An American physicist, a Japanese mathematician and a German cosmologist walk into a lab; what do you get? Based on recent outcomes, you'll get ground-breaking science. And lately, governments have begun paying heed to evidence1 that suggests international, multidisciplinary collaborations such as these will yield high-impact results.

Tea time at Kavli Institute allows for an organized and informal exchange of collaborative ideas. Credit: Kavli Institute

Policymakers from diverse countries, including China, Japan, Australia, Chile and Germany, have sought to foster excellent science and technological innovation — and reap the associated economic benefits — by promoting collaboration across borders and disciplines, and setting up specialist centres with the necessary resources (see 'Conduits to collaboration').

figure 1

Governments around the world have established special centres with the ambition of fostering high quality research through collaboration. This graphic shows up to 10 of the top centres per country in the Nature Index 2014.

But does this kind of top-down approach really increase the number of collaborations beyond what may have occurred organically? And, more importantly, is it producing the best possible research? Here, the answers aren't so definitive. Nature Index data show that some specialist centres are much more collaborative than others, and that their collaboration scores range from several hundred down to a handful.

Tea and teamwork

The scenario described in the opening line is what you may find any weekday afternoon in the atrium of the Kavli Institute for the Physics and Mathematics of the Universe (IPMU) on the campus of the University of Tokyo, Japan. The institute has a mandatory staff tea time during which ideas and avenues for potential collaboration are shared.

I don't think taxpayers like paying for tea, but it was worth fighting for.

Tea time is a custom that Kavli IPMU director and particle physicist, Hitoshi Murayama, brought back to Japan from the United States after witnessing its benefits in labs he had worked in. He was so convinced of its value in promoting collaboration among researchers that he put it in the initial proposal for the institute and battled with the funders to have it included. “I don't think taxpayers anywhere like to pay for tea, but I knew that it was worth fighting for,” he says.

The Kavli IPMU is a product of the Japanese Government's World Premier International Research Center Initiative (WPI), launched in 2007 with the aim of creating “globally visible research centres”. With more than 350 collaborating partners in index papers in 2014, Kavli IPMU is the most prolific of the nine WPIs (see 'Teams of all sizes').

figure 2

Japan’s World Premier Institutes (WPI) illustrate the variability of collaborative patterns across disciplines.

These WPIs are part of a growing collection of initiatives established by governments around the world with similar goals. Prominent examples include the ARC Centres of Excellence in Australia, the Millennium Science Initiative in Chile, Germany's Clusters of Excellence, and, most recently, the Collaborative Innovation Centres in China.

The common goal underpinning these schemes is to gather a critical mass of leading researchers to work together on cutting-edge projects that will potentially have an impact on the world stage. Each of the initiatives mentioned has an explicit mandate to increase research collaboration, both domestically and internationally, although there is less emphasis on domestic partnerships in the Japanese and Chilean models. The research bodies created under these schemes are given significant autonomy to determine their own research priorities, and funding is granted for substantial periods, a decade or more in the case of WPIs, to enable longer-term research projects to be planned with confidence.

For the WPIs, guidelines specify that at least 30% of the researchers at any institute must be from overseas, and English is to be the primary language. Still less than a decade old, Kavli IPMU has ticked off all requirements, and is achieving impressive research collaboration.

Kavli's core business, astrophysics and astronomy, both demand enormous collaborative efforts. Along with the geosciences and high-energy physics — where a single paper can have as many as 3,000 authors — these are the realms of science most obviously suited to such collaborative centres. In Europe, for example, the TUM Cluster of Excellence for the Origin and Structure of the Universe has the highest collaboration score among Germany's Clusters of Excellence.

Away from physics, astronomy and geosciences, the Munich Cluster for Systems Neurology (SyNergy) is the most collaborative of all the government-backed collaborative research institutions included in the Nature Index analysis. In 2014, it shared authorship with well over 100 separate research bodies — the majority fellow German institutions.

At the other end of the scale, China's relatively young Collaborative Innovation Centers — the first were established in 2012 — have fewer partners, particularly when it comes to international institutions. But it seems likely that numbers will increase before long.

Research policy risk

Some observers caution that such government-directed research initiatives remain focused on generating the highest quality science at the expense of other policy goals.

Caroline Wagner, an Ohio State University expert in the links between science, policy, society, and innovation, advises that a distinction should be made between the kind of collaborative activity organized by scientists themselves — where the research determines the organizational structure of the teams — and the government-driven collaborations that have different objectives. “These can be about a lot of other things in addition to science, for example, national prestige, industrial competitiveness and encouraging intellectual property development.”

She says researchers should resist policy initiatives that get in the way of achieving the highest scientific standards — pointing to European Union rules dictating that participants from at least three different EU states must be involved in any successful proposals — a policy designed to lift the performance of those states with weaker science records.

“Adding these policy goals into the science almost always reduces the quality of the science, but policymakers are willing to buy down the quality of the science in order to reach other kinds of social or political goals.”

Les Field, secretary of policy with the Australian Academy of Science, shares Wagner's observations about the potential pitfalls of mixing science and policy, but says the success he has seen achieved by the best Australian centres of excellence and their international equivalents, producing great science and establishing collaborations at the highest levels, should be acknowledged and encouraged by governments. “It's important to recognize the benefits that you get out of the continuity of some very high-powered teams which can contribute very significantly to the reputation and economic progress of the country.

Indeed, there is good science being produced by collaborative institutions included in the index.

At the Kavli IPMU, for example, collaboration with researchers from a range of Japanese institutions and from Duke University in the United States, led to the recent discovery of a gravitational lens in space created by a hitherto unknown galaxy which explained the super-luminescence of an unusual supernova that had been puzzling astronomers2.

Elsewhere, research from Australia's ARC Centre of Excellence for Climate System Science, in collaboration with researchers from France's Laboratoire de Meteorologie Dynamique, has significantly increased the sensitivity ascribed to the effects of increased carbon in the earth's atmosphere in climate modelling3.

Director of the ARC centre, Andrew Pitman, says the rapidly increasing complexity of climate modelling over the last decade, requiring computer coding with as many as one million lines, means that the field has become dependent on collaboration facilitated by centres such as his. “No one single country can build the climate models and maintain international competitiveness in modelling systems so we collaborate with the big groups in the US, Germany and Britain.”

Pitman says that the rules applied to work at the centres of excellence, which force collaboration, allow them to more easily overcome the structural biases that exist against sharing research credits with other institutions.

Meanwhile, he says, having a critical mass of researchers working together with good funding over a long period enables necessary software experts and technicians to be employed and widely shared, and for younger scientists to receive training.

A critical mass means that international linkages form far more easily.

Bringing researchers into such centres also allows them to travel and work with peers around the world — establishing the links necessary for future collaborations — Pitman says. “It may just occur naturally but I think having the critical mass means that the international linkages form far more easily.” A.M.