Illuminating the Shadow of the Future:
Scientific Prediction and the Human Condition
September 23 – 25, 2005
Conference Home | Prospectus | Agenda | Participants

Conference Prospectus
prepared by Frank Wayman, Aug. 11, 2005:

We are proposing an inter-disciplinary conference, hopefully to be disseminated as an edited volume on prediction of the human condition, with focus on predicting several inter-related variables:

• Global Shifts In Material Standard of Living, including economic, social, health, and environmental well-being, and Sustainable Development
• Spread of Free Society/Democracy
• Decline, Spread, or Altered Distribution of War
• Evolving Patterns of Genocide, Ethnic Cleansing, and other Mass Political Killings
• Proliferation of Weapons of Mass Destruction

Our interest is in global modeling, by which we mean efforts to forecast and explain patterns that cover our entire planet and have a time-span of up to a few decades, (or, more ambitiously, the two centuries of the industrial age). While it is useful to predict future patterns from past patterns of the same variable (for instance, predicting world population in 2025 or 2050 from world population now and in the recent past), we intend to explore the hypothesis that the above variables are indeed inter-related, so we are especially interested in papers that test those inter-relationships. For instance, we would be especially interested in a paper that examined the relationship of war not only to past war, but also to shifts in material standard of living, concomitant population surges, and spread of democracy.

The leadership of the Research Club of the University of Michigan (specifically, Prof. Frank Wayman, president, and J. David Singer, immediate past president) had devised the idea for such a conference, and discovered that Prof. Bruce Bueno de Mesquita (Senior Fellow at the Hoover Institution at Stanford and Chair of the Department of Politics at NYU) was planning a very similar conference together with Dr. Paul Williamson of Global Vision (a non-profit focused on global forecasting). It made sense to pool the funds for the two conferences and run the merged conference at the University of Michigan in Ann Arbor (where Dr. Williamson resides, and where Dr. Bueno de Mesquita has earned his doctorate). Prof. Bueno de Mesquita graciously offered to match the Univ. of Michigan funds even though the conference would not be held at his institution. The merged effort of these initiatives has allowed us to generate an exceptionally diverse conference of scientists, assembled for the purposes of scientific prediction of global conditions.

While modern life has led to academic specialization, we wish to hold a conference that is as inter-disciplinary as possible, based on the idea of consilience (Wilson 1998), within the context of what we are studying. Hence, we are interested in what cross-fertilization of ideas we can engender from a variety of perspectives, including not only varieties of political science and economics, but also mathematics, demography, public health, evolutionary biology, and, where possible, even the physical sciences and certain humanities. Beyond that we wish to inquire whether at a higher level these apparently diverse perspectives can be merged into an underlying or some underlying modes of inquiry. While scholars in those fields are not always interested in devoting substantial chunks of time to study in the topics we are exploring, several who do (for example, Lewis Richardson, Jay Forester, Isaac Asimov) have made a good start in the direction of forecasting. We are interested in any predictive method that works, and that can be applied in a scientific test against evidence from after the date the prediction was written.

Modern growth and concomitant instability

From Everest it’s just a step: off the edge, you hit Tibet. Applying this notion that the higher you go the harder the fall, one might say that modernization, while creating an unprecedented prosperity among the economically developed nations, has left those nations in an artificial and perhaps easily destabilized level of well-being. Ghengis Khan and his hordes were hard-pressed to devastate a peasant society, but nineteen terrorists may have substantially damaged the world economy on Sept. 11, 2001. Advances in econometrics have made forecasts of next year’s national product increasingly accurate, but forecasts of war and revolution remain less validated. Even the otherwise predictable economic forecasts are thrown awry by political shocks such as the 1973 OPEC oil embargo, the 1979 revolution in Iran, and the 1990 Iraqi invasion of Kuwait (Goldstein, Huang, and Akan 1997). This points to the need for the higher level of integration we seek, such as incorporating political shocks into prediction of economic outcomes.

