Why institutional innovation matters for the future of science and technology

Science and technology have long been shaped by public institutions that fund, regulate and shape the work of researchers. Some are very famous, such as DARPA, CERN or Japan’s METI, and some are very influential: seventy years ago the US Department of Defense accounted for a third of all global R&D funding (the figure now is about 3%).

The UK has done its fair share of institution building over the years. But if you look closely at the recent creations – including the Catapults, UKRI and ARIA – a strange pattern becomes apparent. Each was based on a very old model. The research councils model dates back more than a century. ARIA was based on DARPA, invented in the late 1950s, the Catapults based on the Fraunhofer Institutes, invented in the late 1940s.

This pattern of institutional conservatism turns out to be common across public sectors worldwide. It contrasts strikingly with business which has radically reinvented itself in recent years with firms based on platforms (like Uber), algorithms (Tiktok) and search (Google). Most of the top 10 most valuable firms in the world are based on data and knowledge (and also now invest very heavily in R&D).

Yet governments have hardly been touched by these trends. Ministries look very similar to a century ago. Finance remains the dominant function within government, and in science and technology, rather than data or knowledge.

This matters as the UK faces big challenges that don’t fit well with the current institutional landscape. The most obvious come from a more competitive global landscape where trade, regulation and security intersect with technology. The UK currently lacks institutional capacities to think these issues through let alone act on them.

Then there are the problems of public trust – only 9% of the public think that publicly funded R&D benefits them a lot. Yet government has de-prioritised public engagement and seems to expect a passive public to be grateful for what science can give them. Consultancies and tech firms have largely squeezed out public input (for a much more detailed analysis of what DSIT is getting right and wrong, see this).

Just as serious is the stark evidence of declining productivity in R&D, long apparent in pharmaceuticals and now also apparent in many sectors, according to research by John Van Reenen who chairs Rachel Reeves Council of Economic Advisers. These trends are never mentioned in reviews or government documents – presumably because they might prompt reforms that would be uncomfortable for incumbents.

Luckily, we can learn from other countries that at least provide prompts. China invests in technology development through over 1500 government guidance funds, like the China Internet Investment Fund, that have deployed well over $1trn in recent years. India has pioneered digital public infrastructures. Germany tried to create a new agency for technology adoption, DATI, a major gap in the UK landscape given stagnant productivity in so much of the economy. Many countries are creating AI regulators, like AESIA in Spain. And many, too, have institutions for social innovation – increasingly seen as a vital complement to technology innovation, but a blindspot for DSIT.

In the UK our default is still to create a committee to review the landscape, and if a new institution is to be created, a law is passed which then in turn creates another committee which appoints a CEO who fills out a traditional, pyramid-shaped organogram, perhaps helped by an expensive consultancy.

Yet there are many other options. Some are more evolutionary, with stage-gated growth and learning by doing rather than top-down blueprints. There are many other organisational options too: not just platforms but also ‘mesh organisations’ that link multiple tiers of government, as well as business and NGOs; more mycelium-like structures that connect multiple players in a system; institutions based on protocols (like the Internet); or stacks, like the India Stack and now the Eurostack for organising digital transactions and communications which provide potential models for many other areas of science and technology.

And there are options that create more empowered teams outside the traditional bureaucracy – the model used by the US in WW2 to transform industry and by India to create its digital identity programme, the biggest public innovation project of this century.

Unfortunately, none of these options was considered when the UK created recent institutions, and there appears to be no centre of expertise on institutional architecture either in government or elsewhere. Instead, we revert to cycles of merging or splitting agencies (the ‘Lego bricks’ approach to government), which takes up huge amounts of management time to little effect, and simply recycles old models.

This cannot be healthy. Last year Daren Acemoglu and his colleagues won the Nobel Prize for economics for research which showed that institutions were the decisive factor shaping prosperity and national success over the last century. If he’s right, then it follows that how well we design institutions is likely to have a big influence on our prosperity in decades to come. This is the premise of the Institutional Architecture Lab which works with many governments from India to Finland, Japan to Germany, as well as the UNDP to broaden horizons and help governments escape from the tyranny of old-fashioned organograms.

Given the central importance of science and technology to future growth the UK’s excessive institutional conservatism could carry a heavy cost. There are innovations underway in funding (eg randomisation) and in philanthropy. Some public institutions - such as ARIA - are acting in fresh ways. And some small teams have been created which are important pointers to the future - such as the Metascience Unit and Regulatory Innovation Office.

But these remain marginal add-ons that leave the big institutions untouched. And there is almost no debate about reshaping the public institutions of science and technology with intelligence at the core - rather than seeing this as the job of small units.

In short, public institutional design is stuck and neglected. Perhaps not surprisingly, the incumbents see little reason to change. Given that the UK has recently committed over £80bn to science over the next four years we urgently need to match creativity in science and technology with creativity in the design of institutions to support them. If we don’t, institutions that are too slow moving, too opaque, too bureaucratic and too hierarchical are likely to hold us back.

[A version of this piece was carried by Research Professional and presented at the MetaScience conference in London in July 2025. Geoff Mulgan is a professor at UCL, a former director of the Government Strategy Unit and author of ‘When Science Meets Power’]