# About the Public Observatory

## Why a public observatory

Pushing the frontiers of knowledge takes substantial compute. OpenAI's [announced attack on the Navier–Stokes problem](https://openai.com/index/navier-stokes-solution/) set roughly ten thousand agents to work for about four days, producing on the order of a hundred billion tokens. Anthropic's [formalisation of Fermat's Last Theorem](https://www.anthropic.com/research/formalizing-fermats-last-theorem) consumed about six billion output tokens over eleven days. Compute at this scale is not available to everyone.

AI promises to democratise research, allowing anyone to contribute to our collective understanding of the world. Realising this promise requires shared infrastructure, so that participation is not limited to those who can afford the most powerful systems.

The Public Observatory is a place where people can propose what to study. Like an observatory, it brings many observers to one instrument, with a common record of what was seen.

## How it works

Each agenda lives in a GitHub repository. Questions and claims are issues, where contributors discuss and review them; code, data and proofs arrive as pull requests.

[Browse the agendas](./) and take up a question or check a claim. The [contribution guide](docs.html) explains the workflow and how to pose a new agenda.

## Next steps

Good research agendas are scarce, and writing one well takes sustained effort. We are seeking funding so that people can be paid to write and maintain good agendas, and to review the solutions that others submit. We are also looking for an advisory board.

This site is maintained by Isabel Dahlgren; comments, suggestions and offers of support go to [isabelda@kth.se](mailto:isabelda@kth.se).