On stories and science
in the wonderful world of politics

Michael Laver, Trinity College Dublin
4. november 2003

The previous people giving this lecture were eminent scientists. I am not. This lecture is named after world-famous storyteller. I am not one of those either. My name is Michael Laver and I am a political scientist. What I want to do today is think aloud about some of the differences between science and storytelling. And about what these tell us about how best to analyse the social world in which we humans all live. My particular focus will be on the wonderful world of politics, which is what I have studied for the past thirty years or so.

The main avenue I explore leads us to think about the implications of the fact that every story depends upon its teller, its Hans Christian Andersen, while science does not, and certainly should not, depend upon who it happens to be who tells us all about it.

There is, for better or for worse, an academic discipline called "political science". But the name of this discipline begs the very question I want to ask. This concerns whether there is any chance at all that we humans can study politics in a way that most people would recognise as "scientific". My answer is that we can indeed do this if we try hard enough, but that too much of what currently passes itself off as political "science" is in fact storytelling wolf dressed up in scientific clothing.

It is as well to be crystal clear about the intellectual baggage I bring to this argument. Personally, I do believe we never cease to find systematic regularities in the social world in general and in the political world in particular. These worlds are of course never completely determined. But only rarely are they completely random. Indeed people who behave as if the social world is completely random tend to be regarded by the rest of us as insane. When intellectually curious people see systematic regularities in the world, they try as hard as they can to make sense of these. In many ways this is what it means to have an intellectual inner life. Making sense of regularities involves building theoretical models that are more general than any particular observation of the word we might make. A "theory" that can explain just one particular observation and no other is no more than a restatement of this observation using different words. It is not an intellectual tool that gives us analytical purchase on a particular class of observations we long to make sense of. What science does is link observed regularities in the world to carefully constructed theoretical models; and it does this using systematic methods that tend to produce the same results no matter who applies them.

It is this role of systematic methods in generating the same results in different hands that distinguishes science from other types of intellectual activity. It sets science apart from storytelling, for example, since a story is a way of putting words to an interpretation of the world that is a personal construction of the story's teller. Faced with the same world, different people tell different stories about it. Faced with the same story, different people draw different interpretations from it.

Storytelling is not a completely random process, of course. Shared languages and cultures, almost by definition, allow different people with something in common to draw some common thread from the same story. Great storytellers have a genius for sensing and manipulating these linguistic and cultural resonances. Take the example of Italy's Hans Christian Andersen - Carlo Collodi, author of Pinocchio - whose stories have a social meaning for Italians quite different from the meaning they have for other people. The key point in this context is that, if Carlo Collodi had never written Pinnochio, it is almost certain that nothing quite like Pinnochio would ever have been written. Umberto Eco's Pinnochio, for example, would have been a different wooden doll entirely. A story, in short, has an auteur, a person who gives it life, just as Collodi gave life to Pinnochio. Without Collodi there would have been another story altogether.

Many might argue that science is no more than a special type of storytelling, claiming that it is impossible to find a way of interrogating the world that tends to produce the same results no matter who does this. This argument can be made at a level so general that it is impossible to refute. I measure the width of a window using a ruler, for example, you measure the same window using the same ruler. But it could be said the "meaning" of that "same" distance is different for the two of us. For you it may seem large, for me it may seem small. I regard this type of argument as true in its own terms but ultimately unproductive. For me, some might say a rather prosaic type of person, what is important is that I can measure the window, send that measurement to anyone else anywhere in the world and, if I measure the window carefully and if they do their job carefully, they can make a blind for me that will almost certainly fit that window. For me, to argue anything else about the measured width of my window is the intellectual equivalent of doing crossword puzzles. It is not wrong in itself. It is without doubt an excellent way to keep the brain in limber and an activity at which some people can become astonishingly proficient. But it is ultimately an intellectual pastime sans issu.