Predictive Methods

Predicting the future seemed to reach clockwork precision with Newton, whose principles allowed one to predict where Mars would be on each day centuries into the future. Modern social science starting in the twentieth century had some success predicting the future, but much failure in emulating Newton. No social science theory, for instance, told the date, month, year, or decade in which the Soviet Union would fall. In recent decades, chaos theory from mathematics has helped us understand some of the reasons: especially if measurement of initial conditions is very poor (or perhaps even slightly imperfect), it is very hard to predict very far into the future of even deterministic complex systems. And the probabilistic nature of social science theories makes failure more likely. Moreover, Darwinian evolutionary theory (purely verbal rather than formal at first) provided a way to conceive the development of complex systems, biological and then social. Axelrod's Evolution of Cooperation (1984) is an example of how this sort of evolution can be modeled in terms pertinent to international conflict. However, the results, while predictable (tit-for-tat tends to win) are probabilistic at best, and somewhat dependent on what sorts of initial strategies are present. It is also somewhat unclear how the utilities in the game can be measured in real biological systems (that is, what the linkage is between utility and variables that affect survival and birth and maturation of offspring). It addition, perhaps because it is written to be a very general template, this evolutionary model does not suggest the actual timing, or even the approximate time scale, of the process it is modeling. Prediction of the future, then, is less certain when one moves away from the simple Newtonian problem of one body rotating around another in a bipolar solar system.

Yet so much in the global human near-term future merits forecasting. Who would not want to know whether and under what circumstances risks will emerge – especially if such information were timely enough to allow counter-measures to avert catastrophe and funding to empower faster progress? Since the industrial revolution, change has not only become more rapid, but has been tightly linked to the process of economic modernization and its non-economic consequences. These include improved material standard of living, environmental stress (over-population, pollution, resource depletion, loss of biodiversity through extinction and invasive species), spread of liberal democracy, emergence and decay of communist and fascist regimes, decline of military rule, discovery and spread of weapons of mass destruction, genocide, and alterations in the patterns of warfare. We propose a conference on these inter-related global phenomena, including attention to:

1. the degree to which the variation in these conditions during the modern era has been systematic enough to allow progress in scientific explanation/prediction of past patterns

2. the degree to which such prediction of the past allows extrapolation to the future

3. the degree to which forecasting has elements different from explanation/prediction

4. the absolute and comparative advantage of competing means of forecasting, including but not limited to
   a. conventional multivariate regression and analogous methods such as logit/probit;
   b. game theoretic and choice theoretic models where strategic interaction may be involved;
   c. prognostications by open-information experts, such as academic area specialists and journalists;
   d. expectations of those with access to classified information, such as government officials;
   e. refinements of the above through Bayesian updating of prior expectations.
   f. dynamic modeling (e.g., Richardson, Forester World Dynamics)

5. the degree to which methods from the various sciences, mathematical, biological, physical, philosophical, and social, lend themselves to the solution of these problems.

In making this proposal, we have tried to strike a balance between, on the one extreme, being so broad as to be of little use in this age of specialization, and, on the other extreme, of being so specialized as to be of only narrow interest. We believe we have tailored a proposal that taps into a broad-based network of scholars from many academic disciplines, but that allows us to bring them together in a way that will bring fresh insights.

We are planning a conference on this topic at the University of Michigan for the weekend of September 23-25, 2005, with an ensuing book. Funding for all this will be provided by the Office of the Vice President for Research funds to the Research Club of the University of Michigan combined with support from the Hoover Institution, Stanford University. Planning the conference is being done by a collaboration of the above two parties and Global Vision, Inc., a non-profit 501 (c)(3) entity having a focal interest in scientific global forecasting.

PREDICTION IN SCIENCE

Science has been able to advance by being based on three principles: (1) looking for connections between things; (2) studying dynamics rather than static situations; and (3) examining evidence to see if the supposed connections really do help one predict the changes occurring in the subject. These three principles are the core of the Global Vision approach. For example, Newton predicted where planets would move based on connections between their motion and things that influenced that motion, namely, their mass, the mass of the Sun, the gravitational attraction between massive objects, and the distance between a given planet and the Sun. Newton connected these variables in a dynamic model that differed from the older, static view of science often attributed to Aristotle. And the predictions were falsifiable, meaning that evidence could be used to confirm whether or not the planets moved to exactly where Newton predicted and did so at exactly the time he predicted.

