Preserving the code we write today

The GitHub Archive Program, first announced at GitHub Universe 2019, has moved from planning into production. The Arctic Code Vault — an archive designed to preserve open source software for a thousand years — is now being built. On February 2, 2020, GitHub captured a snapshot of all active public repositories to be stored in the vault.

Which repositories made the cut

The snapshot includes repositories meeting any of the following criteria:

  • Any commits made between the Universe announcement on November 13, 2019, and February 2, 2020
  • At least one star and any commits from the year prior to the snapshot, between February 3, 2019, and February 2, 2020
  • At least 250 stars, regardless of when the repository was last active

A guide for future readers

Every archive includes a human-readable index and guide that itemizes the location of each repository and explains how to recover the data. The guide provides an overview of what software is, explains open source and its ethos, and gives a technical walkthrough for unpacking the archive's contents.

GitHub open sourced draft 0.1 of the guide on January 23. Contributions are welcome via pull request in the GitHub Archive Program repository until February 29, 2020.

Expert input on maximizing value

GitHub convened an advisory board with expertise in anthropology, archaeology, archiving, history, linguistics, science, and long-term projects. At the first Advisory Summit, held January 16–17, the board identified three key recommendations:

  • Visualization: Embed visual representations within the archive content, such as pairing scientific plotting code with its output, and make the physical artifacts both visually striking and aesthetically appealing.
  • Metadata: Include repository metadata (description, language, commit logs, associated wiki, and similar) along with broader context such as recent State of the Octoverse reports and a Wikipedia snapshot.
  • Redundancy: Create smaller "fractional" deposits containing the 10,000 most-starred and most-depended-upon repositories plus a small random sample of others. Copies of these deposits will also be donated to prominent archives and libraries worldwide, including Oxford's Bodleian Library.

Production timeline and next steps

Production of the Arctic Code Vault will take roughly two months. In spring 2020, GitHub will return to Svalbard to make the official deposit in the Arctic World Archive. Further details about the Archive Program are expected at GitHub Satellite in May 2020.