Hacker News: basilikum

New comment by basilikum in "Why AI Sucks at Front End"

basilikum — Mon, 13 Apr 2026 16:43:45 +0000

It's really weird. I don't even care if they used AI as part of their development process. But most AI™ developed stuff is just so insanely soulless crap I can instantly can tell and instinctively close the tab.

If you're AI developed software was so great, I couldn't tell it's AI.

Like I cannot wrap my head around how anyone vibe slops and think "No, this is good. I will now proudly show this off."

New comment by basilikum in "Why AI Sucks at Front End"

basilikum — Mon, 13 Apr 2026 12:48:57 +0000

It shows.

The landing page looks like every other AI slopped product page out there.

New comment by basilikum in "We have a 99% email reputation, but Gmail disagrees"

basilikum — Sun, 12 Apr 2026 14:25:12 +0000

Why is this blog on a sudomain of wpcomstaging.com?

Is this actually an official site by fontawsome? If yes, what a pack of clowns. I hope their spam emails rot in every spam filter forever.

New comment by basilikum in "Adobe modifies hosts file to detect whether Creative Cloud is installed"

basilikum — Tue, 07 Apr 2026 00:09:16 +0000

I'd like to answer the closing question

> At what point does a commercial software suite become malware?

The vast majority of commercial software is malware.

New comment by basilikum in "Got kicked out of uni and had the cops called for a social media website I made"

basilikum — Mon, 06 Apr 2026 21:53:31 +0000

I really hope you'll find something in life to live for that actually benefits people.

Flatseal – GUI for managing Flatpak permissions

basilikum — Thu, 02 Apr 2026 13:22:41 +0000

Article URL: https://github.com/tchx84/Flatseal

Comments URL: https://news.ycombinator.com/item?id=47614155

Points: 3

# Comments: 0

Ask HN: How do you get LLMs to stop spewing corpo speak?

basilikum — Thu, 02 Apr 2026 12:39:39 +0000

Whenever I use an LLM chat bot for anything it always answers with this fake positive corp tone that we all know. And I absolutely hate that. I cannot stand corpo speak and the LLM style makes want to puke. To cut the rant short: How exactly do you stop it from doing this? I do not just mean the style of the response but also the response pattern: making assumptions instead of asking questions, always agreeing with me, having this fake differentiated view on everything, telling me it understood something and then doing the exact opposite.

I know I cannot make LLMs more intelligent by prompting, they will always hallucinate and get some things wrong, but I want to at least get it to be less of an asshole about it.

So, what do you do? I followed some generic advice like using positive instructions instead of negations, using examples, but I cannot get the corpo slop tone out of it.

General advice is welcome, but I'm mostly looking for specific verbatim examples of system prompts that you use.

Comments URL: https://news.ycombinator.com/item?id=47613671

Points: 4

# Comments: 0

New comment by basilikum in "Show HN: Git bayesect – Bayesian Git bisection for non-deterministic bugs"

basilikum — Thu, 02 Apr 2026 12:05:52 +0000

Please stop spamming HN wih AI slop.

New comment by basilikum in "Ask HN: Who wants to be hired? (April 2026)"

basilikum — Thu, 02 Apr 2026 12:04:03 +0000

I found your comment here from another comment of you elsewhere.

HackerNews is not a place for you to puke your AI slop into. Stop doing that. It's against the site rules.

I get trying to find a (better) job and doin this on HN in a threat like this one is fine, but please contribute something of value otherwise instead of just spamming LLM slop.

New comment by basilikum in "Gone (Almost) Phishin'"

basilikum — Thu, 02 Apr 2026 11:33:39 +0000

It is unfortunately normal for companies to impersonate scammers.

We can teach people as much as we want about security against phishing. It won't matter because people have to break these rules constantly. Companies actively train people to fall for phishing by doing everything in their power to be indistinguishable from phishing themselves.

New comment by basilikum in "Voyager 1 runs on 69 KB of memory and an 8-track tape recorder"

basilikum — Sun, 29 Mar 2026 21:57:52 +0000

Our comments don't really contradict each other. The page size without any linked documents like an external style sheet grew to 140KB after your comment. But just the text is 30KB.

New comment by basilikum in "Voyager 1 runs on 69 KB of memory and an 8-track tape recorder"

basilikum — Sun, 29 Mar 2026 20:04:57 +0000

  curl https://news.ycombinator.com/item?id=47564421#47564679 | wc -c

143927

  curl https://news.ycombinator.com/item?id=47564421#47564679 | pup -p --charset utf8 'text{}' | wc -c

30954

New comment by basilikum in "Folk are getting dangerously attached to AI that always tells them they're right"

basilikum — Sat, 28 Mar 2026 20:25:49 +0000

It's exactly what Carnegie says not to do.

New comment by basilikum in "Treason in the Futures Markets"

basilikum — Sat, 28 Mar 2026 12:26:21 +0000

It's you. Every trader who does not have the insider information loses. That's how markets work after all. They collect information by rewarding the use of information. Anyone who has information and uses it is rewarded and anyone who does not is punished.

Even when you are a passive investor you lose. You essentially buy shares at random points in time. When that point happens to fall between the trading of an insider and the public disclosure of the insider information you will get a worse price for that trade.

It isn't even relevant whether the insider buys stocks or other securities directly or trades in futures instead. All information you enter into the market through trades permeates the whole market through arbitrage regardless of where you enter that information.

New comment by basilikum in "Make macOS consistently bad unironically"

basilikum — Sat, 28 Mar 2026 10:37:06 +0000

Have you tried disabling animations in the System Settings? Some apps respond to that.

New comment by basilikum in "21,864 Yugoslavian .yu domains"

basilikum — Sat, 28 Mar 2026 04:24:33 +0000

certificate transparency logs will give you a lot of that data.

New comment by basilikum in "AMD's Ryzen 9 9950X3D2 Dual Edition crams 208MB of cache into a single chip"

basilikum — Sat, 28 Mar 2026 04:01:28 +0000

Well, yeah, reality strikes again. All you need is an exploit in the microcode to gain access to AMD's equivalent to the ME and now you can just map the cache as memory directly. Maybe. Can microcode do this or is there still hardware that cannot be overcome by the black magic of CPU microcode?

New comment by basilikum in "AMD's Ryzen 9 9950X3D2 Dual Edition crams 208MB of cache into a single chip"

basilikum — Sat, 28 Mar 2026 03:54:23 +0000

KolibriOS would fit in there, even with the data in memory. You cannot load it into the cache directly, but when the cache capacity is larger than all the data you read there should be no cache eviction and the OS and all data should end up in the cache more or less entirely. In other words it should be really, really fast, which KolibriOS already is to begin with.

Consumer Rights Wiki

basilikum — Sat, 28 Mar 2026 03:19:23 +0000

Article URL: https://consumerrights.wiki/w/Main_Page

Comments URL: https://news.ycombinator.com/item?id=47551253

Points: 3

# Comments: 0

KolibriOS – tiny, feature rich OS written in x86 assembly

basilikum — Thu, 26 Mar 2026 19:54:56 +0000

Article URL: https://www.kolibrios.org/en/

Comments URL: https://news.ycombinator.com/item?id=47534922

Points: 4

# Comments: 1