Predictions can be about the present. By 'present' is meant a span of time during which the variables of interest may be regarded as unchanging. As often the case in political or other societal inquiry, the period of one calendar year may be so regarded. Other short or (more usually) longer time periods (e.g., decades) may also be regarded as constituting the present. It should be clear that this idea of "present" is judgmental and contextual. One also conventionally regards the present to include whatever moment we, the "observers," "now" occupy. Relative to any such definition of the present, the past consists of those moments in time that preceded the present; the future consists of moments of time that will follow the present. In sum, the terms past, present, and future refer to a 3-fold stratification based on time.

Using this terminology, we note that predictions are most limited when they only "predict" what is in the present based on other things also in the present. Let us refer to this as the first type of prediction. This happens when the prediction takes the form of estimating the present value of one variable based on its correlation to other variables and on their present values. This is a type of static prediction, in the sense that the variation is across "space" only, rather than across time and space; this has been called "correlational design" (Campbell and Stanley, Experimental and Quasi-Experimental Designs for Research, 1966; Cook and Campbell 1979). In contrast, by "forecasting" we mean predictions over periods of time during which the variables are regarded to show changes in value, such that prediction requires that we predict the changes. One form of such forecasting is "postdiction," when information taken from what was once a "past" stratum is used to predict what then was in the present. This form, which we may call the second type, is of greater interest than the first type, for the reason that this second type may be used to test dynamic models. Of course, the power of any such test is limited in that all the data are known a priori. This catch is what Niels Bohr had in mind when he said, "Prediction is difficult, especially about the future." Of course, as Bohr was implying, of still greater intrinsic interest is forecasting from one's own time to the future. Let us call this the third type.

This scientific method of prediction through dynamic modeling of the connection between observed variables can be embodied in one equation, which we can call the dynamic equation. It is that y at t+1 = f (y at t, x at t, and possibly x at t-1). Incidentally, this contrasts to a static model, in which y at t = f (x at t, and possibly some accumulated x from the more distant past to t-1). (In these definitions of dynamic and static prediction, while the symbol t-1 is meant to denote x at the immediately prior point in time, it could be generalized to x at some previous point in time, or a several previous points in time.)

Models predict what will happen in the future when they say what will happen at a point after the research has been completed and published. For example, Newton's theory predicts where Mars will be years after Newton worked. Even static models can be predictions of the future if the analyst has an understanding that there are logical reasons for the future to probably be like the past. To be relevant to policy, social science models, whether static or dynamic, must be generalizable enough to hold for the future as well as the past in which they were tested.

Prediction is usually thought of as the other side of explanation, but some predictions can be made (and be accurate) even when an explanation is not yet available. If there is an explanation, that gives us more confidence that the future will be like the past, because we expect the correlation between the predictor and outcome variable to be more stable if we see that there is an explanation for why it has its current value.

Extrapolating from a trend is saying what the future y will be like using only the prior y variable, i.e., without bringing in the x variables. Such extrapolation may not be prediction and explanation, because those two things may require some other variable to be the thing producing change in the outcome. If a non-scientific but savvy prognosticator, such as Jules Verne or H.G. Wells, successfully anticipates the future, it is probably best to call that intuitive forecasting rather than prediction, because the prognosticator has not explicitly spelled out the variables that explain the future forecast.

FOUR VARIATIONS

1. It is possible to do dynamic modeling/prediction just regarding past and present, with implications for future left implicit. In fact, there is no way to validate future predictions except by testing the model on historic data, so any model that has met empirical tests has been used to study the past.

2. If the x to y link is weak and x varies unpredictably, yet y varies in a relatively stable manner, it is acceptable to just predict future y based on past and present y. (Demography is a relatively stable y.)

