May 2020 – It was big news, not only for the tech industry, when Twitter’s CEO Jack Dorsey announced that the company would go fully remote. Now, we can easily imagine how the future will look like for Twitter’s employees. During the lockdown, video calls, instant messaging and cloud services have entered our work routine. This mode of remote work seems  novel to many of us, but it builds on a long tradition in software development.

Programmers, developers and data scientists of the global open source communities have been sharing code and know-how on digital platforms such as Github and Stack Overflow for more than a decade.

In fact, many of the most successful tools of digital age – the programming language Python, the operating system Linux, or the artificial intelligence toolbox TensorFlow are the result of open source collaborations.

For example, on the world’s largest question-and-answer platform for programming questions, Stack Overflow, users help each other with coding related advice, independently of who or where you are. Such platforms carry on the Internet’s promise: participation in the global knowledge community independently of time and space. A promise too good to be true?

We geo-coded historical contribution data from more than two million users of the platform from 2009 to 2017 in our recent study ‘Global networks in collaborative programming’. Our results show that the global geographies of contributions to the platform are highly concentrated in a limited number of places.

Figure 1 Global contributions to Stack Overflow in 2009 (blue) and 2017 (red): the geographies of the collaborations are changing

Figure 1 – Global contributions to Stack Overflow in 2009 (blue) and 2017 (red): the geographies of the collaborations are changing.

In addition, the map in Figure 1 reveals that the geographies of collaborative open innovation on Stack Overflow have changed: in 2009, most collaborations happened between users in metropolitan areas in North America and Europe; now India has become a global hub, too. This adds a cautious remark to the Internet’s promise: just because digital participation can happen from anywhere in the world, it does not imply that it will happen everywhere.

Where do these distinct geographies come from? Why are the contributions to a global platform such as Stack Overflow so highly concentrated? This is the question we were curious to investigate in our study ‘Coding together – coding alone: the role of trust in collaborative programming’, recently published in Information, Communication & Society.

In assigning the user contributions to 188 countries from around the world and 266 metropolitan areas in OECD countries, we find that Stack Overflow participation is related to a set of local characteristics. Contributions are highest in large cities, with a sound Internet Infrastructure, and a highly developed economy. Most interestingly, apart from economic and infrastructure characteristics, our findings indicate that the contributions benefit from societal values. Places with higher levels of trust and civic engagement are found to contribute more actively to the platform.

Conclusion

Our daily lives are shaped by digital technologies. Powerful tools such as Python or Linux are the product of open innovation. They are the best examples of what delocalised, asynchronous collaboration can achieve. Just like for Twitter employees, in the future, our work will be shaped by remote collaboration, too. A spirit of trust and knowledge sharing is needed to facilitate innovation in such a context. Our study indicates that a collaborative mindset, built of on trust, matters just as much for digital innovation as fast Internet access and computing power.

Read more:

Fabian Stephany, Fabian Braesemann & Mark Graham (2020) Coding together – coding alone: the role of trust in collaborative programming, Information, Communication & Society, DOI: 10.1080/1369118X.2020.1749699

Fabian Braesemann, Niklas Stoehr & Mark Graham (2019) Global networks in collaborative programming, Regional Studies, Regional Science, 6:1, 371-373, DOI: 10.1080/21681376.2019.1588155

For reproducibility, find all code and data at: https://github.com/Braesemann/CodingTogether