Many might argue that this way of thinking sidesteps the key issue in relation to the possibility of a social or political science because, while windows are windows, people are different. They might argue that people think, that different people think differently, that how people think affects how they act, and thus that my account of some particular social action essentially depends upon my own interpretation, inevitably subjective, of how other people think. On this line or argument, other analysts could easily come up with different interpretations of the same social world. I agree that windows do not think the same way as Hans Christian Andersen's darning needle, cute precisely because it thinks like a human.  But don't agree that this makes social science impossible.

Some people do feel hungry and they would rather not. Some people do feel that the government making important decisions on their behalf takes no account of their opinions, and they don't like this. To argue that such people do not actually feel hungry or disfranchised, that this is just my personal view of their state of mind, seems to me to be fatuous. It is also beyond reasonable doubt that things can be done to change these states of the world. Different people with different minds come together to stage bloody revolutions or engage in systematic programmes of social reform. These processes are not random. They are systematic regularities in the social world. We can construct systematic models of these regularities. These models imply that particular stimuli tend to generate particular effects. Changing the auteurs of these models does not change their effects. Both reform and revolution are indeed possible and neither depends ultimately on any one auteur to give them life. Thus, for me, social science is possible and to argue otherwise is to set angels dancing on the head of a pin. That is the baggage I bring to this talk. If others reject it out of hand then I can do no more than bang the table a little harder.

Now it is vitally important to emphasise in all of this that I am not arguing for one single second that political "science" is good while political "non-science" is bad. Nothing could be further from the truth. A story, whether it is Orwell's 1984, Kafka's The Castle, or indeed The Emperor's New Clothes, can offer us profound social and political insights. The same can be said of a work of graphic art, whether Picasso's Guernica of Warhol's Campbell's Soup Can. Take the opening lines of the poem, Lough Derg by Patrick Kavanagh, dealing with religious practice in rural Ireland:

From Cavan and from Leitrim and from Mayo,

From all the thin-faced parishes where hills

Are perished noses running peaty water,

They come to Lough Derg to fast and pray and beg

With all the bitterness of nonentities, and the envy

Of the inarticulate when dealing with an artist.

Their hands push closed the doors that God holds open.

Love-sunlit is an enchanter in June's hours

And flowers and light. These to shopkeepers and small lawyers

Are heresies up beauty's sleeve.

A social scientist could take a million years to say the same thing less well and indeed, on reading a poem judged by literati to be far from Kavanagh's best, might be tempted to opt for early retirement. But, and this is a big but, Kavanagh's model of the social world is a story told by him, however potent its impact. If Kavanagh had never been born, then nobody would have told this precise story. It would never have existed in anything vaguely like the same form even though the social reality to which it refers would have been unchanged. This particular story depends fundamentally on its auteur. It is absolutely none the worse for that and indeed its very idiosyncrasy lies at the heart of its intellectual power.  But it is not science - at least I have yet to meet anyone who claims that Patrick Kavanagh was a scientist.

So I am certainly not here to turn all forms of social analysis into science, in the process making the world a far drearier place. What I argue is that it is nonetheless possible to be scientific when we study society; that doing this can help us to understand, and therefore to act upon, important social problems. This is hardly an ignoble objective. My cross-hairs in this talk are thus not trained upon "non-science" at all, they are trained upon bad science. Cloaking itself in the mantle of the scientific method, there is alas much bad science in the social "sciences" and to my mind this is worse than useless. The trouble with bad science is that its scientific aura seduces us to believe in, even worse to act upon, its conclusions, despite the fact that these ultimately lack any rigorous basis.

Political science and non-science part 1

I now take a little longer to explore the respective roles of the auteur in science and non-science. In the past when thinking about this I have used the example of Warhol's Campbell's Soup Can. Looking at this image, does it make any difference whether it is signed by Andy Warhol or by an advertising agency working for the folks who make Campbell's Soup? The two identical images have different auteurs, and I suspect for many people this makes all the difference in the world. Laver's Campbell's Soup Can, even if visually indistinguishable from that of the master, would be seen as little short of pathetic.

