Brexistential Angst and the Paradoxes of Populism: On the Contingency,Predictability and Intelligibility of Seismic Shifts

The Predictability and Intelligibility of Brexit

The parable of the time-travelling political analyst suggests the need to draw a clear distinction between predictability (the capacity to anticipate), on the one hand, and retrospective intelligibility (the capacity to make sense of something after the fact with already existing analytic resources), on the other. Put bluntly, the capacity to anticipate a social fact is not a good (nor, indeed, a legitimate) test of social scientific theory; the capacity to render a social fact intelligible retrospectively is. From such a perspective, the failure to predict that Brexit would happen is not, in and of itself, a theoretical failing in that Brexit is not knowable in advance of the crystallisation of the specific causal sequence bringing it to fruition. But the incapacity to make sense of Brexit in causal terms after the fact of Brexit is a failing (the failure of a legitimate test or expectation).

In this respect, the failure of political scientists and political economists to anticipate Brexit and/or to accord to it a low probability as an outcome is different in kind to the failure of mainstream economics to anticipate the global financial crisis (though cf. Blyth and Matthijs, 2017; Hopkin and Blyth, 2018). For in the latter case, the charge (at least) is that mainstream economics does not have the conceptual and analytic resources and capacity (based, as it is, on essentially equilibrium assumptions) to render intelligible the crisis it failed to anticipate.

While our time-travelling political analyst may lack the analytic resources required to make sense of what, to her, look like twenty-first-century anomalies, the same is not really the case for her twenty-first century peers. For, arguably, each of the exceptional (and unforeseen) outcomes identified above can be rendered intelligible (and has now already been rendered intelligible) in terms of the (contingent) articulation, interaction and interplay of quite familiar processes (some, indeed, familiar even to a political scientist who learned her trade in the 1960s and 1970s). Among such factors are the following: (1) the growing regional and geographical differentiation of political cultures in a more dis-United Kingdom; (2) the increasing presence of non-majoritarian electoral systems in the UK polity and the ability to carry success from one electoral system to another; (3) the declining electoral appeal of valence politics; (4) declining partisan alignment and identification; (5) increasing electoral volatility; the ‘taint of office’; (6) the declining electoral appeal of valence politics; (7) growing divisions within the Conservative Party over Europe; (8) the increasing use of referendums to resolve positional issues; (9) growing anxieties about immigration and competition for jobs associated (rightly or wrongly) with EU enlargement and the free mobility of labour; and (10) growing economic inequality in a context of austerity and persistently low growth. If Brexit can be seen as a (potentially) seismic shift, then these are the fault-lines making it possible.

The point is that one can go a long way to explaining each and every one of the seemingly exceptional political outcomes identified above, Brexit included, in terms of the interplay of a series of broader political trends already widely identified in the existing literature. If this is right, and the ‘exceptional’ is in fact explicable at least retrospectively in more general terms, the impression of a new condition of radical contingency afflicting our polity and our politics may well be misleading and something of an illusion. It may well also prove rather transitory. At times, British politics has been quite predictable; at other times, less so. Today it would appear less so. Yet this is not because it is, in some fundamental way, more profoundly contingent than it has ever been. Rather, and more simply, it is that a variety of long-term processes and trends that political scientists would claim to understand quite well come together and interact to produce a conjuncture in which a great many major issues will be resolved together in an unusually short space of time. In this conjuncture, British politics will not just develop, but also potentially change decisively with enduring consequences.

Hobsbawm’s Choice: The Inevitable Contingency or the Contingent Inevitability of Brexit

A second window on all of this is opened by an intuitively appealing if paradoxical aphorism of the late, great and much lamented historian Eric Hobsbawm. It looks immediately like an antidote to the kind of argument that I have been making. The line appears only once in his published writings – a characteristically provocative essay on the use of counterfactuals in the history of Britain’s post-war reconstruction (Hobsbawm, 1993: 29). But apparently he used it a great deal in the seminar room, which perhaps gives me the licence to misquote it just a little. It goes something like this: ‘things turned out the way they did, and because they turned out the way they did they couldn’t have turned out any other way’ (see also Hay, 2002: 269). Recast in the terms of the present discussion, we get ‘the referendum on Brexit, the vote for Brexit and the Brexit to which it gave rise each turned out the way they did, and because they turned out the way they did they couldn’t have turned out any other way’ (on the contingency and inevitability of Brexit see also Thompson, 2017; Wincott, 2017).

