The why, how and where of mapping innovation
“If we don’t create and spread new ideas, our economy has no future”
The Minister knows that innovation is vital for the country. Everybody is talking about the impact of AI, Industry 4.0 and blockchain. But when she looks at the official statistics, she can’t see any of that. She is flying blind.
A couple of floors down, a civil servant reviews a pile of proposals for new research centres in hot ‘disruptive’ technologies. The one about Brain Computer Interfaces centre looks very good. But is this the right location? Will the investment also benefit industry and the rest of the country? He just doesn’t know.
Zip out of the country’s capital and into its industrial heartland, to the office of a seasoned innovation fixer. She has spent the last decade connecting local entrepreneurs and universities, and chasing innovation funding. It has been an uphill struggle, but things are looking good right now, she can feel the buzz in the city. She only wishes she had better numbers to show how things have changed and convince big investors to come over and see.
Not far away, in a brand new co-working space set up by our innovation fixer, a tech entrepreneur thinks about her next steps. She has a new idea for a Brain Computer Interface to control drones. She is excited, but also daunted. She needs knowledge, collaborators and advice - but how to find them? A lot of the business support is generic. The local university will try to get her to work with them, but she feels there might be a better fit further afield. Sadly, she doesn’t even know where to start looking.
Like many thousands of others like them across the UK and beyond, these adventurers all have one thing in common: They need the right map.
Creating and applying new ideas is more important now than ever, but the maps that policymakers and practitioners need in order to understand how and where this innovation is happening, and what to do about it, are seriously lacking.
The stakes are high. Take the UK, for example: the country’s productivity record is abysmal, and its economic geography extremely lopsided. In spite of its world-leading universities and vibrant start-up scene, UK businesses are between a quarter and a third less productive than their counterparts in Germany and France. GVA per head of population is 70% higher in London than the UK average. These challenges have been recognised far and wide, from the final report of the Industrial Strategy Commission to London School of Economics’ recent Atlas of Industry.
But what to do about this?
Innovation policymakers are becoming more activist and directional. In its recent Industrial Strategy white paper, UK Government sets out sector deals and challenges to strengthen specific sectors and technologies like AI or Life Sciences, and catalyse innovation coalitions to deal with big economic and social problems like ageing or climate change. But how will we know if we are headed the right way, and at what speed? How do we avoid the bog of vested interests, that hobgoblin of all activist economic policies?
Obviously, we need better maps.
Existing maps and measures of innovative activity based on official business surveys and economic statistics, numbers of patents and papers published are not enough[i]: They were created to monitor a slow-changing economy where decennial updates in the industrial codes were enough to capture new industries. They take countries as the main unit of analysis (hence the gaps in regional and local data) and target national policymakers as their primary audience, ignoring the data needs of many other important groups locally and outside government. They assume that innovation concentrates in a small number of science and tech-based sectors, and lack information about the identity of individual companies and their unique quirks.
How can these data help us understand an innovative economy where novelty is the norm and takes many different forms beyond patents and widgets, an economy where space and place matter, where understanding and managing complex networks is an important policy goal, and where outliers are often the most interesting observations?
We need new maps to navigate this territory. New sources of data (including web, open and so-called ‘big’ data), analytics methods such as machine learning and text mining and interactive tools for data visualisation can help us to build and present these maps. These methods are being applied to monitor the evolution of scientific research, but their adoption has been slower in industrial and innovation policy. We have been working to fill this gap through projects such as Arloesiadur, Tech Nation, The Geography of Creativity and UK Games Map. These experiments have demonstrated the potential of the approach - see for example the papers we presented at the Data for Policy conference earlier this year, where we use text mining to analyse research trends and business activity.
In the coming months we will be scaling up our innovation mapping efforts with an ambitions programme of work where...
Imagine using a 2007 newspaper to try to understand what is happening in the world today. It would not be very useful. Yet this is what we do with the economy. The UK’s economic statistics are based on industrial codes agreed in 2007. This means that they cannot be used to measure the new industries that appeared since, from Blockchain to the Internet of Things. Unfortunately, these are some of the sectors that innovation policymakers are most interested in measuring. Paraphrasing US economist Rober Solow, we can see innovative firms everywhere except in the official statistics!
Immersive technology (including Virtual and Augmented Reality) is one of those promising sectors that is almost impossible to measure with conventional data sources. Our Immersive Economy Mapping project for ImmerseUK aims to overcome this hurdle using a new approach developed together with GlassAI, a tech start-up that collects web data about hundreds of thousands of UK organisations, and MTM London, a creative research agency. In a nutshell, we look for organisations that talk about Immersive technologies and tools in their websites - this allows us to to not only map immersive tech developers and creative companies producing Immersive content, but also organisations in other sectors who are using immersive tech to deliver new products and services.[ii]
In another ongoing project, we are working with the Mohammed Bin Rashid Centre for Government Innovation in Dubai to map innovation in United Arab Emirates. There, we are mining research paper abstracts, business websites, online job ads, university courses, patents and information about tech communities to track research and tech trends in areas of strategic importance for UAE, ranging from big data to fintech. We will publish an interactive data visualisation with key findings very soon.
