So far I built a minimal tech integration so far, just pipewire moving a triangle in Vulkan. It's just a proof-of-life.

Modern Beat Detection

The tech I'm going to going to bring to the table is some ML for beat detection using a technique that doesn't require back-propagation training. I have some experience with such systems from years ago and it has been coming back to life for me while characterizing LLMs and diffusion models.

The point is to develop extremely small, real-time AIs in the open source, aimed at the very forgiving problem of music visualization, where behaviors like hallucination are a feature rather than a bug. We can cut training time and runtime costs while also have a faster feedback loop and no requirement for differentiable forward architectures.

The Carrot for PrizeForge

I'm developing µTate (Mu Tate) in order to provide a concrete motivation for consumers (not us) to sign up for PrizeForge, where I'm developing new social finance & social decision models.

Most open source projects have pretty bad monetization. Contributing on someone's project is normally just scratching our own itches. For µTate, the fund raising can flow to anyone who works on the project. I run PrizeForge, but the point is to leave control with the user community, not an individual repo owner or campaign creator. A funding stream on PrizeForge is more like a foundation that can direct funds at what they think will benefit the backers.

This is Viable

µTate is very clearly a consumer facing product. Social finance has applications well beyond open source software. 2D fund raising creates meaningful coordination between businesses and groups of consumers to fund tech and products they both rely on. Our social decision systems are designed to overcome the accountability problems that plagued Kickstarter and others. I know we can do a lot of good.

If you are interested in social decision and social finance problems, I'm going to be recruiting co-founders. Everything we use is in Rust. If you like FFTs and dynamic programming, you will probably have fun implementing crowd cognition (our social decision system) with me. I have a stack of business cards from investors who just wanted to know how we will pull consumers onto the platform, and µTate can do that.

1 post - 1 participant

Read full topic