3. It is possible to forecast based on intuition. Jensen (1972) has shown that State Dept. officials and journalists are better than academics at this. It is possible that, in addition to such journalists and diplomats, science fiction authors such as Verne and H.G. Wells and Asimov, and others who use intuition and logic to anticipate future events, may be better than more academic analysis when the goal is anticipating what the future holds. It has been said that H.G. Wells' epitaph is basically, "I told you so." A full consideration of alternative futures has to be open to the hypothesis that such literary intuitions may have merit.

4. Intuition may also work as input into a mathematical model for forecasting. While Jensen shows you can just forecast with area specialists' raw opinions, Bruce Bueno de Mesquita (1985) uses these intuitions as a basis of a rational choice type forecasting model.

GAME THEORY RELEVANCE

Axelrod (1984), Evolution of Cooperation, shows how conflict/cooperation predictions and evolutionary dynamics may be affected by strategic interaction. This is more Darwinian than Newtonian. How does this get incorporated in what sort of predictions work best? Axelrod brings math and computation to what with Darwin had been initially a verbal model. (Erik Gartzke has emphasized to me -- if I may express it in my own words -- that powerful decision makers in the global system have their own expectations of what will be effective (often novel) strategies, based on their own intuitions about the future, and that these can confound simple predictions based on past conditions.)

ORGANIZATION OF THE BOOK

To make a contribution to these larger questions, we must focus on a subset of areas in which progress has recently been made, further progress is possible, and we can bring together scholars who can truly interact with each other. We focus particularly on the relationship of economic change, comparative politics, and international relations [recently explored in Bueno de Mesquita, et al., Logic of Political Survival (2003) and in the democratic peace literature], and on danger of nuclear proliferation and nuclear war.

Conceptual outline of Proposed chapters/presentations

NOTE: NOT ALL NAMES ARE COMMITTED TO WRITING, AND ARE FOR EDITORS' PLANNING PURPOSES ONLY
1. Introduction: Consilience (Prof. Edward Wilson, keynote speaker)

2. Overview of themes of the book (Frank Wayman)

3. A basis on which to organize diverse contributions to global forecasting (Paul Williamson)

4. Impact of natural environment on violent conflict. (Urs Luterbacher)

5. Impact of climate change on politics (Detlef Sprinz)

6. Modeling sustainable development I (Doyne Farmer)

7. Modeling sustainable development II (John Holland)

8. Forecasting nuclear weapons proliferation – use of a hazard model (Atsushi Tago)

9. Severity of war as a function of time – past patterns and concomitant forecasts (Claudio Cioffi-Rivella)

10. "Cultural Collisions: Forecasting The Evolution of History Using 'Boundary Value Behaviors' " (Myron S. Karasik)

11. "Predictability in evolutionary biology, especially as applied to human sociality" (Dick Alexander)

12. "Bankers as Oracles? Forecasting Political Developments with the Help of Financial Markets" (Gerald Schneider)

13. On identifying the coalescence of individuals into groups, and the movement from rhetoric to action – application of computational linguistics (Verspoor)

14. On the evolution of international conflict (Wayman)

15. “Analysis of Power-Structure Fluctuations in The ‘Longue Durée’ of The South Asian World System” (Wilkinson, with Tsirel)

16. “System Change and Richardson Processes – Application of Social Field Theory” (Williamson)

17. "Critical Time Scales in Neural and Global Systems" (Mayer-Kress)

18. Academic research versus policy needs (Bueno de Mesquita)

19. Possible integrative chapter based on some schema such as Barry Hughes's International Futures model (Wayman and Williamson)

20. Methodological critique of chapters and suggestions for next steps (Polachek)

21. Conclusions: assessment of current and anticipation of future of forecasting methods (Bueno de Mesquita)

Review of all topics to be addressed by Bruce Bueno de Mesquita, Jim Morrow, Sol Polachek, J. David Singer, Frank Wayman, Paul Williamson. (Some of those in this category are now listed also to present papers, or may choose to do so.)