The quote itself, and the implicit challenge it poses, is unsettling (intentionally so). But it is an altogether easier trope to trip from the lips of a historian (with the benefit of hindsight) than it is for a political analyst (who typically lacks such hindsight). Either way, it challenges directly the preceding argument. For it suggests that uncertainty is simply a problem of lack of knowledge – and, consequently, that if we knew more we would be able to predict better. That sounds intuitively plausible. But the clear implication is that prediction is both a legitimate and an epistemological test of a theory. The outcome (the thing that turned out the way it did), in Hobsbawm’s parable, is certain; it is just our knowledge of it that is lacking (it is only uncertain to us). As such, there is no condition of uncertainty in Hobsbawm’s ontological universe. Uncertainty, for him, is a purely epistemological phenomenon – epistemologically contingent not ontologically necessary. ³ The problem, as one can almost hear him adding, is that we are simply never capable of acquiring the knowledge that would be required to know how things will turn out in advance of them turning out the way they were always going to turn out. That, for him, I suppose is the paradox of (our) uncertainty, and it is certainly frustrating for the political analyst. It is reminiscent of the meteorologist who, when asked for a weather forecast for 2020, responds that the causal sequence that will produce the weather in 2020 does not yet exist and that, as such, no prediction is currently possible. The weather will be there to be predicted; just not yet!

While there is no problem with the meteorologist’s (convenient) retort, social and political systems are different in kind from natural systems – and the problem of certainty and uncertainty presents itself differently. By contrast, Hobsbawm I think is wrong. His position, perhaps more significantly, is an indulgence that I think a political analyst cannot afford; for it is a profound disavowal of the political. Politics is made; and at the moment of its making it is indeterminate. As another, no less renowned historian, Charles Tilly (1991: 86) puts it, ‘we now live in one of many possible worlds’. From this perspective, there are always multiple possible outcomes. That does not make all outcomes possible (alas) and it does not make the unlikely more likely (alas). But it means that our future is always uncertain – and uncertain ontologically (it could, and at the moment of its making, can be made differently) not (just) epistemologically (we do not know enough to know how it will turn out).

That is neither good nor bad, though it is the very condition I would contend of things being political. The implication is, again, that while we might hope to make sense of Brexit we shouldn’t expect to be able to predict it; and that, in turn, has (or should have) implications for what we might credibly say about the consequences of Brexit (a point to which we return presently).

But there is a second problem, at least in the context of the present discussion, of Hobsbawm’s aphorism. For the things we are interested in here (Brexit and its consequences) are still in play. As such, they have yet to turn out the way they will. And since they haven’t yet turned out that way, we remain in the dark as to the only way they could turn out. This reminds us again of the difference between the historian (concerned largely with things already resolved) and political analysts (concerned, certainly to a far greater extent, with things still in play). Yet the more one thinks about it, the more problematic even this is. For it is in fact rare for political analysts to be as heavily invested in processes that are still unfolding as they are with Brexit. We tend in fact to have more in common with the historians, typically sharing with them the comparative luxury of waiting for things to have turned out the way they will before we seek to make sense of them. With Brexit, the stakes are simply too high for that (certainly for analysts of British and EU politics and political economy); that means that we need to think very seriously about what we can legitimately expect to know about something still very much in the process of being made, the confidence with which we profess that knowledge and the appropriate caveats surrounding it.

That takes us well beyond our typical comfort zone. Yet if the value of the epistemological distinction that I have sought to draw between predictability and the rendering intelligible of things after the fact is accepted, then it might provide some help and guidance here. Above all it suggests that, if we cannot expect to be able to predict that which we accept to be political and (hence necessarily) contingent, the best we can do is to describe prospectively the field of outcomes that our current approach is capable of rendering intelligible retrospectively. In addition, and by way of commentary, we might perhaps also offer a tentative assessment of the distribution of relative probabilities across this spectrum of conceivable outcomes. But it is difficult to see how we can credibly do more. In short, those who failed to predict Brexit (and even those lucky enough to do so) need to be careful before telling us what its consequences will turn out to be.

To consolidate the point, it is perhaps worth considering Hobsbawm’s implicit epistemological test when applied directly to Brexit. If our general and collective failure to predict Brexit was simply a reflection of our lack of knowledge, it is surely worth asking what would we have needed to know in order to predict that an EU referendum would take place in the way in which it did (and not in some other way) and that it would result in the vote for Brexit that it generated (and not some other vote for Brexit).

