Making science work
In my annual Anniversary Addresses I want to tackle important themes and issues of current interest related to science. This year I will discuss how to make good decisions about what scientific research should be supported, particularly with respect to applications of science for the public good. The term public good is meant in the widest possible sense: improving health and quality of life, driving innovation in support of the economy, securing sustainability and protection of the environment, contributing to culture and enhancing our civilization. Making good funding decisions is concerned with making science work well, generating reliable scientific knowledge and using science to bring about benefits for society. Better knowledge of ourselves and the natural world enhances our culture and is a central aspect of civilization, aims that science shares with the humanities. This year I will not consider this specific benefit of science for society further but will do so in a later Anniversary Address.
The discovery of new scientific knowledge and the application of scientific knowledge are sometimes presented as being very different from each other. The fact is, however, that scientific enquiry has always been concerned with acquiring knowledge both of the natural world and of ourselves, and with using that knowledge for the public good. Francis Bacon argued that science improved learning and knowledge, and that led to the relief of man's estate. This argument was reinforced by Robert Hooke at the birth of the Royal Society, who emphasized how scientific discoveries concerning motion, light, gravity, magnetism and the heavens helped to improve shipping, watches, optics and engines for trade and carriage. There is a continuum from discovery science acquiring new knowledge, through research aimed at translating scientific knowledge for application, onto subsequent innovation. This spectrum should be considered as an interactive system, with knowledge generated at different places within the continuum influencing both upstream in the creation of new discoveries and downstream in the production of new applications. A historic example of how investigations downstream can influence research upstream was work on improving the steam engine, which greatly informed the subsequent formulation of thermodynamics. It is important to emphasize this continuum of science spanning discovery through translation to innovation. We should reject attempts to place artificial barriers in the continuum and those who argue that different parts are superior to other parts. Science throughout the continuum shares the same values, skill sets and methodologies, although as I shall discuss there can be differences in emphasis in how research is carried out.
What factors have to be considered when deciding which scientific research should be supported? Several are important but the one that I think is crucial is the scientist carrying out the research. Major discoveries in science are usually associated with highly talented individuals who combine a number of qualities: they should have in-depth knowledge, be creative, understand the values of science and how research is done, be well motivated, and be effective in achieving what they set out to do. In-depth knowledge of an area of science is essential but this needs to be combined with what John Cadogan has called ‘peripheral vision’, an understanding of and openness to what other sciences can contribute. This is especially required when solution of a research problem needs multidisciplinary and interdisciplinary approaches. Carrying out good scientific research is a creative activity, and scientists have more similarities than might be imagined with those pursuing other creative activities such as the arts, writing and the media. Like other creative workers, scientists thrive on freedom and organizing them is like ‘herding cats’. Freedom of thought, to pursue a line of investigation wherever it may lead and to uncover uncomfortable truths, is crucial to an effective scientific endeavour. A scientist whose thoughts are restrained, who is too strongly directed, or who is unable to freely exchange ideas will not be an effective scientist. Similarly, societies that are not free and do not encourage the free exchange of ideas or respect the values of science cannot be leading scientific powers because that freedom is closely connected with the creativity required for good science.
Scientists need to embrace the values of science, to have respect for reliable and reproducible data, a sceptical approach that challenges orthodoxy and the scientists' own ideas, an abhorrence of the falsification or cherry-picking of data, and a commitment to the pursuit of truth. Scientific research is hard and, to be effective, research scientists need to be highly motivated. Often this motivation is provided by a passionate curiosity about the natural world, a desire to know how things work or how they can be directed to achieve particular outcomes. But other motivations are also important, a desire to undertake public good through the eradication of disease for example, to make something useful, to create economic wealth, or to become famous. Whatever the motivation, it needs to be strong because the pursuit of research is long and difficult. So in deciding what research should be supported, much attention should be paid to the scientists carrying out the work, and as far as possible decisions about research projects should be closely associated with assessments of the individuals proposing the projects.
