Predictions for International Security: The Knowledge Practice Enigma

Copy/paste this URL to link to this paper
This conference brings scholars from different disciplines - political science and international relations, sociology, history, economics, demography and philosophy - into a discussion on the nature of predictions in international security. It will address the following questions. How are claims about the future made in international security, a professional realm that is obsessed with knowing the future? How are these claims “sold” on the public marketplace of ideas? Is anticipating the future about anticipating change?

We will meet on Friday March 16th and Saturday March 17th 2012.

The conference will take place at CERI-Sciences Po (on the 16th, 56, rue Jacob) and at ENS (on the 17th, 46, rue d'Ulm).

The conference is organized sponsored by CERI Sciences Po in cooperation with Ecole Normale Supérieure – Institut Jean Nicod. It benefits from the financial support of Institut Jean Nicod, the French Ministry of Defense, the Alliance Program (Columbia-Sciences Po), the CERI (Sciences Po).

Papers open for discussion

Introduction: Can We Predict the Future?

Date of publication: 10 January 2012
Social scientists argue over the possibility to make valid claims about the future of politics. Some are very skeptical, but others are much more confident. Indeed, can we predict the future of world politics? Is science the solution? Or, what kind of sciences would bring the better answers? Do predictive claims affect the course of events and how? Most of the videos were shot in New York in 2010. />

On the Ontological and Ethico-Political Presuppositions of Assigning Probabilities to Possible Futures

Date of publication: 31 January 2012
Social sciences have found very few if any stable, law-like regularities. In their absence, also prediction seems impossible. There are conceptions of causation that explain this absence of universal empirical regularities, especially the Critical Realist conception of cause as an INUS-condition in open systems (causeis an Insufficient but Non-redundant element of a complex which is itself Unnecessary but Sufficient for the production of a result). However, Critical Realism (CR) seems to lack an ethico-political theory about our choices in a world of limited knowledge about the future, something that was developed by Keynes in his A Treatise on Probability. The juxtaposition of Keynes and Critical Realism can reveal many important ontological and ethico-political presuppositions about attempts to predict the likely consequences of particular courses of action, policies or reforms. Keynes discussed the issue whether probabilities are stable. From an ontological point of view, what he seems to have argued is that all systems are open and closed to a degree, although he did not use this terminology. Because the world is not unchanging, any probability of X may change too. On the other hand, a possible interpretation of the reason why CR is so categorical about the asymmetry between explanation and prediction is that it is embedded in a Marxist framework of emancipation being a one-time revolutionary moment in world history, or a series of them. However, any concrete policy or reform or emancipatory change requires some knowledge about the future, which, in the absence of certainty, presupposes a rational basis for assigning probabilities and revising them in the light of new evidence or events/developments. Keynes's "Middle Way" politics was based, in part, on the conviction that we have very little basis for knowing the long-term future. Hence, a relatively secure improvement here and now is better than an abstract but highly uncertain hope about a radically better future. This is a rather "flat" argument, as changes can also be (devised to be) cumulative and as "militant hope" (Bloch) can inform policies and reforms. What is clear, however, is that discussions about the ontological conditions of the possibility of prediction are interwoven in interesting ways with the normative basis of transformative politics. Moreover, there is room for a synthesis overcoming the limitations of both CR and Keynes.

When predictors say what cannot be done Revisiting the authority of nuclear proliferation forecasters

Date of publication: 01 February 2012
This paper intends to reveal problems in the literature about the political impact of expertise; this literature exists in sociology, mainstream International Relations and critical security studies. Over the last two decades, the study of the political impact of “epistemic communities” in mainstream IR has narrowly focused on their ability to promote policy options; it has focused too exclusively on the elites to measure the impact of these experts. The sociology of intellectuals and experts originally tackled the problems of experts’ allegiance to power but has moved away from them towards a study of experts’ interventions. Critical security studies, which could have tackled the problem of the authority of experts, have moved away from nuclear issues.

This paper intends to shed light on these problems in the literature by introducing four criteria by which to evaluate expert authority and by asking the following question: what are the effects of experts’ authority in the nuclear field? It focuses on the practice of forecasting the next nuclear-weapon state. This is a useful focus for two reasons. One being that the authority of unelected experts is a problem in every democratic society (a problem that the quasi-instantaneity and irreversibility of the destructiveness of nuclear weapons exacerbate). The other being that this practice is as old as the nuclear age.

The paper answers the question of the effects of experts’ authority in the nuclear field in three ways.