Since I am in Odense, I leave the Campbell's Soup on the shelf and focus instead on the stories of the Danish Andy Warhol, Hans Christian Andersen. When you read Thumbelina, The Ugly Duckling or The Steadfast Tin Soldier, you are reading a story that was "there" in some important sense long before Hans Christian Andersen was born. So why are these Hans Christian Andersen's fairy tales? We think of them in this way because they are his distinctive telling of what are in a sense universal stories. Andersen is remembered both because of the universal nature of "his" stories and because of his own distinctive and powerful telling of these. The same stories would be quite different when told by someone else. In my book this makes Hans Christian Andersen not a scientist but an auteur, a teller of stories.

Contrast this with our current understanding of the relationship between energy and matter, captured in the famous equation e = mc2. Many people associate the name of Albert Einstein with this equation and credit him as a great genius for coming up with it. But, if anyone else in the world had come up with e = mc2, quite unlike a book of fairy stories, it would have been exactly the same equation and no less right or wrong. Einstein was not the auteur of e = mc2, which is not his own personal "spin" on the relationship between energy and matter. He was simply the first man with the deep insight to formulate a theory with profound practical implications and is justly remembered for this. It is "his" theory in this sense but its profound implications depend not one iota upon the name of Albert Einstein. Hiroshima and Nagasaki would have happened even if Hans Christian Andersen had first come up with e = mc2.

This is even more strikingly true for many of the recent advances in information technology. My father, an old computer hand, associates the invention of the computer with Lady Lovelace (1815-1852). Others might name Alan Turing as having had the first real vision of what a modern "universal" computer might be able to do. But the current state of the art in IT has come to pass as the result of what, standing back from it all, is truly a mind-boggling collective endeavour by a huge group of scientists and engineers, working to more or less the same rules as part of a global scientific community. Any one person who claimed authorship over all of this would be regarded as a worthy of no more than our pity, since the whole concept of authorship does not arise in this "scientific" context. Things happened. People made discoveries. But none of this had an auteur.

In case there are people in the audience getting ready to riot because they have heard nothing yet about political science, perhaps I should start talking about things I can at least pretend to know something about. There is a useful term in French - politologue - that seems to translate well into other Latin languages. Roughly, a politologue, is someone who writes or talks about politics. A politologue has a name, and this name is an important professional resource. Political analysis from a politologue is taken seriously because it is Professor X's political analysis. Professor X is a wise and experienced person with a considerable reputation - a person who knows people, who hears things, who has the inside track. Professor X may be talking perfect sense but he is not talking perfect science, because we need to know Professor X's name before we can decide whether or not to take seriously what it is that is being said.

Many political "scientists" are politologues, at least in their spare time. I myself, for example, am sometimes asked to appear on TV or radio, often to talk about subjects about which I know next to nothing. If I reveal that I know next to nothing about some subject, and suggest an excellent graduate student working on precisely the right theme, there is always a weary sigh from the media types and never yet in my experience has that graduate student hit the airwaves. Frankly, though they are never explicit about this, the media would rather have a "name" talking on something about which he is ignorant, than a "no-name" talking perfect and well-informed sense. This is not science. It is asking the Oracle - and we like to ask Oracles with fabled names because these names reassure us. The words say the same thing, they just don't have that swing when the speaker has no name.

My past connection with Odense is not with Hans Christian Andersen but with Mogens Pedersen, with whom I edited a political science journal for some time. One thing I like to think about those times, and about how things are done by other good scientific journals, is that the humblest graduate student, indeed the most ordinary person in the street, could have published an article and have had an impact on the profession if his or her work had come up to the required standards. And what are these standards? To be honest, I have no systematic answer but I do know that the earthly manifestation of "coming up to the required standards" is passing a process of anonymous peer review. The name of the author is removed before a submitted piece is sent out to reviewers, and in a good journal every self-referential statement is also removed.

