Uncovering the Digital Complexity of Nations: A Q&A on How GitHub Data Reveals Economic Insights

In a groundbreaking study published in Research Policy, four researchers—Sándor Juhász, Johannes Wachs, Jermain Kaminski, and César A. Hidalgo—used data from the GitHub Innovation Graph to explore whether open-source software production can measure what they call the “digital complexity” of nations. Their work reveals that software contributions can predict economic outcomes like GDP growth, inequality, and emissions in ways traditional metrics cannot. Below, we dive into their findings through a series of questions and answers.

What is the “digital complexity” of a nation, and why is it important?

Digital complexity refers to the knowledge embedded in a country's software production, measured by the diversity and uniqueness of programming languages used by its developers. For over a decade, economists have used the Economic Complexity Index (ECI) to assess nations' productive knowledge based on physical exports, patents, and research publications. These measures are powerful predictors of economic growth, inequality, and environmental impact. However, they entirely miss software—a critical component of modern economies. Code doesn't cross borders through customs; it moves via cloud services and package managers, creating a blind spot that researchers call “digital dark matter.” By applying the ECI to GitHub data, the team quantified this hidden knowledge, showing that software complexity fills a gap in traditional economic analysis and offers fresh insights into national prosperity and challenges.

How did you use GitHub Innovation Graph data to measure digital complexity?

We leveraged the GitHub Innovation Graph, which tracks developer activity globally by IP address, showing how many developers in each country push code in different programming languages. We then applied the Economic Complexity Index (ECI) to this data—a method originally used for physical products. The ECI measures both the diversity of languages a country uses and the ubiquity of those languages across countries. A nation that uses a wide range of languages that few others employ is considered more complex. This approach revealed distinct patterns: countries with high software complexity tend to have advanced digital economies, while low complexity often correlates with reliance on common languages. The data allowed us to create a “software ECI” that mirrors traditional complexity metrics but captures the intangible digital dimension.

Why is software a blind spot in traditional economic complexity measures?

Traditional economic complexity measures focus on tangible outputs—physical products that pass through customs, patents filed with national offices, and peer-reviewed research papers. Software, however, is intangible and globally distributed. A developer in one country can contribute to code hosted on servers in another, and a “git push” can instantly share new knowledge across borders. This digital activity is invisible to standard trade and innovation statistics. The result is that economists have been missing a huge chunk of productive knowledge: the code that powers everything from apps to infrastructure. As Jermain Kaminski notes, this has been called the “digital dark matter” of the economy. Our work using GitHub data shines a light on this dark matter, revealing that software complexity not only exists but also predicts macroeconomic outcomes independently of traditional measures.

What are the key findings of your study?

Our study found that the “software ECI”—derived from GitHub developer language data—is a robust predictor of several key economic indicators. Specifically, nations with higher software complexity tend to have higher GDP per capita, lower income inequality, and lower carbon emissions, even after controlling for traditional complexity measures. This suggests that software production captures unique aspects of a country's productive knowledge that physical exports or patents miss. For example, a country that produces a diverse range of niche programming languages likely has a sophisticated digital ecosystem, fostering innovation and efficiency. Moreover, software complexity provides early signals of economic shifts, as code is written long before products are shipped or patents granted. This makes it a valuable tool for forward-looking policy analysis and research.

How does software complexity compare to traditional complexity measures based on physical exports?

While traditional measures of economic complexity—based on what a country exports physically—are well-established predictors of growth, they fall short in capturing the digital economy. Software complexity is distinct: it correlates only moderately with physical export complexity, meaning it provides new information. In many cases, software complexity predicts outcomes like GDP and inequality better than traditional measures alone. For instance, a country like India, with a strong software services sector but moderate physical exports, shows high software complexity. Conversely, resource-rich nations may have high export complexity but lower software complexity. This complementary nature means that combining both measures gives a fuller picture of a nation's productive capabilities. The study emphasizes that ignoring software leaves a significant explanatory gap—one that the GitHub Innovation Graph data helps fill.

What implications does this research have for policymakers and economists?

Our findings suggest that governments and international organizations should incorporate digital metrics into their economic monitoring and development strategies. Traditional indicators like export baskets and patent counts are insufficient for understanding modern economies. By using GitHub Innovation Graph data, policymakers can gauge a nation's digital complexity and identify areas for growth—such as promoting diverse programming language skills or fostering niche digital industries. This could inform education, immigration, and technology policies. Additionally, the connection between software complexity and lower inequality and emissions offers a new angle for sustainable development. For economists, the research opens up a new data source for studying innovation, knowledge flows, and economic dynamics. The GitHub Innovation Graph is freely available, enabling further studies to build on this work.