Arguably, for things to turn out the way they did, many if not all of the following had to happen: ⁴

Arguably, that just about gets us from the General Election Campaign of 2015 to the outcome of the Brexit vote itself; to get us from there to where we are today, we need to add a few more similarly contingent factors:

Put together, this is already quite a list. Clearly some of these factors are the product of short-term, almost whimsical, decisions and may well turn out to be of little or no long-term consequence (Bercow’s exercise of the constitutional discretion afforded the Speaker of the House of Commons, for instance). But the point is that we simply do not know what will turn out to be significant in advance. ⁷ Brexit has been made through the dynamic interplay between the highly contingent and sometimes prosaic on the one hand and the rather more predictable and seemingly significant on the other. Knowledge of the latter can only get us so far. And in terms of the compound contingencies of the Brexit process to date, this is arguably just the tip of the proverbial iceberg – not least since it makes no reference at all to the negotiating stance of the EU27. But the point is surely made already. While some of these contingencies were clearly more predictable than others, it is difficult even to render intelligible, let alone to have predicted in advance, how things turned out the way they did. To do so would, in effect, be the equivalent of having placed the largest successful one-way accumulator bet in history. ⁸

The preceding discussion is intended to be chastening. My aim has been to remind us of, and to focus in on, the reasons for, the inherent difficulty of seeking to make sense of unfolding processes of potentially seismic change in real time. As I have sought to suggest, and strange though it may seem, this is not something that we do frequently and it is, arguably, something for which modern political science does not prepare us especially well. But that is most definitely not an argument for foregoing such a task. It is an argument for thinking seriously about our responsibilities as ostensible experts (purveyors of an expertise to others) in a context in which the use and misuse of that expertise is both highly contested and, crucially, stands accused by those on one side of a highly charged and polarised political debate of a disingenuous and misleading neutrality that disguises an essentially normative commitment to the status quo.

This is, perhaps, the curse of living and writing in Br-exiting times! It is also a challenge. In what follows, I return first to the making of Brexit and then to the potentially paradoxical character of the Brexit that Britain ⁹ is still in the process of making in the light of my own cautionary remarks. In so doing I seek to address the challenge directly, by exploring how such cautionary reflections might lead me (and conceivably others) to revise or temper what I (and they) might otherwise be inclined to inject into the public debate.

So Why Did Brexit Succeed?

Why did British citizens vote to leave the EU, albeit by a narrow margin, on 23June 2016? Four factors, I will suggest, are crucial, though really only one of them has been extensively discussed in the wider literature on the subject to date, which has tended to focus on the psephological variables and the composition of the vote rather than the context in which the vote took place (though, see the various contributions to British Journal of Politics and International Relations volume 19 (3/4), 2017; Diamond et al., 2018; Jennings and Lodge, 2018). They are (1) Cameron’s negotiating strategy – similar in certain respects to that of Harold Wilson in 1975, but playing out very differently in a very different political context; (2) differential turnout between potential leave and potential remain supporters (associated, in turn, with the differential mobilisation of those for whom the issue was a valence question as against those for whom it was a positional question, with turnout significantly higher in the latter category); (3) the combination of political disaffection and socio-economic dislocation – with strong support for Brexit among those feeling displaced by globalisation, by labour-market precarity and the perception that such precarity was a product of high levels of in-migration (significantly from East European accession states after 2004) and (4) the failure of the political establishment to convince the electorate of the (singularly malign, in its terms) economic consequences of Brexit – and the wider failure of a valence politics of what might be termed ‘expert paternalism’. Each factor is worth exploring in a little more detail.

Cameron’s Political Brinkmanship

The first is Cameron’s negotiating strategy with the EU and the parallels with that of Harold Wilson in 1975 (on which, see Butler and Kitzinger, 1976; Daddow, 2016; King, 1977). It is difficult to think of a more disastrous piece of political brinkmanship by any British Prime Minister (certainly since the Suez Crisis of 1956) than Cameron’s decision to put to the British electorate the question of continued EU membership and the politics leading up to that decision (see also Tooze, 2018: 545–548).

To be fair to him, the commitment to such a referendum in the Conservative manifesto for the 2015 General Election was made at a time when it seemed exceptionally unlikely that it would need to be honoured (though, for a different view, see Thompson, 2017: 444). ¹² While it was not unthinkable that the Conservatives might emerge as the largest single party after the election, few if any commentators and none of the Party’s advisors anticipated the Conservatives governing alone after 2015. Moreover, of potential coalition partners it was only UKIP who would be likely to accept a referendum as part of any coalition governing agenda (and they were likely to make the chance to vote for Brexit, a condition of any coalition agreement anyway). As such, the referendum pledge might well have seemed at the time a costless gesture.