Of course nothing is perfect and there is a Catch-22 paradox at the heart of the process of anonymous peer review. The best reviewers for any given paper will be so familiar with the cutting edge of the field that they will recognise the "anonymous" work of the paper's author. But at least in spirit, and quite often indeed in practice, the scientific value of anonymous peer review is that the arbiters of whether a piece of work is "good enough" to publish in a scientific journal make their judgement in ignorance of the work's authorship. This is science and it is the polar opposite of a critical review of a new piece of art, music or literature, which typically revolves around the name of its auteur.

Science, by my lights therefore, is among many other things an analysis of the world that people can evaluate as being good or bad without knowing, or even feeling an overwhelming need to know, the name of the author. If you crucially need to know who the author is before you can evaluate it, then the work in question is non-science. Political "science" can in my view be science in this sense that if it tries hard enough. Sadly, however, political science does also have its pantheon of "great names", who command attention by virtue of their very names and reputations rather than by virtue of their essentially anonymous contribution to a larger collective enterprise.

Interestingly, we can cut this cake according to whether work is published in scientific journals or in books. Great names tend to write great books rather than great articles and this trend becomes stronger the grander these authors become. This may be in part because great names in the profession become increasingly irked by the need to submit themselves to the criticism of no-name, and they suspect far more junior, reviewers. In part it may be because the economics of book publishing put a premium on the name and reputation of the author. In practice, furthermore, even proposals for quite serious and boring academic books tend to be reviewed in the light of the identity of the author. Books almost invariably have auteurs, and they thus are not part of science in the strict sense. One interesting indicator of the scientific status of a given profession is the extent to which its cutting edge work is published in journals using strictly anonymous peer review, as opposed to being published in books. In this sense political "science" is indeed becoming more scientific as it comes of age.

Political science and non-science part 2

I now set out to sneak up on the same target from a rather different direction, considering the ability of one person to replicate the work of another.

Recall that part of the overall ethos of science is that results can be replicated by different people using the same methods. It might take an extraordinary genius to come up with a brilliant new idea that has profound implications for the real world. Once this idea has been placed in the public arena, however, it is open to anyone at all to observe these real world implications for themselves. In the world of political science, this argument has an eloquent advocate in Gary King, whose famous and cunningly-titled piece "Replication, replication" makes the point clearly and convincingly. [1] On this point you would do better to read King rather than listen to me.

Many of us had the replication standard dunned into us in school science labs. Follow the instructions carefully, replicate the experiment, observe the predicted effect. Of course the real world is replete with random events so things are never quite as simple as this. If you follow the instructions carefully, however, and replicate the experiment then you should tend to observe the predicted effect.

A scientific community of scholars pays attention to surprising new findings (whether concerning cold fusion or perpetual motion) only after these have been replicated. And replicated again and again and again. No scientific progress has been made on the basis of an effect that cannot be replicated (once more, setting replication in an appropriate probabilistic framework). This is one of the things that make science science. Inability to replicate results is one of the hallmarks of non-science.

Political "science" in particular, and social "science" in general, both sadly still have a long way to go in this regard. There are two basic problems. The first concerns whether, if data and method are made available to all, anyone else can observe the effects being claimed. The second concerns whether data and method are made available to all in the first place.

Take the second problem first. We are all familiar with stories, and to be honest I don't know if they are true of not but they do make the point, about historians or specialists in some branch or other of literature who are driven into paroxysms of ecstasy by the discovery of some hitherto undiscovered "secret" source. What do they do, at least in the stories? Do they publish this source immediately and allow the entire community of scholars to have at it? They do not. They mine it privately for as long as they can, since their scholarly reputation depends in part upon access to information that is denied to other scholars.

In the political context, the politologues I discussed earlier often cherish and even fiercely "protect" their private privileged access to information and to people in the know. Indeed an eminent politologue may well be eminent precisely because of such privileged access.

