HACKER Q&A
📣 xdgrulez

HSL 0.1 – The Human Source License. Please help refining


As a side project, I am building Kafi Streams, a new stream processing library for Python based, partly, on Kafka Streams, but using the novel (Py)DBSP (Database Stream Processing) engine under the covers. DBSP gives you "database semantics" for truly incremental stream processing - accessible for anybody, not just the few deep down stream processing experts.

I was thinking about how to license Kafi Streams. How I could build something exciting but not just give it away for free to the IaaS/SaaS vendors like AWS, and the OpenAIs and Anthropics of this world.

What came out of this thinking is a new software license called "Human Source License" (HSL) (https://github.com/xdgulez/human-source-license).

Why "human"? The license is Open Source for every human being who'd like to use it in their software or the software they build for their company.

But there is a red line. As soon as the software is either used inside a managed service e.g. by AWS, or, more importantly now, for training AI models, the license bites: If the license is violated, I get back my share. And more. How about 10% of the entire yearly revenue of OpenAI?

Imagine a lot of important Open Source projects had a license like this. Then scraping GitHub for new training data would actually become a minefield for AI companies. A minefield that would maybe sway the world to become a little bit fairer after all.

What I am asking you is to help me work on finalizing the license and maybe make it viral.


  👤 eimrine Accepted Answer ✓
There is no minefield. Ask your favorite LLM is it bad to read torrented books. Then ask it whether it did this. Or try to set up robots.txt file on your website than see what happens.

It is even legal for them to "clear room" rewriting some GPL software into a permissive license.

What person do you consider will protect your license in front of Big Tech, wasn't he the one named Donny?