Juraj's avatar
Juraj
juraj@tamersofentropy.net
npub1m2mv...r8p9
I don’t seek rigid structure — I seek resonance Vibe coding, reality bending, cypherpunk visions. Author of Tamers of Entropy: https://tamersofentropy.net/ I like teaching, get my books and courses here: https://hackyourself.io/shop https://juraj.bednar.io/shop (You'll learn skills no one else is teaching!) Podcasts 🎙️: Option Plus - https://optionplus.io/ Reči o živote, vesmíre a vôbec: https://juraj.bednar.io/reci-o-zivote/ Ako vyhackovať otcovstvo: https://otcovia.com/
Juraj's avatar
Juraj 1 month ago
Greatest moment on the mesh yesterday. Yes, it was very lively during the day on bitchat, but it ended up being me and @mariannasadecka on the mesh while she was singing (with 4xSuper) and I was sending her photos of her singing (photo that I meshed @No Good Kid ). Two people being massaged on the stage. @Bitcoin je retro was the most Lunarpunk thing that I ever experienced in my life. image
Juraj's avatar
Juraj 1 month ago
Hot tip: If you have shitloads of subscriptions and you don't need them anymore, tell your AI agent to find a way to migrate away. Get a @Start9, nextcloud, or whatever, let agents write migration scripts and verify it all went out. Cancel all the corporate bullshit. Before, the cost for subscriptions was much lower than cost (mental cost, time) to migrate. Not anymore. Get free. I'm migrating.
Juraj's avatar
Juraj 1 month ago
How do I know that my talk on Saturday at @Bitcoin je retro will be good? When putting it together with Claude, it kept saying something about some legalese and terms of service 😃 Switched to a chinese model (Kimi K2.6), it was way less censored. I don't know exactly what it did not like, I just gave it my research, dumped all my Nostr posts for the last year and asked it if I did not forget anything important. And I have! So it will be even better. Yes, I will make all you beautiful people a recording.
Juraj's avatar
Juraj 1 month ago
My PulseLibre app for vagus nerve stimulator is on Zapstore! Get your Pulsetto here (you will support app development): https://juraj.bednar.io/pulsetto Beware, you have to cancel trial right after ordering the device, because they will bill you for the premium app automatically after 14 days! Get the Fit version, it fits better. And yes, it works, measured on HRV. Good thing if you are chronically stressed and need immune or gut support (both are greatly modulated by the vagus nerve).
Juraj's avatar
Juraj 1 month ago
This weekend, there is @Bitcoin je retro in Bratislava (many English language talks!). It's very cool. Also, there will be a Bitcoin sauna tomorrow, with singing and whisks, etc. Tickets on the same site. "Bitcoin is retro" means Bitcoin is now almost old enough to vote, drive and drink :). I am teasing my novel Tamers of entropy, which starts in 2003, when the protagonists were sketching the same sovereign stack out of necessity — mesh networking, encrypted communication, bearer money, no single point of failure. Twenty+ years later the tools exist and the characters are compounding them for something magical. We're catching up to the vision. That is what my talk at Bitcoin je retro is about - did we catch up to the vision? (Spoiler alert: We are way beyond old cypherpunk dreams. Cypherpunk future is now). If you want to know when you can preorder Tamers of entropy, sign-up here (choose EN, SK or CZ language - and yes, you can put an npub there, @Tamers of Entropy are Nostr native):
Juraj's avatar
Juraj 1 month ago
Picking a Qwen3.6 model in Ollama? More parameters ≠ better model. Architecture matters. qwen3.6:27b is a dense model — every token goes through the entire network. Slower, but higher-quality output. qwen3.6:35b-a3b is MoE (mixture of experts): a router picks a few experts, and only ~3B of the 35B parameters are active per token. Much faster, usually lower quality. You still need all 35B loaded into RAM. The benchmarks back this up: Qwen claims the 27B dense beats their own previous flagship 397B-A17B MoE on agentic coding. Quantization is rounding the weights (4-bit, 8-bit, …). Smaller file, faster inference, mild quality loss. Running unquantized is usually slower because of memory-bandwidth overhead. MLX is Apple Silicon–native inference, the fastest path on a Mac. In Ollama, though, the MLX builds of Qwen3.6 are text-only for now — vision isn't wired up on Ollama's MLX path yet, and Unsloth's vision-capable builds don't currently work on Ollama either. If you need images, you're stuck waiting, or going outside Ollama (LM Studio, mlx-vlm). TL;DR: dense is better for quality, MoE for speed, quantization almost always yes, MLX on a Mac — but in Ollama, no images yet. Pick here:
Juraj's avatar
Juraj 1 month ago
Dan Sullivan napísal Free Zone Frontier pre podnikateľov. Netušil, že píše operačný manuál pre cypherpunkerov. Bitcoin, Aave, @Liberation Travel — všetko podobný scenár: úskok z mainstream hry do takej, ktorú regulátori nevedia rozoznať. Spolupracuj s niekým mimo svojho odboru a úplne unikneš konkurencii. A každá Free Zone má dátum exspirácie. image
Juraj's avatar
Juraj 1 month ago
Dan Sullivan wrote Free Zone Frontier for entrepreneurs. He didn't realize he was writing an operations manual for cypherpunks. Bitcoin, Aave, Liberation.travel — all the same move: step sideways into a game regulators can't see. Cooperate with someone out of your field and completely escape competition. And every Free Zone has an expiry date. image
Juraj's avatar
Juraj 2 months ago
Just for people who think how people are moving to the little country of Paraguay that noone knows about... @Liberation Travel
Juraj's avatar
Juraj 2 months ago
"They" are doing COVID dystopia, "they" are batshit crazy warmongers but somehow "we" are going to the moon. Nope sisters and brothers, you are not going to the moon. At best, you're just paying for that.
Juraj's avatar
Juraj 2 months ago
GM! Today I realized that people in the walled gardens can't even say GM anymore. Everything breaking for them. So GM to you all!