It is just about possible, then, to see the Brexit referendum to which Cameron felt he had committed himself in the manifesto, as an unintended (and even slightly embarrassing) consequence of his own success at the polls – and, in particular, his success is obliterating his former coalition partners, the Liberal Democrats, in an election he had never expected to win.

But the more important point is that a Brexit referendum would have been unthinkable, even to Cameron, in the absence of the lingering internecine warfare in his own party over the question of Europe. In the 1970s, it was Labour that was divided over Europe (leading to the 1975 referendum); since the mid 1980s and especially since Thatcher’s Bruges Speech (1988) and Black Wednesday (1992), it was the Tories. The point is that the only conceivable question to which a Brexit referendum was the answer was one that started and finished with Conservative Party unity. The positing and eventual calling of a referendum on Britain’s EU membership was only ever a strategy for managing internal party discipline.

In a way, Cameron gambled what he regarded to be Britain’s economic future on an always rather tenuous strategy for re-uniting his party – and one that failed so spectacularly that it ended up costing him his own job and which still threatens to split the party today. Yet in so doing, he was arguably not entirely unlike Harold Wilson in 1975 (see also Saunders, 2016). But what Wilson got right, Cameron got catastrophically wrong – in large part because of the rather different political context in which their respective forms of brinkmanship were played out.

Wilson, too, was something of a reluctant European. He had argued passionately against Britain’s membership of the EEC in 1972. Yet, by 1975, he was (in today’s parlance) a pragmatic ‘remainer’, unimpressed by the, for him, nauseatingly integrationist tenor of the European ‘project’, but convinced in a rather more prosaic and practical way of the economic advantages of continued membership. Thus, when it came to preparing the ground for his own ‘in/out’ referendum on continued EEC membership, he picked, as negotiating points, substantive issues on which he was likely to gain significant concessions (unlike Cameron), he secured essentially all the concessions that he sought during the renegotiation (unlike Cameron), he refused to campaign actively for the remain that he tacitly embraced while making it very clear that his position was one of simple pragmatism (unlike Cameron) and he was positively surprised by the result, a significant majority in his favour in each region of the UK (most unlike Cameron).

It could be argued that Wilson was lucky – he, too, gambled what he also regarded to be Britain’s economic fate on political contingency. But he did so in a context in which pragmatic economic paternalism (the expert adjudication of what was presented as a narrowly economic issue) was significantly more likely to pay off. What is striking today is quite how difficult it is for political elites to make the case to citizens for something being in their shared or collective interest (Hay, 2007). In British politics, at least, it seems that the electorate is no longer capable of believing that something is in their interests if it is not directly opposed to the interests of someone else. The campaigners for Brexit seemed to understand that intuitively, and certainly, far better than their Remain counterparts with Michael Gove in particular seeking at times almost to turn the vote for leave into a rejection of the political appeal to dispassionate expertise – to considerable effect.

Turnout and Turnout Differentials

The second factor to which I would draw attention is that of turnout differentials – which, strangely, are hardly debated at all in the existing literature on Brexit (though for a rare exception, see Clarke et al., 2017: 208–211). It is almost certainly not the case that a majority of voting age British citizens supported Brexit on 23 June 2016. For, as ever, there were significant differences between the preferences of eligible potential voters, on the one hand, and the preferences of those of the electorate who were both registered to vote and who chose to vote on the other. This is not, in anyway, to question the democratic legitimacy of the outcome – turnout was, of course, high and the furnishing of a democratic mandate has never relied upon 100% participation. So Brexit was the democratically legitimate outcome of the referendum. As such, it is difficult to argue that it should not now be honoured (whatever one might think of the potential economic consequences).

But it is nonetheless instructive to reflect on the fact that the vote for Brexit was the product of significant turnout differentials between potential ‘remainers’ and potential ‘Brexiteers’. What is now clear is that, had there been a 100% turnout, Britain would have voted for remain and would probably have voted clearly for remain. Perhaps more significantly, had remain-minded and leave-minded voters turned out at a common and rather more credible 70%, the result would almost certainly have been different. The point is that predictable differences in turnout made Brexit always more likely than the polls suggested (for precisely such a ‘prediction’, see Hay, 2016).