Given this emphasis on the primacy of the individuals carrying out the research, decisions should be guided by the effectiveness of the researchers making the research proposal. The most useful criterion for effectiveness is immediate past progress. Those that have recently carried out high-quality research are most likely to continue to do so. In coming to research funding decisions the objective is not simply to support those that write good-quality grant proposals but those that will actually carry out good-quality research. So more attention should be given to actual performance rather than planned activity. Obviously this emphasis needs to be tempered for those who have only a limited recent past record, such as researchers early in their career or those with a break in their careers. In these cases making more use of face-to-face interviews can be very helpful in determining the quality of the researcher making the application. The greater costs involved in direct interviews will be more than recompensed by the greater quality of the decisions that will be made. So making good decisions about research funding requires a focus on the quality, passion and past performance of the scientist proposing the research.
A perennially vexing question is how prescriptive research funding agencies should be in determining what research areas should be supported. This recurring issue arises because of the tensions between scientists wanting the freedom to decide what projects they should pursue and society which supports science not simply as a cultural activity increasing knowledge but also as an activity aimed at improving the lot of humankind through achieving specific useful objectives. One possible response of funding agencies faced with this issue is to carry out a strategic review to decide priorities and identify research areas judged either as being especially timely for future scientific advances or as reflecting particular needs for society. This can lead to initiatives that shape or sponsor research, sometimes with ring-fenced allocations of research funding. Although well intentioned and attempting to address an important issue, these approaches run the risk of funding lower-quality research. One problem is that decisions are separated from consideration both of specific projects and of the scientist carrying out that project. As a consequence such initiatives may attract less creative and effective scientists who simply follow where resources are being made available. A second problem is that the identification of favoured and non-favoured research areas is usually made by committees made up of senior researchers who are sometimes not particularly active in research any more themselves. Such committees are prone to coming up with the rather obvious and being behind the cutting edge. Better judgements are more likely to be made by the scientists actually carrying out specific areas of research who are much closer to the research problem being pursued.
So how can this difficult tension be resolved? In my opinion there are three issues that are relevant: the Haldane Principle, or rather what we understand the Haldane Principle to be; a different approach when considering programmes aimed at achieving applications and specific goals; and a more imaginative role for scientific leadership in influencing funding. The Haldane Principle is usually interpreted as meaning that researchers and not politicians should decide how to spend funds, although the original Haldane Report made no reference to any specific principle. Science Minister David Willetts has recently expressed his understanding of the Haldane Principle as meaning that politicians, informed by external advice, should decide on the overall science budget and the allocation between Research Councils along with identifying key priorities such as specific challenges or key infrastructures. He argued that politicians should not be involved in decisions on specific funding proposals, which should be made by researchers using peer review. This is a sensible view that I would extend further by arguing more generally that decisions should be made as close as possible to the researchers actually carrying out the research. Such thinking can be extended to decision makers further down the funding chain. Those leading research-funding bodies should focus their attention on high-level priorities, avoiding the temptation to become too prescriptive and fine-grained in recommendations concerning what areas should be funded. This should be left to those closer to the research. The point I am making here can be illustrated by a metaphor derived from geographical exploration. In the nineteenth century the Royal Geographical Society based in London supporting an expedition might decide that it wanted to sponsor exploration of the Amazon basin, the source of the Nile, or the Antarctic. But it would have been ill advised to be too fine-grained in its deliberations to specify which Amazon tributary or African lake or South Polar glacier should be the focus of attention. That should be left to the explorer on the ground, not the committee in London. The funder's role should be to define the general geographical region of interest, identify the best explorer and then properly equip that explorer so they can be most effective in the field. Research funders should behave in the same way. They should put their trust most in the explorer scientist carrying out the research rather than in a committee in London. As far as possible, research funding decisions, especially at the discovery end of the research spectrum, should be driven by the scientists carrying out the research because they are the ones best placed to shape the research agenda. This is the reason why response-mode funding is an effective way to deliver new scientific knowledge.
However, this approach needs modification when a research programme is directed at achieving specific goals or applications; this does require a more prescriptive behaviour. Goal-directed research can occur anywhere in the scientific spectrum—whole-genome sequencing would be an example at the discovery end—but tends to be more prevalent when thinking about applications through translation and innovation. It is necessary and valuable to identify sectors that are close to application as being areas that are worth supporting. However, identification of sectors worthy of support should be broadly scoped and should involve both those carrying out the research and those who want to use outcomes of the research being supported. Generally this involvement should also include financial contribution from those wanting to exploit the research as a statement of their support. This more prescriptive approach applies to research close to application across the whole spectrum, both for-profit activities driving the economy and not-for-profit activities such as improving health and protecting the environment. But even when decisions are more prescriptive they always need to be driven by quality, both of the research proposed and of the researcher. Two further points need to be made. The first is that not all research close to application should be prescriptive; there is an important role for bottom-up response-mode funding in the translation and innovation parts of the research continuum. The second is that more prescriptive approaches are also sometimes needed at the discovery end of research, for example when assembling large data sets such as genome sequences and meteorological data, or when investing in large infrastructures such as particle accelerators.
