When we think about innovation, we usually picture something new: a new machine, a new medicine, a new energy technology, or a new digital tool. Progress is often imagined as a stream of fresh ideas entering the world.

But innovation also happens in a quieter way. Sometimes, an existing idea is moved into a new category. A technology once seen as belonging to one field suddenly becomes relevant in another. A patent written years ago may later be recognised as useful for a new application. A classification system changes, and with it, our understanding of what a technology is connected to.

This process is called reclassification. It sounds technical, but the idea is simple: old knowledge can gain new meaning.

A good example is solar technology. Solar panels were once mainly associated with satellites and space applications. Later, they became part of energy systems, rooftops, climate policy, and everyday infrastructure. The technology itself did not appear from nowhere at that later moment, but its place in the wider knowledge system changed. That change matters.

My research is centered around a simple question: can this kind of reorganisation help technological innovation move faster?

The answer appears to be: quite possibly, yes.

Using patent data, I find that reclassification is not random noise in the innovation system. It follows clear patterns. First, newer patents are more likely to be reclassified than older ones. That makes intuitive sense. Recent inventions are often still being understood. Their uses are not always obvious at the moment they are filed. As more people work with them, new connections become visible.

Second, larger technology classes tend to attract more reclassified patents. In other words, areas of technology that already contain many inventions are also more likely to pull in older inventions from elsewhere. This may be because large technology areas have a broader scope. They touch more problems, more industries, and more possible applications. Once a field becomes widely relevant, more existing ideas can suddenly be seen as belonging to it.

These finding allow me to build a mathematical model to connect these patterns to technological growth. The model combines two processes. One is the familiar process of invention: existing knowledge helps trigger new knowledge. The other is reclassification: existing patents are moved into new technological categories as our understanding changes.

The key result is that reclassification and growth are closely linked. Technologies that are reclassified more often also tend to grow faster. This does not prove that reclassification alone causes faster innovation. I have to be careful with such claims.  But the relationship is strong enough to suggest that how we organise knowledge may affect how easily future ideas can build on it.

That is an encouraging message. Innovation policy often focuses on money, infrastructure, skills, and regulation. These are all important. But my research suggests that the structure of knowledge itself also matters. Better maps of technology may help researchers, firms, and policymakers see useful links sooner.

There is another useful implication. Patent counts often appear to decline in the most recent years of a dataset. This is usually treated as a data problem: recent patents have not all been published yet, or databases have not fully caught up. That is partly true. But my findings show that reclassification can also create an apparent decline. Recent patents have had less time to be reclassified into all the categories where they may eventually belong. So a technology may look like it is slowing down, even when the apparent dip is partly caused by the way classifications evolve.

This matters especially for fast-changing areas such as green technology. If recent green patents appear to fall, it may be tempting to conclude that innovation is weakening. But some of that pattern may simply reflect the fact that these technologies are still being sorted, reinterpreted, and connected to new uses. The boundary of a technology is not fixed. It evolves.

The broader lesson of my research is that knowledge is more flexible than it looks. A patent is not just a fixed record of an invention. Over time, it can become relevant to new fields, new problems, and new technological pathways. Reclassification captures part of that movement.

This gives a more hopeful view of innovation. Progress is not limited to waiting for entirely new ideas. We can also create value by seeing old ideas differently, by improving classification systems, and by making connections easier to find.

For researchers, this means that classification systems are not merely administrative tools. They shape how knowledge is searched, compared, and reused. For policymakers, it suggests that better knowledge infrastructure may support faster innovation. For the public, it offers a simple but powerful insight: invention is not only about creating more pieces of knowledge. It is also about arranging the pieces in ways that reveal new possibilities.

Please note I do not claim that reclassification is a magic button for innovation or that every reclassification reflects a major conceptual change. Some are practical updates to help patent examiners search more efficiently. Still, at scale, the patterns are meaningful.

The optimistic conclusion is this: technological progress may depend not only on how many ideas we produce, but also on how well we keep rediscovering, reorganising, and reconnecting the ideas we already have. Innovation is not just about the next invention. It is also about learning to see existing knowledge with new eyes.

I am very grateful to the Oxford Martin School and in particular the Oxford Martin Programme on Technological and Economic Change for making this research possible. I am also grateful to Kerstin Hotte and Nicolo Barbieri for joint developments of the data strategy and fruitful discussions.

This blog was created partly using the aid of a large language model. 

About the author: Peter Persoon worked on the Oxford Martin Programme on Technological and Economic Change as a Postdoctoral Researcher from February 2022 to September 2024.

This opinion piece reflects the views of the author, and does not necessarily reflect the position of the Oxford Martin School or the University of Oxford. Any errors or omissions are those of the author.