A year of GitHub Innovation Graph data

This first 2025 release of the GitHub Innovation Graph also marks the first data release since the project's first anniversary. Looking back at the past year-plus, the dataset has found steady uptake among researchers, policymakers, and developers—though measuring that success is itself a challenge. As a rough proxy, the github/innovationgraph repository's star history shows steady growth, with increases clustering around open source-related events:

A line chart of the cumulative star count of the github/innovationgraph repo over time, starting from September 2023 through January 2025. The number of stars increased rapidly in the two weeks following its launch, then grew steadily (but more slowly) for the subsequent year and a quarter. The line chart is annotated with various events, including from conferences, data releases, academic paper publications, and news articles.

Release cadence and versioning

Since launch, five additional quarters of data have been released through the releases page. Each release brings a modest bump in stars, but a recent working paper on semantic versioning and software adoption suggests the project may be leaving "stars on the table" by only incrementing the patch version with each dataset update—the research indicates that new major releases are associated with significantly more adoption growth than minor or patch releases.

Notable research using the data

The following papers stood out for aligning with open source research questions on our wishlist, or for examining AI's effect on software production—often both.

  • The Value of Open Source Software. This paper estimates the demand-side value of open source software at $8.8 trillion. (Hoffmann, Nagle, and Zhou, Harvard Business School Strategy Unit Working Paper No. 24-038, SSRN)
  • Generative AI and the Nature of Work. Access to GitHub Copilot was shown to shift open source maintainers' efforts toward coding and away from project management, while also encouraging exploration of more lucrative programming languages. See our Q&A with co-authors. (Hoffmann, Boysel, Nagle, Peng, and Xu, Working Paper No. 25-021, SSRN)
  • Measuring Software Innovation with Open Source Software Development Data. Major OSS package releases can be treated as innovation units alongside scientific publications, patents, and standards—useful for policymakers, managers, and researchers. (Brown et al., arXiv:2411.05087)
  • Open Source Software Policy in Industry Equilibrium. Simulations of China and US government restrictions, disincentives, and subsidies on OSS contribution found that restrictions don't boost domestic investment, disincentives raise costs domestically and globally, and subsidies increase domestic investment while lowering global costs. (Gortmaker, working paper)
  • Impact of the Availability of ChatGPT on Software Development. ChatGPT's availability led to more Git pushes per 100,000 inhabitants and greater developer engagement in high-level languages like Python and JavaScript, while effects on domain-specific languages like HTML and SQL were mixed. See our Q&A with co-authors. (Quispe and Grijalba, arXiv:2406.11046)
  • From GitHub to GDP. This work develops a framework for estimating OSS investment consistent with US national accounting methods, putting 2019 US investment in OSS at $37.8 billion with a current-cost net stock of $74.3 billion. (Korkmaz et al., Research Policy, doi:10.1016/j.respol.2024.104954)

Conference and media pickups

Innovation Graph data appeared in presentations at too many conferences to list fully, including the OpenForum Academy Symposium (2023 and 2024), NBER Summer Institute 2024, AI at Wharton's 2nd Annual Business & Generative AI Workshop, Tilburg University's Conference of the Workplace of the Future, and State of Open Con 24. Many of the papers above were also presented at these gatherings.

On the journalism side, The Economist and Rest of World both used Innovation Graph data in 2024 coverage. We're considering adding year-over-year percentage growth charts to the Innovation Graph site to lighten the load for data journalists.

Reports and surveys

The data continued to feed macro-level publications: the WIPO Global Innovation Index (2023 and 2024 editions) and the Stanford AI Index Report (2023 and 2024), alongside the annual State of the Octoverse. Complementing these large-scale aggregate analyses, the project also supported two surveys aimed at understanding the open source community's composition and motivations: the 2024 Open Source Survey and the inaugural Open Source Software Funding Survey.

For the year ahead, the goal is straightforward: more collaboration, more researchers and policymakers engaging with the data, and continued growth of the evidence base for open source's impact.