Science, we often hear, is too important to be left to scientists. It is a daft thing to say, really. Scientists undergo long and arduous training, work implausibly hard and take specialisation to a degree rarely matched in other disciplines. If you want science done at all, leaving it to them seems an excellent idea.

The argument that others need to get involved still has real force, though. Science and the technologies it helps into existence are highly consequential. Those consequences are the sort of thing we all feel entitled to have opinions about. But in the face of all that training, and the esoteric knowledge that goes with it, how far should the opinions of the public count? What, in short, is the place of expertise in democratic discussion?

This unresolved - perhaps unresolvable - question lies behind many of the efforts over the past 25 years to improve public understanding of science (more recently rebranded as "public engagement"). A fascinating variety of initiatives now bring scientists together with laypeople, including consensus conferences, citizens' juries and various other orchestrated dialogues, on and offline. All count as small good deeds in a world where deliberation is often drowned out by declamation. But they sometimes seem to proceed from a simple assumption that "it's good to talk". That is true, but is it enough?

In that light, it is bracing to encounter a rigorous effort to explore how high the stakes are, to define the most exacting standards we could apply to discussions of how, when, where and why research gets done, and what we should make of the results. Philip Kitcher, the expatriate British philosopher of science who now teaches at Columbia University, does this in Science in a Democratic Society (2011). He considers the tension between science and democracy as an inevitable outcome of any epistemic division of labour. And he explores what a genuinely democratic discussion of science would look like. That exploration is driven by two concerns: one is that "the conception that guides our public rhetoric about democracy is the most superficial"; the other is a wish to establish a proper role for scientific authority, a task made harder, he argues, by a common insistence that genuine science is value free.

ÖйúAƬ

He proposes that the value judgements that are in fact always being made - for example, on the balance between satisfying our curiosity and serving human welfare, and on which scientific topics should have priority - be discussed as widely as possible. How widely is that? Clearly, it goes further than simple representative democracy can do. But the issues that arise in complex societies mean that fully participatory democracy is out of the question: it just takes up too much time.

So our debates are going to be staged in the broad deliberative space between these two extremes. What might their ideal form look like? Generally, Kitcher wants a "well-ordered" discussion that requires representatives of all points of view and a commitment to mutual engagement. Those general conditions expand for a well-ordered discussion of science. The issues in play often affect everybody on the planet and their descendants, so the range of interests that need to be represented is unusually wide.

ÖйúAƬ

That is hard enough to envisage. But then comes the tricky part when scientific experts have knowledge the rest of us do not. The discussion should work towards what Kitcher calls "tutored opinions", "free of misapprehensions that contemporary science can correct".

That sounds as though it would give scientists privileges in debate that some people are reluctant to grant, although citizens' juries and consensus conferences, in which scientists answer laypeople's questions in conversation, can contrive something approaching the tutored opinions Kitcher seeks.

Equally interesting, though, is what he demands from the research system in return. In particular, he suggests that "well-ordered" science, in his sense, means that the agenda for scientific research is in line with judgements reached under the conditions he defines for ideal discussion. Research choices, in other words - at least at the level of the questions addressed - are definitely part of the necessary democratic discussion.

This already happens in a narrow sense in some of the deliberative exercises mounted in the name of public engagement. What is still missing, I think, is any real effort to discuss research priorities across the board. Such a discussion, Kitcher suggests, would need what he calls an "Atlas of Scientific Significance". Mapping the promise of different research topics and cross-referencing it with a catalogue of human needs would be a pretty daunting task. The current fashion for throwing out "grand challenges" in a range of areas from global health to defence technology and computing is a gesture in this direction, but Kitcher has in mind something much more comprehensive - and more widely discussed.

In conversation, he emphasises that he sees his role as a philosopher as proposing an ideal in the hope of starting a conversation about how far to move towards it. His own ideas in this area mostly draw on the kinds of exercises I have already cited - consensus conferences and deliberative polling, for instance. But how far could we really go in extending the democratic discussion about science?

The latest UK evidence from the surveys carried out for the Government Office for Science shows that people favour public discussion about science, but mostly want others to do it. Fifty per cent of those polled indicate that they would like to know that the public are consulted on science issues, but only 7 per cent want to get involved personally.

There's also a quite widespread feeling that "experts and not the public should advise the government about the implications of scientific developments", with 64 per cent agreeing or strongly agreeing. On the other hand, nearly 75 per cent agree that "the government should act in accordance with public concerns about science and technology".

Both sets of wording raise lots of questions, of course, but it looks as if, on the whole, people want the government to take their interests into account, but judge (correctly) that direct involvement with issues with a technical component would be pretty time-consuming.

ÖйúAƬ

That seems to limit movement toward one simple notion of democratisation: involving more people. However, it leaves scope for expansion in the other dimension - the breadth of the research agenda that is discussed. At the moment, the exemplary efforts focus on a few areas where the need for discussion is perceived as pressing.

A recent newsletter from Sciencewise, the UK's national effort to promote dialogue about science, applauds "opportunities for the public to contribute to the development of policy on emerging areas of science and technology", including geoengineering - trying to tackle climate change by deliberate intervention in the atmosphere - and synthetic biology - making designer life forms.

ÖйúAƬ

Those would be high on most lists of potentially controversial research areas. Other recurrent topics include genetically modified foods and the hazier area of nanotechnology. But you will not find invitations to consider the overall research agenda, the kind of thing that could help compile Kitcher's Atlas of Scientific Significance. Scientists tend to shy away from that sort of thing, a defensive response to what they hear as a threat to their autonomy. That is understandable in the face of funders' ever-present demands to tailor research to applications sought by the government or industry. But we have little idea, really, whether national research priorities bear much relation to public aspirations, let alone to global needs.

An exception is a recent critique of the European Commission's multibillion-euro Framework Programme from an impressively large coalition of non-governmental organisations - a rare example of direct confrontation between framing a research agenda tightly in terms of a fairly narrow concept of economic development and a different set of values. However, it is a bit of a compendium of third-sector clich¨¦s, suggesting that the EU framework should be "geared towards the needs of society and the environment rather than those of big business". The argument about why these needs are different proceeds by assertion.

The NGOs go on to say that work on "nuclear energy, pharmaceuticals, agricultural genetic engineering, synthetic biology, nanotechnologies, space and military research" involves generous public subsidies for the aforementioned big business. On the other hand, the ungainsayable goal of making the world "an environmentally sustainable, healthy and peaceful place to live" really needs research in "environmental protection, preventative health policy, organic and low-input agriculture, energy saving and renewable energies, toxicology, water supply issues, and environmentally sustainable fisheries, as well as for research in social sciences which contributes to social change and problem-solving that are not focused on technological fixes". One could find plenty to disagree with there, but it is at least a contribution to a conversation that is not happening otherwise.

So the challenge Kitcher poses, in my reading, is whether we can have a real discussion that would begin the large-scale mapping his atlas would need. That would require us to move beyond the fragmentation of the current discussion of global research priorities, which basically goes: yes, it is a bit embarrassing that most research in the West ignores the health problems of half the globe; there does seem to be an awful lot of defence research going on; it would be really, really good to have some security of energy supply; and we want a better idea what our current energy system is doing to the planet.

A serious effort to go further could be an effective vehicle to promote further public engagement with science. Any takers?