All of these people are doing something that, by definition, others cannot do because the necessary resources are withheld. All of this is, including much that currently passes under the misnomer political "science", is therefore non-science. Again I should emphasise that this does not at all mean that such non-science is inevitably nonsense. It implies simply that it has no claim to the virtues of the scientific method. If people act on the basis of such work, then they are acting under the influence of an Oracle they trust, but whose analysis they cannot reproduce for themselves.

One welcome development, arising in part from the funding of "big" social science by agencies that follow the traditions of the natural sciences, is that new data whose collection is funded by public sources must be archived within a set period and made publicly available to all. Gary King's Virtual Data Center project, making such archiving very much less burdensome, will hopefully speed this process on its way. Nonetheless it is a process still in its early stages for political science. It is still too easy to find colleagues who have a strong sense of ownership of what they see as "their" data. These data can be difficult and sometimes impossible to prise from the hands of an author who has published a paper making an argument in which you are interested and that you want to explore for yourself. But, if you publish something but won't open up your data and method to me, there is no reason why I should take you seriously. You have in fact set yourself up as an Oracle, since I can do no more if I want to take you seriously than put my trust in your personal reputation. As we have seen this is non-science and there is still far too much of it to be found in political "science".

We all, as individuals, have to wrestle with this problem once we have sweat long and hard over the collection of new data and have proudly published analyses of these for all the world to admire. There is an auteur in all of us, however hard we deny it. That auteur does not like other people criticising his or her beautiful story, perhaps even constructing a story using the same raw materials that others like even better than our own. The auteur in many people tends to take criticism personally. A good scientist, however, shoots his or her own personal auteur every morning before breakfast. If there is a weakness in some analysis, it is better for science that this be opened up as wide as possible, neither obscured by withholding the means to do this nor spun with the rhetoric of an ever more elaborate story. Science very often progresses by opening up mistakes in the work of others, mistakes that cannot be opened up if they are not made in the first place.

And, it must be said, even the most vain of scientists has nothing personally to fear from this way of doing things. Did Newton or Einstein or Galileo make mistakes? Of course they did but only idiots think less of these great scientists as a result.

We can now return to the first problem of replication. I do have your data and your method but can I replicate the salient effects you observed? This is where, sad to say, much of what passes itself off as political science falls far short of anything we might regard as ideal. Gary King, for example, has shown that frighteningly few of the findings in articles published in the world's most prestigious political science journals could actually be replicated using the authors' own data. This problem has two facets, one banal but nonetheless worrying, one more profound.

The banal problem is best illustrated by thinking of two people analysing precisely the same data using precisely the same computer program. Even in these circumstances it can happen that the two get different results. How could this possibly be? The cause has nothing to do with sunspot activity deranging computers and making these sometimes come up with different answers to the same sums. Rather, it arises because political "scientists" too often have little or no training in how to behave like real scientists. Again, those casting their minds back to school science labs will recall the ever-present laboratory notebook, in which careful records were kept of every operational decision taken along the way from the design of the experiment to the observation of the consequent effect.

To take a simple example, if we want to know the relationship between social class and voting in elections, we do not just get our hands on a social survey and type the question "what is the relationship between social class and voting in elections?" into the computer. (Although I swear I have indeed seen students in undergraduate methods classes doing precisely that.) The answer to this hard question is torn from the data at the end of a surprisingly long sequence of detailed decisions. Some of these decisions may seem incredibly trivial at the time; others may be made almost without thinking about them. A "bad case" may be excluded for some reason or another. Certain weights may be applied to certain cases. One of several similar variables may be selected for use in a particular context. Indices may be constructed. Variables are recoded and otherwise transformed. Particular settings are used for particular methods. And so on.

I have personally had the experience on more than one occasion of failing to reproduce some widely-cited published result using the author's own data, only to find that the author did not keep a record of all operational decisions taken en route to the final destination, which has become effectively irretrievable, even from the dataset that generated it. This is certainly not good science and is something we all need to pay much more attention to. But political science can indeed do this if it does pay close enough attention. For this reason I heartily commend the laboratory notebook to all who think of themselves as political scientists.