We have long insisted that new types of data - for example, big datasets from company websites - aren’t here to replace traditional economic statistics and indicators, but to complement them: Big data is not just about volume but also about variety, bringing together the novel and the traditional in ways that strengthen each other. Arloesiadur, our innovation dashboard for Wales is a good example of this. In it, we have combined official economic data, open research data and web data about tech meetups to provide a holistic view of Wales’ innovation system.
Creative Nation, our forthcoming analysis of creative clustering will also integrate data from many sources to answer some vital questions about the geography of creativity in the UK: What clusters are growing faster? Are the UK creative industries becoming more or less concentrated in a small number of places? Does growth in a location help or hinder its neighbours? What sectors outside of the creative industries are becoming ‘more creative’ in their capabilities? Where are the strongest research collaborations between creative industries and universities? Are UK clusters suffering a ‘creative brain drain’ after the EU Referendum?
We will publish our findings at the beginning of 2018, together with an open dataset and an interactive data visualisation where users can explore the data. This analysis will provide a strong evidence base for ongoing policy efforts to strengthen creative clusters outside London and the South East of England, addressing the geographical imbalances we mentioned before.
Although we are very excited about the potential of data science for innovation policy, we are not oblivious to its risks: New data could be biased (if for example, video games developers are more active in social media than biotech companies) or track the wrong thing (are you measuring an increase in the number of tech meetups in London, or in the popularity of the Meetup website?). The results of machine learning algorithms that, for example, classify companies into sectors can be hard to explain and interpret. We worry that these risks could be discouraging innovation policymakers and researchers from experimenting with new data sources, creating a catch 22 situation where mistrust prevents exploration, and lack of exploration prevents learning.
We are working with the European Commission in two projects to tackle this issue, taking new data, indicators and methods from the periphery of innovation policy to its core.
The first one, EURITO (EU Relevant, Inclusive, Trusted and Open indicators for Innovation Policy) is a three year project funded by the Commission and done in partnership with DTU, Fraunhofer ISI and Fundación Cotec. This project will follow a step-by-step process to identify burning innovation policy questions that could be addressed with new data, explore those opportunities through agile pilots and strenuously validate the results with other data and domain experts to build trust in our new indicators. Going beyond new indicators for innovation policy, the project will create an open data and code infrastructure that can be reused by others, and interactive data visualisations to explore the findings.
We are also working with MERIT, Deloitte and SPI in the latest refresh of the European Innovation Scoreboard. The Scoreboard is an important source of evidence about innovation in the EU until now based on traditional data sources such as business surveys, patents and numbers of papers published. At the beginning of 2018, we will run exploratory pilots to identify new opportunities to enhance the Scoreboard with indicators based on big data, hopefully informing its revision after 2019.
The best maps allow you to zoom in, going from a bird's eye view right down to the street level. We can do the same with many of the data we collect, drilling down from aggregate trends and national and regional indices to individual organisations, communities and projects. In doing this, we can create valuable information for policymakers and other stakeholders - entrepreneurs, investors, researchers - looking for the right lead. For example, in Arloesiadur we identified technology communities working in particular domains such as big data or cybersecurity, and built a prototype recommendation engine to help Welsh innovators find new partners.
We will continue pushing this idea from January in a 15 month Health Innovation Mapping project supported by the Robert Woods Johnson Foundation. There, we will explore multiple data sources such as open source software repositories and crowdfunding sites, academic conferences and data about research papers and grants in order to identify interesting health innovations. We will present the results in a health innovation scanner providing a comprehensive, highly detailed view of the health innovation landscape, including as much as possible innovations in developing countries, and social innovations that promote a Culture of Health. We hope that all this information will help the Robert Woods Johnson Foundation and other funders (including Nesta) find new ideas to support and scale-up.
One (meta) way to think about all these mapping projects is as explorations of a new and uncharted space of opportunity for innovation policy. In the coming months we will be working with many different data sources and analytics methods, and presenting our data in a variety of ways. Some of the things we try will work, and others won’t. We will be open and transparent about what we learn, so that others can build on our successes and avoid our mistakes.
We will also be thinking hard about the adoption and diffusion of our results, and particularly what they mean for the organisation of innovation policy, addressing questions such as:
In addressing those questions, we want to make innovation policies smarter, more inclusive and dynamic, helping our societies and economies create more and better ideas that are applied for the benefit for all. Now that’s a trip worth taking.
Get in touch with [email protected] if you are interested in our work or are working in this space. We’d love to talk to you!
This blog received helpful comments from Kirsten Bound
Image credits: Stitch of panel A and B of the Carta Marina map (Wikimedia)
[ii] This illustrates another advantage of ‘bottom up’ categories based on rich web data over rigid sectoral classifications: we can tag companies with more than one sector, recognising the fact that innovative companies often sit across sectors and that innovation is as much about adopting new ideas as it is about creating them. It has in fact been argued that the UK’s productivity problem hinges on the lack of diffusion of new ideas from companies at the cutting edge to those coming behind.