Hello HN! Divvi Up is a system for privacy preserving telemetry. That may sound like a misnomer but the idea behind it is pretty simple: split telemetric data from users into two parts and let two different non-colluding servers summarize a collection of data from a set of users independently. Then do one final summarization of the parts created by the server to generate the final report. At the end of this process Divvi Up user gets the aggregate data they care about and no no single server gets any original value from the user. Confused? We have an in-depth explanation on our docs[0].
Anyways, we have had the system in production users for a while. But, it has been a bit hard to get started with the project. And Show HN is all about stuff you can try out. So...
Today, we are happy to say that Divvi Up can now easily run under `docker compose` and we have built a command line tool, `divviup`, that can be used to exercise most of the system without using a client library.
You can read the docs on how to do that over here:
https://docs.divviup.org/command-line-tutorial
Or if you prefer watch a narrated video:
https://docs.divviup.org/command-line-video-demo
The project was built by ISRG (the Let's Encrypt folks) and evolved from the COVID-19 Exposure Notification system that Apple/Google worked on[1]. And, since the exposure system was discontinued Divvi Up has found other use cases in application metrics, federated machine learning, and survey response processing.
Oh, and also all components of the Divvi Up system are open source and available at https://github.com/divviup. Further, the core protocol, Distributed Aggregation Protocol (DAP)], is on a path to become an Internet Engineering Task Force (IETF) standard co-developed by Internet Security Research Group (ISRG, best known known for Let's Encrypt), Cloudflare, and Mozilla[3].
We feel this all a very promising technology to help solve telemetry privacy. But, it still feels like early days and we would love your feedback. Thanks!
[0]: https://docs.divviup.org/how-it-works
[1]: https://covid19-static.cdn-apple.com/applications/covid19/cu...
[2]: https://divviup.org/blog/prio-services-for-covid-en/
[3]: https://github.com/ietf-wg-ppm/draft-ietf-ppm-dap
