Erik Larson

Sep 30, 2007

Project Outline

Introduction: We’re at another moment, alas

There are moments in history; events that change the entire course of things. They are hard to predict, and they may even be small, taken by themselves. They may be recognizable as such right away, or they may reveal themselves completely only in retrospect. But nonetheless, they are inescapable. The assasination of Archduke Franz Ferdinand in 1914 was a moment—an event of tremendous future significance that later plunged us into The Great War (after which the old European world order was gone, a vacuum filled by new leaders and political ideologies—among them Fascism, later Nazism). Just 25 years later, Germany invaded Poland, the beginning of second world war in Europe. Another moment.

Historical moments tend to lead to comprehensive changes in human societies, from changes in politics and policies to fuzzier changes in pop culture and broad themes in intellectual culture. We get new “isms”, like atheistic existentialism in Paris, 1940s. Comprehensive changes in society happen after moments beacause humans—like every other living thing—have to adapt to continue to survive. We adapt, mainly, by thinking. Humans are constantly trying to adapt to events that we cannot predict by rethinking and retooling our ideas, beliefs, and values.

Adapting to historical moments has become almost constantly necessary. The world really seems unstable, viewed through the lens of the last hundred years. The Cold War was, in a sense, stability, but it was the wrong kind. When the wall came down, we all realized the price of that awful stability, decades of locking horns with the terrible other side. The adaption that began after the Berlin Wall fell was accelerated by information technology: within a few years the conceptual gap between then (big clunky, pre Internet Soviet era then ) and some new, exciting word order was palpable. Thomas Friedman gave us “pop” globalization, Clinton Free Trade, and the Internet something really big to think about. And really progressive. The Gulf War was a blip after which discussions of possibilities defined the times.

But we can’t get away from these moments. Not a decade of relative peace, and a world of possibilities exploded in the most fantastic way conceivable with the fall of the towers in September 11, 2001. Instability everywhere.

What's interesting about our world today, however, is that we have very little consensus on what 9/11 meant, and how we _ought to_ adapt now. It is perhaps distinctive of these times that we have so many opinions informed by so much information, and no consensus. We speak a common language—I understand what you are saying—but we have no shared vocabulary for evaluating competing claims and hence reaching consensus. (Watch the "news" talk shows and understand this point exactly: just add volume and hope for the best.)

The danger, then, is not just in the temporary problems that occupy political discourse—who’s in office this year, when we pull out of Iraq, whether we should raise the CAFE standards. The danger is that fundamentally we’re not problem solving very well anymore. There are, I argue, some very deep reasons for this much worth exploring. Changing our conceptual landscape—or at least shining some light on it—ought to get us going in more promising directions. That’s my hope in this book, if even I’ve shined only enough light to get others talking on productive terms. I hope it helps.

Part 1: Theory

  1. Of causes, statistics, and chaos

The Mechanistic Model

History (Newton)

Laplace, Civil War era thinkers

Success of statistics, new way of viewing things (from Plato's essences, Agassiz to Darwin). Anecdote about the Pierces.

Quantum Mechanics - Shrodinger's Cat

Penrose's suggestion that the world is not deterministic - Godel's Argument

The Statistical Model

But whether the world is, at root, completely deterministic, so that in principle we could predict every state of the world at every instant of time, hardly matters to us today. Supercomputers can't tell us whether it will be overcast or sunny on Monday, when we're planning the big vacation Saturday. Our predictive capacities in the physical world are limited, an observation that is of course consistent with both the Laplacian view and the "New Physics" as quantum mechanics has been coined.

But we do of course use models of the world for prediction. I've spent years building NLP systems that learn predictive models from annotated data (so called "supervised" learning techniques where there is a gold standard of correct examples available to the learning algorithm). When running such systems on unseen data (new text that the learning did not see), we expect with even successful implementations some error. Error is part of engineering. Practically speaking, our scientific interactions with the physical world are exersizes in minimizing error. But error is always there.

