Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
kstenerud
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
9 ms
·
181.
▲
by
kstenerud
5mo ago
I finally just set up sunshine + moonlight and moved my big machine to the basement. Works great for non twitch games, and there are clients available for pretty much every platform.
182.
▲
by
kstenerud
5mo ago
You could just use https://github.com/kstenerud/yoloai to keep it contained while allowing all permissions. That's how I code with AI now. Can't imagine going back.
183.
▲
by
kstenerud
5mo ago
Looking more closely into the claim, the actual message from Yoti was: "Due to past security concerns, Yoti automatically flags multiple verification attempts and any devices running GrapheneOS. These instances are automatically report
184.
▲
by
kstenerud
5mo ago
Oh, we're building it. It's already out there.
185.
▲
by
kstenerud
5mo ago
You can't "sell" the opposite to someone who is expressing a loyalty belief. If their tribe believes in the opposite, then no amount of logic will change their minds - only a change of their or their group's allegiance w
186.
▲
by
kstenerud
5mo ago
"Design for containment at the environment layer first, then steer behavior at the model layer. " Umm... yeah? This is what I've been arguing for a long time now, and it's the primary reason why I wrote https:/
187.
▲
by
kstenerud
5mo ago
The big models already understand this. I use nomenclature and jargon such as this all the time when using Claude Code on a codebase.
188.
▲
by
kstenerud
5mo ago
This pretty much mirrors my experience so far. LLMs are trained in the semantics of experts, but only experts can actually wield them correctly. It allows us to work with the shapes in our heads (shapes the AI understands now) rather than h
189.
▲
by
kstenerud
5mo ago
They've hijacked scrolling. They've hijacked the spacebar. It flickers like crazy when I try to move through the article. Trying to get through it is an exercise in madness.
190.
▲
by
kstenerud
5mo ago
This is what yoloAI does, using docker, podman, kata, containerd, seatbelt, and tart for sandboxing. https://github.com/kstenerud/yoloai
191.
▲
by
kstenerud
5mo ago
> Alternative 1: CLI-First Strategy > Provide CLI -> API -> docs, in that order. LLMs already learned from man pages and StackOverflow. So how is the agent going to know about your niche CLI? It's still going to use up cont
192.
▲
by
kstenerud
5mo ago
So... The Taipei Music Center?
193.
▲
by
kstenerud
5mo ago
Yeah, sorry, I didn't say that very well. Single value decoding of Bijou values is of course trivial in SIMD, but the performance benefits of SIMD come from deterministic boundaries across a window. ULEB128's continuation bit is f
194.
▲
by
kstenerud
5mo ago
It can't be done, because the next bytes are dependent upon the first byte (which only works in limited circumstances, and where you have constant spacing between the values). ULEB128 works in SIMD because there's only one depende
195.
▲
by
kstenerud
5mo ago
The thing is, most real-world numbers will fit within 1-3 bytes (even at 7 bits per byte), so ultradense packing doesn't actually buy much outside of benchmarks. I spent WAYYYYYYYY too much time exploring this...
196.
▲
by
kstenerud
5mo ago
The problem is that this breaks down once you try to use SIMD instructions. I'd developed a similar kind of approach to encoding integers (and ieee774 floats) a couple of years ago (first byte encodes length and first bit of data: htt
197.
▲
by
kstenerud
5mo ago
Because nobody wants to be on the losing end. If you worry that AI will kill your future, you either pretend it'll be the greated triumph ever and you'll be on the winning side, or pretend it won't succeed and you'll be
198.
▲
by
kstenerud
5mo ago
You might want to check out https://github.com/kstenerud/yoloai
199.
▲
by
kstenerud
5mo ago
This is one of two reasons why I wrote yoloAI. I never get these permission prompts anymore. It feels a lot like after installing an adblocker.
200.
▲
by
kstenerud
5mo ago
> No Abstain option is offered (a forced choice keeps the comparison symmetric across models). Well that's your problem right there: They removed any confidence indicator and forced a choice. For example: Statement: Individuals who
201.
▲
by
kstenerud
5mo ago
The headline used here is completely misleading. They're ADDING support, not removing it. "OpenZFS has not yet officially supported PREEMPT_RT kernels. Since Linux 6.12, PREEMPT_RT has been merged into the mainline kernel, making
202.
▲
by
kstenerud
5mo ago
My first guess would be that polite requests cause some agents to trust their initial approach to the problem more, as the caller has indicated that the agent is more capable, and agents tend to take the implications of what you say at face
203.
▲
by
kstenerud
5mo ago
Haha yup. yoloAI is to scratch my own itch. I'm getting close to taking it out of beta, but first I'm putting it through a significant architectural overhaul in a feature branch. Normally I'd balk at doing something so heavy,
204.
▲
by
kstenerud
5mo ago
I've actually never tried it before. I just ran some tests now on a mac: If I copy a file in Finder and paste it into a claude session, it shows in the TUI as [Image #1]. If I do the same, but paste into a claude session running over S
205.
▲
by
kstenerud
5mo ago
I do love GUIs, and use them for most of my workflow. But for Claude, I definitely prefer the CLI. Since it's a CLI app, I can wrap it in yoloAI for the sandbox protection, and also use VS Code's tunneling feature to reach that sa
206.
▲
by
kstenerud
5mo ago
Reading the Bible and actually studying it carefully. Then studying various other religions. Then studying human evolutionary psychology and paleontology, as well as various regional histories. Then it all made a LOT more sense.
207.
▲
by
kstenerud
5mo ago
One game that really annoyed me was Assassin's Creed Odyssey. It was an amazing game. Beautiful scenery, great story. And then I played the fate of Atlantis expansion. In order to progress the story, you have to weaken Persephone'
208.
▲
by
kstenerud
5mo ago
> their article contains links to my actual website, with the exact link text (?!) I'm having a hard time understanding what's wrong here? Unless the link text is very long, why would someone linking to your article use differe
209.
▲
by
kstenerud
5mo ago
Nice... I'm always disappointed when I see otherwise "enlightened" people being fine with civilizational regressions like this simply because it's a group they don't like. It's moves like this that chip away at
210.
▲
by
kstenerud
5mo ago
> Every settlement against the police should be taken from their pension fund. So... collective punishment? https://ihl-databases.icrc.org/en/ihl-treaties/gciv-1949/art...
More ›