There are, in fact, a number of differential turnout effects at work in the Brexit result. Among these, three are familiar, well established and relatively widely discussed. They are: (1) age cohort effects – the propensity for the elderly to turnout at much higher rates than the young and disproportionately to vote for leave (Franklin, 2004; Hay, 2007: 49–54); (2) national variations in turnout levels, with participation rates in remain-supporting Scotland (67%) and Northern Ireland (63%) being significantly lower than those in leave-supporting England (73%) and Wales (72%) (and with turnout typically lowest in areas where there was either little or no campaign for ‘leave’, as in much of Scotland, or where electorates were accustomed to having little or no impact on the result of UK-wide elections); and (3) a London (and more general metropolitan) effect – with turnout both low in London and lowest in those areas of London in which support for remain was strongest (typically areas characterised by high levels of both migration and poverty). The only significant factor pushing in the other direction was education – which was strongly correlated with both turnout and with support for remain (see, for instance, Alabrese et al., 2018: Table A.2).

These factors are all well known and well documented. But arguably there is a fifth, that although largely absent from the debate is arguably at least as important. It is the difference in engagement, mobilisation and, ultimately, turnout between those for whom the question of Brexit was a valence issue (a question of determining the shared collective good of the community) and those for whom it was a positional question (a simple question of political preference). From the perspective of the remain campaign, the question of continued membership of the EU was a valence issue. Brexit was simply bad for all. The question was, in effect, a technical and largely economic matter to be determined through the use of appropriate expertise. In stark contrast, for both the official and unofficial leave campaigns, Brexit was a positional issue – a question of politics not of economics, a question of values, personal conviction and identity which simply couldn’t be reduced to a set of technical considerations that might be adjudicated through the use of such expertise (or even sub-contracted to the experts).

In drawing attention to this distinction, my aim is not to pass a judgement on the political merits or otherwise of the remain campaign’s valence construal of the question of Britain’s EU membership. It is not difficult to understand why a more emotional and more political appeal was rejected in favour of a form of ‘expert paternalism’ as I shall call it. Instead, my aim is to point to one of the potential consequences of that choice – namely, a lower capacity to mobilise voters, especially in a campaign in which one’s opponents were engaged in a very different style of (positional) politics. The valence politics of remain was always likely to leave more of its natural support at home than the positional politics of Brexit, particularly with many voters discounting the possibility that leave could win. In terms of turnout, positional politics trumps valence politics every time.

Bringing these factors together may well make Brexit seem almost inevitable, at least with the benefit of hindsight (and not just in the Hobsbawmian sense that since things turned out the way they did they couldn’t possibly have turned out any other way). But that is the wrong inference to draw here. The success of the Brexit campaign in ‘getting out its vote’ should not be underestimated – and there are lessons to learn from this. While we should not exaggerate the extent to which the vote for Brexit was a product of political disaffection and economic alienation (the majority of those who voted for Brexit were, after all, typical and habitual Conservative voters), we should not underestimate the capacity of the Brexit campaign (official, unofficial and in combination) to mobilise those typically characterised in the literature on political disengagement and disaffection as apathetic non-voters. The Brexit vote gave many of them a voice. Arguably, this is its most remarkable feature.

This leads to a final observation about turnout and expertise, albeit expertise of a rather different kind – the expertise of the pollster. It is credible, I think, to suggest that the pollsters played quite a significant, if inadvertent, role in the making of the vote for Brexit. Throughout the campaign, it is now clear, they systematically underestimated the likely size of the leave vote. In large part, they did so because they underestimated the capacity of the leave campaign to mobilise the serially disaffected and those serially accustomed not to vote. There is nothing terribly surprising about that failure, especially given the methodological difficulties inherent in seeking to construct a vote projection from polling data for a one-off referendum of this kind (in which all sorts of potentially problematic turnout estimations need to be made).

But three points of some significance nonetheless follow from this. First, getting it wrong in this way may well have contributed to keeping portions of the remain vote at home in that it encouraged them to think that they did not need to vote in order for Britain to remain an EU member state. Second, and conversely, it may well have contributed to boosting the vote for leave, by making it easier to think of this as a relatively costless and largely symbolic protest against the political establishment (both domestic and European).

