertheless it breaks the workflow of at least one person. My father's Linux password is one character. I didn't knew this when I supported him over screen sharing methods, because I couldn't see it. He told me, so now I know. But the silent prompt protected that fact. It is still a good decision, an one character password is useless from a security standpoint.

New comment by ahofmann in "Ask HN: How do you deal with people who trust LLMs?"

ahofmann — Thu, 19 Mar 2026 07:07:20 +0000

I have access to the ChatGPT account of my boss and it is unusable sycophancy slop, horrible to read because every information is buried under endless emojis and the like. And it is almost indistinguishable if the LLM is wrong or right, every answer looks the same, often with a "my final answer" at the end. It's a mess.

I'm using Claude Opus 4.6 and it is much calmer, or "professional" in tone and much more information and almost no fluff.

New comment by ahofmann in "Digg is gone again"

ahofmann — Sat, 14 Mar 2026 08:30:21 +0000

150m page views a month is peanuts and very far away from the "social" networks numbers. I don't have those numbers, but I know how many page views we had 2011 while running a german browser game community.

New comment by ahofmann in "Google closes deal to acquire Wiz"

ahofmann — Wed, 11 Mar 2026 21:51:30 +0000

And "Witz" means "joke" in german.

New comment by ahofmann in "The dead Internet is not a theory anymore"

ahofmann — Wed, 11 Mar 2026 21:47:25 +0000

Small internet isn't very attractive for most bots. Also, I use websites that are invite-only. This is effectively a web of trust. This works pretty well, bots aren't a real problem there.

New comment by ahofmann in "The dead Internet is not a theory anymore"

ahofmann — Wed, 11 Mar 2026 21:44:32 +0000

Bandwidth is only expensive in the US, somehow. Here in Germany I didn't bother about bots and their additional traffic since 1998 (there are other annoying things about bots though).

New comment by ahofmann in "100M-Row Challenge with PHP"

ahofmann — Tue, 03 Mar 2026 08:45:30 +0000

They are down to 2,6 seconds now, wow.

New comment by ahofmann in "Claude Code LSP"

ahofmann — Mon, 02 Mar 2026 12:36:20 +0000

This blog post is in the uncanny valley of "looks and sounds nice, but a bit too nice, could be useful, could be useless AI-slop, idk".

New comment by ahofmann in "HN is drowning in AI comments"

ahofmann — Sun, 01 Mar 2026 09:07:55 +0000

I would like to encourage you to step back a little and try to get a broader view on this issue.

I don't want to discuss whether there is more LLM-generated content or not. There clearly is, and there is no feasible way to get rid of it , because there is simply no reliable way to distinguish what was made by humans and what wasn't. Regardless of what is claimed, it is just not possible, if only because hybrid forms exist as well. This text was written by me, but reviewed and stylistically adjusted by an LLM.

It is therefore completely pointless to get upset about LLM content and demand anything from the moderators. All we are left with are our votes and the "reputation" of our user handles - and the awareness that we need to learn to consume content with a great deal of skepticism. We should have been doing this all along, but it seems to be something we struggle with.

On the internet, nobody knows you're a dog.

And yes, this may mean that we stop using anonymous online platforms altogether, because nothing on them can be relied upon anymore. That is a shame, but it cannot be stopped — Pandora's LLM box has been open for at least five years.

I therefore consider any discussion about banning LLM content to be futile. We are witnessing another Eternal September here: it couldn't be stopped back then, and it won't be stopped today either.

So what is there left to discuss?

New comment by ahofmann in "Dear Time Lords: Freeze Computers in 1993"

ahofmann — Fri, 27 Feb 2026 14:04:01 +0000

This naming scheme made things very obvious. While it wasn't the most creative, it was objectively the best ;-)