First, this paper questions the side of proliferation predictions that consists of defining what is considered to be politically impossible, or impractical. In other words, it focuses on what could not and therefore should not be tried, thereby breaking with the mainstream IR approach of epistemic communities sticking to the advocacy dimension.

Second, instead of focusing on the influence of experts on policymakers alone, this paper adopts a broader definition of political influence that includes public opinion.

Third, by offering a better understanding of the effects of experts’ authority, this paper attempts to reintroduce normative considerations into the discussion and to draw normative conclusions on the ability of experts to shape, or even hijack, the debate about national security issues at two levels. First, by showing how determining the impossible in the future deprives the political authorities of their responsibility to decide what can be changed and what new can be introduced in the world. (In this sense, expert judgments about what is presented as “impossible”have a very strong political and under-researched impact.) And second, by showing that these expert judgments are based on political assumptions and preferences even when they are denied.

Forecasting in the Security Community: An American Perspective

Date of publication: 05 February 2012
 Intelligence aims to reduce uncertainty for leaders and decision-makers.   There are many perils in the business of prediction, however.  The most devastating events for a nation involve strategic surprise, be it a military attack, technological breakthrough, or geopolitical maneuver.  With each such development in recent U.S. history, the expectations and demands of intelligence have escalated.  Critiques and calls for re-organization, enhancement of intelligence capabilities, refinement of analytic methods, and bridging of barriers between intelligence agencies and their bureaucratic cultures have grown as well. 

With each such purported U.S. “intelligence failure,” legislative hearings are convoked, study commissions formed, “lessons learned” exercises ensue, reforms are mandated, and U.S. intelligence is often re-organized, most recently under the 2004 Intelligence Reform and Terrorism Prevention Act.  For their part, intelligence consumers ask what good is intelligence if it cannot give them useful forecasts with which to forestall catastrophes that would imperil U.S. security and interests.   All efforts to improve performance, however well-intentioned, do not engender realistic expectations of intelligence’s accuracy and timeliness. 

The expanding complexity and inter-relationship of developments brought on by cascading globalization and exploding worldwide inter-connectedness present new and more daunting challenges for intelligence analysis and forecasting.  Pinpointing geopolitical variables and drivers, envisioning alternative hypotheses, issuing indications and warnings, estimating strategic trends, and anticipating tipping points all demand improvement.  Minimizing analyst bias, explicitly stating assumptions, assigning probabilities, and sifting the strategic wheat from the current intelligence chaff also remain obstacles to cogent, relevant, analysis.  Sadly, strategic forecasting remains an under-valued, under-developed endeavor in the U.S.  All such reform and improvement efforts fail to fulfill the presumed -- but superhuman -- mission of intelligence, i.e., to foresee threats, identify opportunities, and credibly forecast the future. 

The Past Future: How and Why Past Population Projections Failed

Date of publication: 05 February 2012

Prediction Markets Predict Well, But Few Care

Date of publication: 05 February 2012
 We have seen enough of prediction markets in action to tentatively conclude that they have low operating costs and make forecasts that are often much more accurate, and rarely much less so, than commonly used sources. However, even when organizations become convinced these advantages are real, such as via a direct internal demonstration, most are not interested in using prediction markets. Similar fates seem to befall related mechanisms for better forecast accountability. I consider several possible explanations for this puzzling disinterest. 

Indicators and Regulation: Implications for Predictions

Date of publication: 05 February 2012
An indicator is a collection of named, rank-ordered, simplified and processed data that purports to represent the past or projected performance of different units (Davis and Kingsbury 2011; Davis, Kingsbury and Merry 2012)  An indicator simplifies and processes data about a named social phenomenon in a way that makes it possible to compare and evaluate units such as countries, communities, organizations, or individuals.   Indicators can be used for comparison and evaluation in ways that other forms of information cannot.  Indicators set an implicit standard of achievement.  They may define what constitutes a named social phenomenon, by deciding what is measured, and they may implicitly incorporate views as to causation and pathways to improvement.  Often these implicit claims are reinforced by the simplifying name given to the indicator.  For example, indicators using labels such as “state failure” or “state fragility” transform particular statistics into the description of a specific social problem.  This paper presents some ideas from work on indicators in global governance that may have significance for the study of social science-based predictions.  Consonant with the theme of the conference, some examples will be given of security indicators, particular indices of state fragility and political instability.