The third implication is of a rather different kind. When the pollsters came to re-assess their methodologies in the light of their (perceived) failure to have anticipated the vote for Brexit (and, indeed, Cameron’s majority in 2015 and May’s vanishing majority in 2017), their conclusion was that they had actually done rather well (see, for instance, Sturgis et al., 2016). Electoral outcomes aren’t easy to predict, especially in times characterised by high electoral volatility, referendums pose even greater challenges and, crucially, their predictions for the 2015 and 2017 General Elections and the EU referendum, though wrong in the rather obvious sense that they failed to anticipate the result, all fell within the declared margin of error. The implication of the preceding discussion is that such a conclusion is simply not good enough.

The question is not whether the pollsters did the best job they could have done methodologically. It is credible to think that they did. Their failure was not methodological but presentational. If polling is inherently tricky, polling in referendums doubly so, then the caveats need to come first and in a rather larger font size. The responsibilities of the political analyst (pollsters included) demand a sharing of the methodological limits of the analysis with those for whom it is intended from the outset. They also demand a much greater sense of responsibility for how the expertise provided is and might be used and misused. It is simply not sufficient to say that the science was good and that the result fell within the margin of error if most of those who consulted the poll were not even aware that there was a margin of error.

That brings us to the third and fourth set of factors, which it makes sense to discuss together.

Brexit and the Politics of Expertise

Brexit represents the failure of a certain style of politics and the triumph of another. That triumph might well prove temporary, but it is certainly credible to think that it might endure. Above all, Brexit represents a rejection of the politics of expertise, at least in its current form, and of expert paternalism in particular. Arguably, the zenith of this expert paternalism was the ‘this is good for you because we tell you that it is’ of the remain campaign in both the Scottish Independence and EU referendums.

In stylised terms, the rise and demise of expert paternalism might be seen to have passed through three stages.

In the 1960s and 1970s, political elites were themselves trusted to deploy appropriate expertise, and the growth in the expertise available to them led to a confidence in their ability to govern (paternalistically) in the collective good.

By the 1980s and 1990s, particularly in the Anglo-liberal democracies, political elites were typically no longer trusted to deploy expertise themselves. Such expertise needed to be depoliticised and placed at one remove from them (Burnham, 2001). Only experts, in this conception, could genuinely discern the collective good – and could only do so if their expertise were protected and insulated from political influence. The result was a politics of ‘expertocracy’ and depoliticisation, in which positional issues were recast as valence issues and placed in the hands of (nominally) independent experts – central bankers and the like.

Brexit rejects that. Its mantra – ‘taking back control’ – is not just about taking back control from Brussels, but taking back control of politics from experts too (and perhaps also taking back control of politics from those political elites who would seek to cede it to experts). In a sense, there is little point and hence little to be gained by taking back control from Brussels if that control is simply to be passed domestically from the elected representatives of the people to their favoured purveyors of expertise. The implied re-politicisation of the domestic polity could take, and might still take, a variety of forms. But, for now, it is predictably the right, and the populist right at that, that has shown itself most at home with, and most deft in deploying, this new vernacular. It is populist; it is nationalist; it is dis-integrationist; it is anti-internationalist; it is, as Albert Weale (2016) puts it, nostalgically democratic – and it has the capacity to recast British domestic politics and, in the process, its political economic interdependence.

It also has potentially profound implications for political scientists as, at least potential, purveyors of expertise, in a context in which such expertise (in the form of ‘impact’ or ‘publicly relevant research’) is an ever more highly incentivised marker of professional standing and achievement. Here is not perhaps the place to pursue these issues in the detail that their significance warrants. Suffice it to note that if there is to be an, arguably, long overdue, recalibration of the relationship between the suppliers of expertise and the exercise of democratic governance, then we will need to think long and hard about the responsibilities that come with the provision of such expertise. Four points immediately present themselves. First, even the genuine, neutral and dispassionate provision of a non-normative scientific knowledge (such as the pollster’s psephology) can have a significant partisan effect, as in the potential contribution to the making of the vote for Brexit played by polling data in the preceding discussion. Second, and rather more generally, the use by political elites of expertise as a mechanism of governance has served to mask, obfuscate, depoliticise and, arguably, de-democratise decision-making processes, not least by turning positional issues and questions (‘who benefits and who suffers from an interest rate rise?’) into valence issues and questions (‘what is the optimal interest rate setting for an economy such as ours at a moment like this?’). Third, any attempt to define more clearly the responsibilities of the expert needs surely to insist on a much more sharply drawn distinction between the expertise itself and the use made of that expertise by decision-makers if the democratic accountability of the decision-making process is to be maintained. And, fourth, a much greater emphasis needs to be placed on communicating the limits to the expertise provided (the margin of error and the methodological reasons for it are just as important as the most probable predicted outcome).