The more profound replication problem is more intimately connected to social science in general and political science in particular. The experimental method is one of the hallmarks of many though not all sciences. And the whole point about experiments is that they can be repeated. History, however, including political history, can never be repeated. This is quite a problem for a political "science", even one doing its very best to be scientific.

For most of my career, for example, I have worked on the making and breaking of national governments in Western Europe. I have developed theories, models and data in this area and have evaluated the theories, models and data of others. However, taking the period since the end of World War 2, only a finite and quite small number of national governments have been made and broken. There is only one dataset of post-war Western European governments, a dataset almost as old as modern political science itself and one being added to at only a glacial rate. Those working seriously in the field have inevitably internalised many of the features of this dataset, whether they realise this or not. So, if some new theory or model seems to work well at explaining the data, we simply can't follow the instincts of the natural scientist, collect a whole new dataset (on Mars, perhaps?) and replicate the results. All we can do is wait 50 years for an equivalent dataset to accrete, but 50 years is longer than the careers of 99.9 percent of human political scientists. I do seriously worry that, in this sub-field at least, it is increasingly difficult to be scientific because the existing data have been mined out and new data are accumulating too slowly to be of significant use. Lots of results have been published over the years on the making and breaking of governments, but scientifically replicating the important effects that have been observed is a lot easier said than done.

This is a particularly sharp facet of the problem but it is a problem with many facets. Take survey research, for example, one of the data generating techniques often considered to be at the scientific end of the discipline. Large scale, carefully run, mass surveys are expensive, yet are very important ways of gathering systematic information about social behaviour and attitudes. Finding funding for a major election study, for example, which generates one of the baseline data resources for political science, is a perpetual headache for people in this line of work. It is easy but not pleasant to imagine the stunned response of a hard-pressed funding agency asked to support several simultaneous election studies in the same country - conducted precisely to see whether all of these would generate the same results. This is not going to happen.

Replicating general results using different independent datasets is never going to be easy for political science given that history cannot be put into reverse, but this is no excuse for pretending the problem doesn't exist. Those with good scientific instincts can still submit their existing data resources to sensitivity analyses, in order to get a feel, for example, for the extent to which particular random subsets of the data generate different results. An election study could be analysed in this way by randomly splitting it in half many times over, to observe the extent to which the effects observed depend upon which random half of the data is analysed. Political science could do things like this if it tried harder. In my experience it does not currently do them very often.

Many replication problems in political science can laid at the feet of the profession as a whole rather than those of individual researchers. Kudos in the profession go to people who do new and exciting things and build their reputations on the back of these. Carefully replicating the results of others, indeed even carefully replicating one's own results, is seen as lower grade work altogether. Yet replication, for the die-hard scientist, is the essential starting point for any new work. Only then is it possible to know in a systematic way how far new work has advanced old ways of thinking. Political science has a long way to go in changing its system of professional rewards and punishments if it wants to make serious scientific replication much more common.

I am now going to break the conventions of giving a public lecture, not by putting on the Emperor's new clothes you will be relieved to hear, but I will press on with some new thoughts rather than summarising what I have already said, which was simple and obvious enough not to need summarising.

My main point so far has been that work claiming the virtues of science should at least be scientific. I want to close this talk by exploring the relationship between what is usually thought of as the "normal science" that characterises 99.9 percent of typical scientific activity and the major breakthroughs and paradigm shifts by which science can generate work that is as exciting, as challenging and indeed as mysterious as the greatest work of art. As with great works of art, such breakthroughs typically depend upon brilliant iconoclasts and eccentrics. In this sense science and non-science are united in the way they are ultimately advanced by untamed, controversial and often unconventional creativity.

Political analysis has indeed enjoyed new paradigms that have unleashed waves of intellectual activity and output. We can think of Marxism, systems theory, behaviourism, rational choice and critical theory as obvious examples. Those working at the cutting edge of a brand new paradigm, whether in art or in science, are typically just Beyond the Pale of the existing intellectual empire. They have often left the existing Pale as renegades from what they see as its unrelenting intellectual poverty.

