Programming technologies are developing rapidly, but the factors contributing to their rise and fall are not yet known. In a new study, Fabian Braesemann, associate scientist at ECDF and lecturer in AI & Work at the Oxford Internet Institute, and Conrad Borchers, doctoral student at the Human-Computer Interaction Institute (HCII) at Carnegie Mellon University in the US, examine online trace data from over a decade to understand the innovation dynamics of the software world. The results offer empirical insight into how innovation works and show that “creative destruction” (one of the best-known theories in economics) appears to drive the development of digital technologies as much as it does the renewal of economies.

Using correlation networks that encode dynamic tag usage patterns, the researchers were able to identify two robust technology clusters.

• Core computing facilities, which include operating systems, databases, and servers, and
• application development technologies, which include frameworks for web development and machine learning.

The researchers did not just examine the popularity of specific programming technologies such as Python or React over time; instead, they wanted to find out which technologies “rise and fall together”: If two technologies experienced a constant increase in usage at the same time, there was a positive connection between them. If one technology gained popularity while another lost popularity, there was a negative connection between them – a sign of competition between the technologies.

The most exciting finding was not only what was popular, but also how it became popular. The study's data supports Joseph Schumpeter's famous theory of “creative destruction.” Schumpeter argued that innovation is not just about adding new things, but that it is a “process of industrial mutation that ceaselessly destroys the old and ceaselessly creates the new” . Braesemann and Borchers' model revealed a formula for success in the world of programming. The technologies that became “winners” (the 100 most-used tags) exhibited a specific pattern of connections.

• They work well in conjunction with other newcomers: Successful new technologies had strong positive connections to other emerging technologies. They did not grow in isolation, but as part of a new ecosystem.
• They compete with the giants: Crucially, successful new technologies often had strong negative connections to established, popular technologies. They did not merely complement the old guard, but replaced it.

These findings have several implications that are directly related to current debates about programming and digital technologies.

1. Early warning signals for technological change

Since Stack Overflow tracks the behavior of millions of developers, changes in these correlation networks could be early indicators of larger changes in programming paradigms—much earlier than official statistics or data on corporate adoption.

2. Labor markets and skills

Recent work shows that developer task profiles derived from online platforms can predict the skills and even salaries required in job postings. Our findings suggest that we can go one step further and identify which skills will gain or lose relevance as technologies compete and recombine.

3. AI-assisted programming and the next wave of creative destruction

The study notes that the rapid proliferation of generative AI tools such as ChatGPT and Copilot is already changing the way developers seek help and share knowledge, reducing activity on Q&A platforms such as Stack Overflow. As AI accelerates experimentation and lowers switching costs between tools, we may see faster and more volatile cycles of technological rise and fall. This would have significant implications for education, productivity, and inequality between regions and companies that may struggle to adapt quickly enough.

4. Economic complexity and national capabilities

Programming technologies are part of countries' “software complexity”—the diversity of their digital capabilities. Understanding the evolution of digital tools can help policymakers predict which regions are well positioned to benefit from new technology clusters and which are at risk of falling behind with declining stacks.

From digital trace data to innovation policy

The study shows that a platform originally developed for debugging code has become an invaluable data set for understanding how digital innovations actually evolve. For researchers, correlation networks of developer activity offer a scalable way to model technological recombination and competition, complementing traditional patent and trade data. For journalists and policymakers, they offer a way to observe the creative destruction of the software world in near real time—not only identifying which technologies are popular, but also how the next generation of tools is quietly reshaping the foundations beneath them.

For the full article: “The innovation dynamics of programming technologies” in the Journal of the Royal Society Interface.