Sign in

Dziban

@dzi-ban.bsky.social
50 followers 286 following 174 posts

Working on Katarineko: katarineko.com a platform to help immerse in Japanese I play ( #GoGame #weiqi #baduk) At @posthog.com

PostsRepliesMedia
Dziban @dzi-ban.bsky.social · 07/10/2026
A persistent chat in a fistful of lines #Forth -like languages are cool
010 0
Dziban @dzi-ban.bsky.social · 06/10/2026
Step 1 of becoming a #mathematician done. What's next? #math
A box of Hagoromo chalk
040 0
Dziban @dzi-ban.bsky.social · 03/10/2026
exactly, the best compression technique is understanding.
000 0
Dziban @dzi-ban.bsky.social · 01/10/2026
I need to start a similar journey soon, not so extreme. Any resources that helped you? Particularly nutrition-wise?
000 0
Dziban @dzi-ban.bsky.social · 30/09/2026
I don't think they took into account that every different scrolling or animation state is a whole new different visualization. So that's the effect they are going for, but it just fucking sucks on 99% of states.
010 0
Dziban @dzi-ban.bsky.social · 30/09/2026
Feel free to hit me up with any question, problem or concern. And you can use EARLYBIRD for a 50% discount on any plan!
000 0
Dziban @dzi-ban.bsky.social · 30/09/2026
I've redesigned Katarineko from the ground up, improved story generation naturalness, added listening and listening reviews and worked over a couple of weeks to make the whole thing spark joy. I've put quite a bit of love into this release and I hope you find it useful. #leanjapanese #日本語
102 0
Dziban @dzi-ban.bsky.social · 25/09/2026
this is deeply idiotic, both because noone gives any kind of a fuck about Urbit and because it's just Linux, so you could already install it there. So it's just a circlejerk between assholes.
020 0
Dziban @dzi-ban.bsky.social · 22/09/2026
Experimental actor-based OS with capability based access to resources. Experiment into mixed agentic + human interaction at the OS level. tangled.org/dzi-ban.bsky...
tangled.org
dzi-ban.bsky.social/KokoroOS
A repository on Tangled
000 0
Dziban @dzi-ban.bsky.social · 13/09/2026
www.youtube.com/watch?v=YFiQ...
youtube.com
Particle Physics in 17 Hours
YouTube video by Kamil Laurent
000 0
Dziban @dzi-ban.bsky.social · 13/09/2026
012 0
Dziban @dzi-ban.bsky.social · 11/09/2026
Voices in my head are telling me to buy these @acollierastro.bsky.social
000 0
Reposted by Dziban
Steve Klabnik @steveklabnik.com · 10/09/2026
We've been cooking at @ersc.io "What comes after git" ersc.io/blog/what-co...
ersc.io
What comes after git
Our thoughts about the future of version control
1626047 0
Dziban @dzi-ban.bsky.social · 10/09/2026
Bars yes! But when I have no signal it becomes dots! I only use one SIM, so I may not be used to this
000 0
Dziban @dzi-ban.bsky.social · 10/09/2026
That's nice, but I'd have to rewire my brain so a dot now means signal, when until now it mean no signal
100 0
Dziban @dzi-ban.bsky.social · 10/09/2026
Avian Intensity
100 0
Dziban @dzi-ban.bsky.social · 09/09/2026
AGI is truly here. 2 years ago models didn't do what you wanted because they forgot. Now models don't do what you want because they ignore you on purpose.
050 0
Dziban @dzi-ban.bsky.social · 09/09/2026
The first few notes of Kyorei played on my Shakuhachi physical model synthesizer
030 0
Dziban @dzi-ban.bsky.social · 09/09/2026
AI vs humans? or together?
110 0
Dziban @dzi-ban.bsky.social · 09/09/2026
Created a demo for the on-going work of the Shakuhachi physical model synth. You can play with it here: shakuhachi.dziban.net
shakuhachi.dziban.net
尺八 · Shakuhachi
000 0
Dziban @dzi-ban.bsky.social · 09/09/2026
It's starting to sound okay, better than the other models achieved tangled.org/dzi-ban.bsky...
tangled.org
dzi-ban.bsky.social/neiro
A shakuhachi physical model synthesizer
000 0
Dziban @dzi-ban.bsky.social · 08/09/2026
diffusion models are very interesting to me. Imagine type-guided diffusion
011 0
Reposted by Dziban
cafkafk @cafkafk.bsky.social · 08/09/2026
I stole a video from twitter for you chat
410510 0
Dziban @dzi-ban.bsky.social · 08/09/2026
how did you get your visa?
030 0
Dziban @dzi-ban.bsky.social · 08/09/2026
This was not a one-shot btw, I did quite a bit of iteration with it, explaining what I didn't like, showing new examples. Lots of iteration.
010 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I did create some concept art with ChatGPT
110 0
Dziban @dzi-ban.bsky.social · 08/09/2026
It looked so whimsical and magical, like from an alternative age of sail
000 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I was def inspired by what I saw at the Museo Naval! The right one is a Flemish Galeon and it kickstarted the idea for the game.
200 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I cannot see where I got the next piece of my idea (that's inference) but I can see the whole cohesive stream of thought (the full token stream). Now, I'm not saying this is how our brain works! Just trying to exemplify how the multiple levels can be seen without thinking they are memoryless.
110 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I'm going to use a very simplistic analogy, please excuse me. This is not an endorsement of a scientific idea. Just a small inarticulate example of where my thought is going. Imagine the token stream as the thought the human sees, and next token inference as a subconscious process for the next word
100 0
Dziban @dzi-ban.bsky.social · 08/09/2026
Yeah, at this point I'm tunnel visioned into the conversation and I need to take a step back and see it in perspective in a few hours. Nonetheless, this was a nice exchange in good faith. So I'm happy!
010 0
Dziban @dzi-ban.bsky.social · 08/09/2026
It also makes me wonder (I have no idea), if we also create stateful memories by a process of stateless mechanisms. Something I'll research later!