A third issue concerns the role of scientific leadership. If after getting good advice a research funding leader decides that a particular research area is important and should receive more support, rather than ring-fencing resources it would be more useful to undertake a process of education and inspiration of researchers so they become motivated to work in that area. Should the area really be as promising as the research leader thinks, then it should be easy to persuade high-quality scientists that there is interesting work to be done and as a consequence they will submit proposals to the normal response mode system. Should it not be so interesting, then high-quality researchers will be less impressed and will be less likely to be persuaded to submit proposals. In this case the research leader should perhaps think again whether his or her enthusiasm is well placed. Research leaders do need to be proactive but not by ring-fencing or micro-management of the research agenda; rather by educating and inspiring the research community.
Are there any other special features concerning decision making with respect to science closer to application? Science across the whole continuum shares many similarities and this includes the importance of supporting talented individuals with the ability and passion to get the job done. However, work closer to application is more likely to be multidisciplinary and may well require greater team work, not only covering more scientific disciplines but also activities outside science, including finance, market analysis and the law. It requires effort to get individuals from such diverse backgrounds to work well together, and attention needs to be paid to encouraging mutual respect and to breaking down barriers between them. This would be encouraged if there were much greater permeability between sectors encouraging the transfer of both ideas and people more freely. We have in place too many barriers and silos that inhibit free transfer and encourage suspicion between the very people that need to be working closely together. One of the problems is that increasing knowledge has led to specialization, making interactions between different scientists, industry, the public services and other professions more difficult. It was easier to make such contacts in the less complex society at the time of the industrial revolution. Take the Lunar Society, for example, made up of chemists, biologists, doctors, industrialists, engineers and social reformers, regularly meeting every month to talk and to exchange ideas. This included intellectuals and entrepreneurs such as James Watt, Josiah Wedgwood and Erasmus Darwin. It was in this atmosphere that the industrial revolution was born, and we need to reproduce it again today. Greater permeability should be promoted, starting with the young by giving them wider intellectual exposure during higher education and their research training. They need more diverse placements early in their careers with easy exchanges between sectors at all career stages. This is a key message: the promotion of translation and innovation requires good permeability across the sectors.
Much is spoken about the ‘valley of death’, the gap between the generation of new knowledge and the application of that new knowledge particularly for commercialization. Usually the focus of discussion is on providing research support to bridge that gap, but attention also needs to be paid to pushing the bridgeheads further out into the valley. There can be a problem when attempts to translate are made too prematurely before knowledge is sufficiently reliable and complete, especially in the biosciences given the complexity of living organisms. To rush into translation runs the risk of becoming lost in translation. A firmer bridgehead needs to be built involving a more extended and secure knowledge base in the area of interest before attempting to pass over the valley of death. Similarly, the bridgehead on the other side needs to be extended out, with more investment from industry in research aimed at capturing new knowledge from the other side of the valley. Without research capacity and knowledge in industry it will be difficult to build back over the valley of death.
I was hoping to finish this Anniversary Address without mentioning impact. Researchers want their research to have impact, to increase knowledge, to contribute to culture, to generate societal benefit, to support the economy. Problems come when crude metrical applications of impact are made a compulsory part of research funding decisions and assessments. The potential impact of research should be clearly identified if it makes sense to do so, but often it does not make any sense to do so. To demand a statement in every research proposal or assessment about impact for societal or economic benefit will often simply result in unhelpful flights of fantasy of no value. Impact is just one aspect out of a number of factors that need to be considered when assessing a research proposal, and should be provided when relevant and not at all if irrelevant.
Funding high-quality research will produce the scientific knowledge needed for the public good, including driving innovation in support of the economy. If we get it right, science can play its proper role for the benefit of society, but getting it wrong will only waste money and lose the great opportunities that science can provide to improve the lot of humankind.
- This journal is © 2012 The Royal Society