Good predictive models have acceptable error. Examples.

So, we give a 70% of showers. When hurricanes are about to make land, computer models, show multiple projected paths, with some more probable (given the model) than others. Mean time to failure informs vehicle manufacturers when parts are likely to fail. Recommendation systems for credit card fraud tell us whether a credit event is fraud subject to some threshold. It could be wrong.

The fact that things break, systems fail, and we all make mistakes is a coin flip observation that things come out as planned, features of the world now give us predictive information about possible feature later. The success of science is mostly a commentary on the success of the Statistical Model.

Turbulence up ahead - The Chaos Model

The publication of Chaos in 1987 introduced the world to nonlinear systems, what makes the world interesting, fantastically complicated, and yes, difficult to predict. Chaos means that predicting the future will continue to be the stuff of science fiction and magicians.

One of the things about chaos is that small changes in intial conditions can lead, over time, to very large changes in the state of a system. The buildings that fell on 9/11 weren't really large targets, judged by, say, fire bombings in Dresden or Nagasaki, or the combined structures hit with missles in Baghdad, to take just a few examples. But the significance of those buildings—the geopolitical conditions that existed then—entered into the intentional world as objects whose destruction by terrorists was of massive significance—the tidal wave set forth from this physical event in the intentional world resulted in two wars—both still ongoing years later, and scores of other changes to the fabric of society well documented in the press (the Patriot Act, for one).

Minds are the ultimate nonlinear system

Penrose (deterministic?)

All of this is rather depressing, actually, when viewed as commentary on the prospects for figuring out what to do next given the very obvious geopolitical instability we witness now nearly everyday. It suggests, of course, that a global system of symbols and facts such as we constitute on our planet is (will continue to be) very resistant to scientific prediction. We literally can't see around the corner with any certainty at all.

  1. Of Plans and Tolstoy

It was predictable that the U.S. would "win", but what about the rest?

What is the best plan relative to some context C (where C contains the best attempt at gathering relevant facts)?

Conclusion: Some predictions are obvious because the initial conditions and final conditions are "tightly coupled". 737s after passing adequate safety inspections, without too many miles, having a good service record, tend not to crash. Even, as Laurence Gonzalez notes about their structure in Deep Survival, given that they have crash potential, nonetheless we predict safe flights all of the time. There is a tight coupling between a conflict whose contestants and their skills are known and objectively comparable. That's why we could predict a quick initial military victory in Iraq. But, let the coupling between prior and subsequent events loosen—for instance, by keeping the American military on the ground after the initial conflict, allowing thousands and thousands of chaotic interactions, decisions, unplanned events to occur, and suddenly it is almost impossible to tell how it will end. We are swamped in chaos.

We might say, then, that we have a horizon of predictability with regard to both the physical and intentional worlds. What exactly is in side our horizon—what we can reasonably predict given what we know—is of course a question of supreme importance to us all. More generally, within our horizon lies our abilities to assess courses of action given initial conditions. In the realm of people "initial conditions" are our traditions, values, and knowledge about each other and the world. What is interesting about people is that they can make changes in the system "from within", as it were. Bad ideas—ones that lead to unwanted outcomes—tend to be replaced with more efficacious ideas and world views over time. We can change the initial conditions. We can, in other words, improve our own view of the world by writing, thinking, talking, and analyzing what is happening today, and how we can make it better.

3. World Views

Pragmatism and the Civil War

The Cold War - welcome back absolutes

Today -

Sam Harris The End of Faith

4. What is necessary today?


We can't stop arguing - why?

MacIntyre's argument

The Rationality of Eric Rodulf (or, something's wrong)

Ethical Theory T - creating new initial conditions in the intentional world - new ways of making sense of why we act


Purpose - what's it all about? nihilism in the modern age

Putting to work the practical progressive mindset - creating different conditions in the physical world

1. Stabilizing Iraq (getting in was "theory", getting out now is too), Middle East

2. Competition: Energy independence

3. Openness to Globalization (nonzero games)