What are we asking about when we ask about the future? A socio-epistemological perspective on probability questions

Date of publication: 05 February 2012
Questions about the future are very often probability questions. Probability is a fascinating field of philosophical enquiry. On the one hand the philosophy of probability has concentrated on the double face of the concept. It is statistical, concerning itself with stochastic laws of chance processes. It is also epistemological, dedicated to assessing reasonable degrees of belief in propositions quite devoid of statistical background. On the other hand, it has concentrated on the relation between probability estimations and decision theory, that is, the utility functions associated with a certain outcome when a subject has to decide how to act upon an estimate of the future. We argue here that a deeper and, surprisingly neglected, philosophical issue about the double face of probability is the following: the probability of a fact and its impact (objective or subjective) are two distinct phenomena, sometimes confused when talking about probability. I can have a precise measure of the impact of a fact (if a tsunami takes place in Jakarta, there will be 5000 deaths) and yet have a much less precise measure of its probability. Impact questions and probability questions are often confused. We will provide example of this distinction and also, if, possible, some remedies.

Forecasting Security and ‘Inefficient Causation’

Date of publication: 05 February 2012
 In this, the second chapter of a book on causation in international relations, I begin by asking how appropriate the general approaches to causation I examined in the previous chapter are to our field.  I evaluate these approaches in terms of two criteria: their comprehensiveness and fit between assumptions and the political world.  Regularity theories and Scientific Realism come off the worst, while process theories and singular causation fare better.  I go on to propose a new approach I call inefficient causation.  It aims to build a bridge between general understandings of international relations and specific features of context.  It assumes that most important international outcomes are context dependent and therefore best characterized as examples of singular causation.  It is constructivist in that it stresses the causal implications of cognitive frames of reference, which include identities, analogies, metaphors as and more general intellectual and visual frameworks.  It also emphasizes the role of mechanisms and processes that mediate between frames of reference and behavior.  It makes the case for construction of complex explanatory narratives and forecasts.  It rejects theory- or regularity-based predictions as impractical.

Focal Points in International Security

Date of publication: 05 February 2012
 What kind of future worlds do experts of international security envision? This paper studies the role of experts in D.C.’s think tanks, a relatively small world socially and culturally highly homogeneous. It underlines the characteristics of this epistemic community that influence the way its analysts make claims about the future for security.

The paper highlights three main features of the relation between those who make claims about the future of security and those to whom these claims are addressed (mainly policy makers). First, it shows that, for epistemic but also for political reasons, the future as it is being imagined in think tanks is relatively stable in linear. Second, it shows that think tanks are also “victims of groupthink”, especially when they make claims about the future. Third, it underlines a paradox: scenarios and predictions create surprises. Claims about the future have a strong tunneling effect. They reinforce preexisting beliefs, create focal points and operate as blinders when, inevitably, the future breaks away from its linear path.

Anticipating the Future of China: Possible Pathways to Modernity

Date of publication: 14 February 2012
            Predicting China’s future is a seemingly impossible task that people choose to take on anyway.  People want to know how wealthy and powerful China will be, how fast it will happen, how China will use that power in international affairs, whether China will become a stable democracy or a nationalistic loose cannon, and whether its citizens will enjoy human rights.  People want to know because they feel their lives—and their actions—depend on it.  Having a vision of possible and likely futures of China matters because outsiders think their choices might affect China’s trajectory and because they want to be prepared to respond to China’s outsize presence.

            Prediction is hard in the social sciences under the best of circumstances, but thinking coherently about the possible path of this kind of complex, contingent, opaque, long-term, strategically interactive, self-reflective process is a crushingly daunting task.  And yet it self-evidently needs to be attempted.  How do people do it, and how should they?

            Mainstream social science starts prediction with a survey of the current facts that our theories tell us are most relevant to the kinds of outcomes we seek to predict (our “initial conditions”), proceeds to survey relevant dynamic causal theories that have passed at least a minimum empirical test, and considers “if, then” hypotheses about how different contingencies might affect the expected trajectory.  Even for small, near-term, well structured problems, social scientists realize they are not very good at this, but we nonetheless often set up our research programs as if they might produce this sort of knowledge.  Even more dismaying, editorial writers and policy advocates likewise talk as if they have this sort of predictive social science knowledge, which they don’t.

            One alternative to impossible prediction is to switch modes to exhortation:  to tell China what it “must” become, and to guide one’s actions by values that would be good to aspire to.  Related to this is the strategy of imposing one’s plan on the world, forcing China to adapt to us and to try to make China absorb the uncertainty of the future.  