PARTICIPANTS

Keynote speaker:
Edward O. Wilson
Pellegrino University Research Professor
Harvard University

Other confirmed participants

Richard D. Alexander
Emeritus Professor
Insect Division, Museum of Zoology
University of Michigan

Bruce Bueno de Mesquita
Chair of the Department of Politics, New York University
Senior Fellow, Hoover Institution, Stanford University

Claudio Cioffi-Revilla
Professor of Computational Social Sciences /
Director, Center for Social Complexity
George Mason University

Doyne Farmer
McKinsey Professor, Physics
Santa Fe Institute
Santa Fe, New Mexico

John Holland
Professor of Psychology and
Professor of Electrical Engineering and Computer Science
University of Michigan

Myron S. Karasik
Partner, Tatum Partners
San Francisco, California /
Chief Financial Officer, EMSO
Palo Alto, California

Urs Luterbacher
Chairman, Environmental Studies Unit
Graduate Institute of International Studies
Geneva, Switzerland

Gottfried Mayer
Adjunct Professor,
Pennsylvania State University /
Editor, Complexity Digest (www.comdig.org)

James D. Morrow
Professor of Political Science
University of Michigan

Solomon Polachek
Distinguished Professor of Economics
SUNY Binghamton

Gerald Schneider
Chair of International Relations
Faculty of Public Administration
University of Konstanz
Germany

J. David Singer
Emeritus Professor of Political Science
and immediate past president, Faculty Research Club,
University of Michigan

Detlef F. Sprinz
Potsdam Institute for Climate Impact Research
Potsdam, Germany

Atsushi Tago
JSPS Research Fellow
University of Tokyo at Komaba

Karin Verspoor
Computational Linguist
Los Alamos National Laboratory

Frank Wayman
Professor of Political Science /
President, Faculty Research Club
University of Michigan

David Wilkinson
Professor of Political Science
University of California, Los Angeles

Paul R. Williamson
President, Global Vision, Inc.
Ann Arbor, Michigan

POSSIBLE PUBLISHERS

The logical place to start would be with M.I.T. press, and in particular their editor John Covell, since they have done Cooper and Layard, eds., What the Future Holds (2002). John Covell is also Bruce Bueno de Mesquita's long-standing editor (first at Yale and now at M.I.T.) and has now brought out at M.I.T. Bueno de Mesquita, Smith, Siverson, and Morrow, The Logic of Political Survival, another text related to our proposed work. Other viable alternatives at this early date include Lynne Reiner (the only publisher we've actually spoken to), and U. of Michigan Press (which has a tradition of publishing political science methodology and has obvious connections to the numerous U. of Michigan faculty on the above list).

REFERENCES

AXELROD, ROBERT (1984) Evolution of Cooperation. N.Y.: Basic Books.

BUENO DE MESQUITA, BRUCE, DAVID NEWMAN, and ALVIN RABUSHKA (1985) Forecasting Political Events: The Future of Hong Kong. New Haven: Yale Univ. Press.

BUENO DE MESQUITA, BRUCE, ALISTAIR SMITH, RANDY SIVERSON, and JIM MORROW (2003) The Logic of Political Survival. Cambridge, Mass.: M.I.T. Press.

CAMPBELL, D.T., and J. STANLEY (1966) Experimental and Quasi-Experimental Designs for Research. Chicago: Rand McNally.

COOK, THOMAS, and D.T. CAMPBELL (1979) Quasi-Experimentation. Chicago: Rand McNally.

COOPER, RICHARD, and RICHARD LAYARD, eds. (2002) What the Future Holds: Insights from Social Science. Cambridge, Mass.: M.I.T. Press.

GOLDSTEIN, JOSHUA S., XIAOMING HUANG, and BURCU AKAN (1997) "Energy in the World Economy, 1950-1992." International Studies Quarterly 41 (No. 2): 241-266

HUGHES, BARRY (2004a) "International Futures: An Introduction to Structure." Lecture and software demonstration, Univ. of Michigan, April 2.

HUGHES, BARRY (2004b) International Futures student version at www.du.edu.

JENSEN, LLOYD (1972) "Predicting International Events," Peace Research Reviews, Vol. IV, No. 6, pp. 1-66.

WILSON, EDWARD (1998) Consilience: the Unity of Knowledge. N.Y..: Knopf.

This site created with the technical support of Nonprofit Enterprise at Work's Nonprofit WebConnect Program, sponsored by SBC.