/informatics/

Towards a supercoordination protocol
Inspired by >>2513 and >>2589 I have been thinking about what a software implementation of a supercoordination protocol might look like. Specifically, this is based on the signal/impact model presented in >>2570. I would be interested to hear what other an... posted 2d ago with 3 replies technology informatics

Towards a supercoord

The way I see it with these supercoordination systems, the hard part besides actually coming up with a good protocol is the right bootstrap network and the right use cases. Data entry is annoying unless it is made fun and rewarding and part of some activit... 9h ago

The way I see it wit

Designing a word frequency auto-tagger for sofiechan
Let's talk about the upcoming auto-tagger architecture. All it has to do is produce affinity scores between threads and tags, and our now existent smart tag selector will select a high-affinity set of tags that also make a good index. To do this, it's goin... posted 7d ago with 3 replies technology meta informatics

Designing a word fre

It works! But slow as all hell. Right now we fully rebuild the "taste machine", which now includes this word frequency tagger, about once a minute. That used to do 50 iterations of the expectation-maximization algorithm we use to refine everyone's tastes, ... 4d ago

It works! But slow a

A Primer on E-Graphs (A technique to control combinatorial blowup in term rewriting systems)
About 10 years ago I was very interested in term rewriting as a basis for exotic programming languages and even AI. One of the big problems in term rewriting is that without a canonical deterministic ordering, you rapidly end up with an uncontrollable numb... (www.cole-k.com) posted 3w ago with no replies technology intelligence informatics

A Primer on E-Graphs

Radiopaper
Radiopaper is an interesting data point in the design space of social media. They've put some thought into it and landed someplace very different from Sofiechan. Their goal is to "improve conversation quality," which sounds goods, but they do it buy leanin... (radiopaper.com) posted 4w ago with 2 replies technology informatics

Radiopaper

interesting indeed. They seem to be building on the substack idea of bringing money into it. I think this results in slop incentives. My usual argument against that is this: High quality thought and speech is a luxury activity for the idle rich and ascetic... 4w ago

interesting indeed.

Why Momentum Really Works. The math of gradient descent with momentum.
https://distill.pub/2017/momentum/ posted 13mo ago with 3 replies technology intelligence informatics

Why Momentum Really

Do you know if there is any predictability to the effect? I would guess it's related to catastrophic forgetting aka the reason stochastic gradient descent has to be stochastic. Basically if you update on one set of evidence without locking in those learnin... 13mo ago

Do you know if there

A pinpoint brain with less than a million neurons, somehow capable of mammalian-level problem-solving.
https://rifters.com/real/2009/01/iterating-towards-bethlehem.html?fbclid=IwAR1b9QURSJnizgy7r4HD7UCYmc06A7uL5muc7igz8uJHxLAIZWEwDMSqyPk posted 2y ago with 15 replies technology philosophy informatics

A pinpoint brain wit

Aristotle described animals as having souls* capable of knowledge derived from the senses and the capability to use that knowledge to support the movements characteristic of their species.... 14mo ago

Aristotle described

Higher-order Virtual Machine (HVM). New functional programming VM based on Interaction Nets.
https://github.com/HigherOrderCO/HVM/tree/master posted 1y ago with 4 replies technology informatics

Higher-order Virtual

Yeah Bend is the language built on the HVM. I haven't looked into it yet but I'd love to try it out when I get a chance. The other one I've recently been wanting to check out is Zig. It really does look like C with the problems fixed. But I have no great p... 14mo ago

Yeah Bend is the lan

Kolmogorov–Arnold Networks, a new architecture for deep learning.
https://github.com/KindXiaoming/pykan posted 1y ago with 1 reply technology intelligence informatics

Kolmogorov–Arnold Ne

This is neat. Instead of having fixed nonlinear activation functions, they have trainable nonlinear functions instead of weights. They claim it has better scalaing properties and performance for many problems, and is more interpretable (because there are f... 1y ago

This is neat. Instea

Hundred Rabbits' "UXN" ultra-portable personal computing stack
They made a simple VM that can be implemented in a weekend by almost anyone that can run their whole suite of personal computing, artistic, gaming, music, creative, etc apps. This is one of the coolest projects out there and it proves that computing could ... (100r.co) posted 2y ago with 15 replies technology lifestyle informatics

Hundred Rabbits' "UX

see also on the general topic of minimalist computing: Suckless... 2y ago

see also on the gene