Another alternative is trial and error: to take tentative steps, see what works, reinforce success, update our hunches and habits, anticipate counteraction, and kick the can down the road.  As Napoleon said, “on s’engage, et puis on voit.” 

Still, these strategies for operating effectively in the face of uncertainty still require some predictive impetus to set them in motion.  This impetus is different for those whom Isaiah Berlin and Philip Tetlock call hedgehogs, who know one big thing, and those whom they call foxes, who know many little things.  In the study of international relations, liberal and realist hedgehogs simplify the task of prediction by assuming that in the long run all trends will conform to a few basic variations on a singular pattern, which can be used to guide conjecture about possible futures and to assess the likely consequences of one’s own actions.  In contrast, foxes, such as area studies experts, may take many independent strands into account, focusing on the complexities and contingencies of how these strands may weave together or pull apart in the short run.  Tetlock shows that foxes are better predictors of specific near term developments, but they set themselves an easier task.

Predictive discourses that trespass on the terrain of extreme uncertainty run the risk of veering into ideology, blinkered by a narrow social interest or perspective, or utopia, blinded by a feel-good solipsism.  Nonetheless, I want to argue that social science theorizing and empirical research, despite its obvious predictive limitations in this kind of subject matter, can and should be a valuable and constructive tool for hedgehogs, foxes, bold system-imposers, timid experimenters, ideologues, and utopians who are trying to come to grips with China’s future.  Let’s see how this can work.


The future perfect of (in)security (P8) : Pre-crime strategy, Proactivity, Preemption, Prevention, Precaution, Profiling, Prediction, & Privacy

Date of publication: 29 February 2012
In this paper I discuss the move to prevention/prediction in relation to security practices.

In a first part I want to insist on the convergence of police practices concerning proactivity, pre-crime strategies, of military discourses concerning pre emption and prevention, of climate change discourses on precaution, on banking system on profiling, and on the general trend towards a prediction oriented towards a worst case scenario which limits the options of the future and the trend to analyse the future as a perfect future, a future already known which justifies the practices of detention and a logic of suspicion associated which what is called now a “preventive justice”. This way to monitor the future, to colonise it in the name of specific knowledge and know-how coming from computer science, criminology, psychology, and military/police security industries is presented as an optimum management of risk by the managers and the transnational guild of professionals of (in)security. Some researchers consider that it leads us to a new form of governmentality rooted in the knowledge of catastrophic events reframing the arrow of time in terms of responsibility and decision, and suggesting that the anticipation and prevention of these events has to trump individual freedom, to reframe the relation between liberty and security in favour of security and prevention. The diagnosis may be accurate in terms of the programme of the discursive practices, but it seems to me that we have to discuss the pretence of these forms of knowledge concerning the future to be scientific when they want to apply to specific individuals and not large trends. The (in)security approach of the future is embedded into a myth of computerisation exchange of information, and profiling, as predicting the true behaviour of very specific individuals while the practices show the limits of the categorisations of future human behaviour, and the tendency to transform prediction into a justification of previous arbitrary actions. Their predictions concerning human beings are forms of sacrificial astrology by other means applied to small groups, and which destabilise and reconfigure presumption of innocence, freedom, privacy, justice, and even anthropological conception of human beings for all. In conclusion I will briefly come back to the link between this trend towards prediction and the (in)securitization process which is the result of a de-differentiation and reshaping of the boundaries of the different security institutions as well as one of the symptom of a governmentality of unease.

From Europe’s Past to the Middle East’s Future: The Constitutive Purpose of Forward Analogies in International Security*

Date of publication: 01 March 2012
Predictions are commonly used in the fields of international security, environmental policy, and financial affairs. In international affairs, different techniques exist to model and test the credibility of these predictions: econometric forecasting, simulation by humans and non-human actors, and “forward analogies.” This paper analyzes how policymakers and experts draw on forward analogies to produce predictions about the future in the field of arms control and nuclear disarmament. More precisely, it will show how these forward analogies between past and present situations in one context (Europe) and present and future situations in another context (the Middle East) were used to advance diplomatic negotiations in a particularly hard context: the deliberations of the Weapons of Mass Destruction (WMD) Free Zone in the Middle East. In situations in which diplomacy is blocked by the unwillingness of parties to start negotiations, this paper claims that such use of forward analogies can serve not only a predictive purpose (to the extent that these analogies produce predictions), but also a constitutive purpose: analogies help “constitute” the reality of regional orders (such as the “Middle East”) when their ontological status as objects of deliberation and intervention is problematic.