Most of what I have said so far has dealt with normal science, not with the paradigm shifts that cause our understanding suddenly and excitingly to race forward in leaps and bounds. But such paradigm shifts are as much part of science as is normal science. Without them, the known intellectual world would become quickly mined out and unproductive. How can paradigm shifts be fostered and encouraged?

Nobody has ever answered this question in a convincing way; it is probably one of life's great mysteries. But there are a few observations we can make about it, and these observations are as relevant to social sciences as to anything else.

The first is that, in the sciences at least, good normal science does also foster paradigm shifts. While ambitious academics often forget this, good normal science, carefully conducted and reported, tells us more about what we don't know than about what we do know. Knowing what we don't know is good. It is trying to gloss over what we don't know that is bad, although again journal editors and article reviewers in political science need to try much harder to keep this close to their hearts when deciding what, and what not, to publish. The better it does its job, the more normal science hammers home what it is that we don't know, thereby setting up the intellectual challenge for new paradigms. With bad science, we have no real idea about what we don't know, and can thus fecklessly amble along the road to nowhere as if nothing at all is the matter.

My second and I promise final observation on this matter is less pious and more practical. It has to do with the funding of basic research. Once upon a time, many great artists and writers and scientists enjoyed either private means or private patrons. These provided the funds that allowed them at least not to die of starvation while single-mindedly pursuing what fired them with intellectual enthusiasm, whether or not this had any immediate practical application whatsoever to the real world. Today, the patron of many scientists is the state, funding scientific research that is very often carried out in universities which, when they operate as research institutes, have become the modern equivalent of the enlightened and curious aristocrats of the renaissance.

There is nothing wrong with this but there are without doubt ugly straws in the wind. The public sector is not so popular these days as it once was and is increasingly judged according to "value for money" principles. While value for money may not be a bad thing in itself and while money is easy to count, the value of basic research, almost by definition, often shimmers just beyond our grasp. Basic scientific research very often involves doing things that have no immediately obvious application to the real world, yet such basic research is the intellectual driving force of any serious scientific community. Applied research tends to apply what essentially we already know; basic research tends to explore uncharted territory. The danger in a "value for money" approach to the public funding of scientific research is that this will favour the more easily evaluated applied research and starve basic research of the resources supporting those who typically have a admittedly tiny chance of making simply enormous scientific breakthroughs.

Allow me to spout just a tiny trickle of actual political science before I sit down. The funding of basic scientific research runs the risk of turning into something of a collective action problem. Since the sciences operate as global intellectual communities, the "value" of basic research, when it materialises, is global - and these days is almost instantly disseminated around the globe. It is a common pool intellectual resource. Those who pay for it pay all of the costs and enjoy only a small share of the benefits. The benefits of applied research, if not completely privatised, are nonetheless much more sharply focussed. This allows funded research programmes to be defined in ways that direct their benefits towards those who pay for them. When times get hard for public funding agencies, it is easy to see why this particular manifestation of the collective action problem may put basic research under the cosh. "Let others pay for the basic research (sotto voce: from which we will nonetheless benefit in the unlikely event it strikes gold). We will pay for applied research that yields enough measurable benefit to allow us to survive a value for money audit."

There is no need for a crystal ball to see whether we will live happily ever after if every funding agency thinks the same way. Any tendency on the part of the modern world's great patrons of the sciences towards funding applied research at the expense of basic research runs counter to the spirit of real science, reducing the probability of unimagined intellectual breakthroughs. Whether we are political scientists or astrophysicists, we must take up the cudgels and fight this evil spectre with all our might.

This was all of course a story; it was not science. Most of you would walk away with a difference message had the very same words been recited by a 21-year-old actress rather than by a man with a grey beard. But both of us do hope that you all live happily ever after.

[1] King, Gary. 1995. Replication, replication.
PS: Political Science and Politics. XXVIII:3 443-449