010 0
Dziban @dzi-ban.bsky.social · 08/09/2026
> Every new token has to start completely fresh without any indication of where the previous tokens where going That's true, but I think that can be seen from a different PoV. The generated tokens *is* the memory, while there is a stateless process that drives what to remember/produce next.
200 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I think it's a good exercise also in case we ever find extraterrestrial life, because they may also have very different ways of reasoning that do not depend on the properties you are saying.
220 0
Dziban @dzi-ban.bsky.social · 08/09/2026
Now, I'm not claiming that they reason now, although tbh I wouldn't be very skeptical of it either. What I'm saying is that a lot of this philosophical thought could probably not even contemplate what we have right now and I think we need to rethink what reasoning means to adapt to other substrates.
130 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I think there is an overfitting of the concept of "reasoning" to existing human biology. A lot of what you are talking about are mechanisms derived from how Humans operate. I think the different substrate from LLMs opens the question to a different abstraction to what "reasoning" means.
101 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I get where you are coming from, I really do and I can up to a point see what you see. I will try to answer each of these more in-depth later and a deeper perspective into what I'm talking about but to summarize my position:
100 0
Dziban @dzi-ban.bsky.social · 08/09/2026
Also some ideas around programming language and platforms: dziban.net/posts/yabu/
dziban.net
Yabu - 藪 — dziban.net
This language is an exploration in many fronts, my personal programming language aesthetics, a programming language that fits how I think, and a programming language to attempt to reduce the unreliabi...
000 0
Reposted by Dziban
Jim Calabro ♟️ @calabro.io · 08/09/2026
I do think the industry is long overdue for some new ideas in really ripe foundational tech Don't just rewrite it for the sake of rewriting, but I want people to bring new ideas to OSes DBs Browsers Languages GUI tooling Network software Etc...
7161 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I've been experimenting a bit on the OS side of things in case something resonates! bsky.app/profile/dzi-...
010 0
Dziban @dzi-ban.bsky.social · 08/09/2026
Shakuhachi produces some of the most beautiful sounds on earth www.youtube.com/watch?v=miUK...
youtube.com
Shakuhachi [The Japanese Flute] - Kohachiro Miyata (Full)
YouTube video by Pássaro Trovão - Sons Ancestrais
000 0
Reposted by Dziban
Grace @gracekind.net · 08/09/2026
👀
Two proposals were offered to me. The first was that we post our Euler result, and that OpenAI post its Navier-Stokes result the next day. The second was that, after posting Euler, I alone write a paper presenting the Navier-Stokes result, acknowledging that an internal OpenAI model had resolved it. Sebastien twice asserted that he wanted Levent removed from authorship, and said it would all be simple if only it were not the case that, and it was so annoying that, Levent works at Anthropic. It was also said that if OpenAI posted after us, they would say that we deserved the Clay Prize, and that we were the "closest humans to the problem". I declined both offers.
I said that if OpenAI released its result in the way proposed I would go public with what happened. The reply was, "Why would you ruin your career?" I replied that I am an academic, and asked why he thought going public would ruin my career. The reply was,
"If you don't want me to be nice, then I don't
have to be nice."
1035343 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I’d like to avoid defining reasoning as biological as that would settle the question by definition. But if you think biology supplies some necessary property, I’m interested in what that is and why it couldn’t be realised another way.
100 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I agree that the process matters. What specific properties do you think it needs to count as reasoning, and how would we determine whether an LLM has them? That seems like the interesting question. I’d also discuss reasoning separately from sentience.
100 0
Dziban @dzi-ban.bsky.social · 08/09/2026
Yes, that’s fair: similar outputs alone don’t establish that the underlying process constitutes reasoning. My objection was narrower. Describing that process as mathematical doesn’t establish that it cannot constitute reasoning, either.
100 0
Dziban @dzi-ban.bsky.social · 08/09/2026
Also assuming that the same problem under different substrates has the same complexity requirements or the same shape makes no sense to me.
000 0
Dziban @dzi-ban.bsky.social · 08/09/2026
So if they both do the same work but one is much easier to understand or reason about this should be exciting because it makes the problem easier to tackle and explore. To be clear I don't think the brain and LLMs are doing the same work. But I think they are on a subset of tasks.
130 0
Dziban @dzi-ban.bsky.social · 08/09/2026
I think there is also always a comparison problem in the arguments. They think that to arrive at the same features you need everything else around it, but I don't think that's true. I can have two complex machines, one MUCH more complex than the other and both doing the same work.
110 0
Dziban @dzi-ban.bsky.social · 08/09/2026
But that's like claiming that you understand the human brain because you have the model of a neuron. It's the complex interrelation of simple composable parts that creates the emergent behavior.
2110 0