Physicists insist that time travel is impossible. For physical objects, the speed of light cannot be exceeded, and that velocity is too firmly tied to the passage of time to be violated. Another form of time travel exists, however, one that's mental instead of physical. Using various numerical tools, we can undertake "travel" into the future. By doing so we shift the foundations of how we undertake planning and make decisions. How we do this and how well we've succeeded and what that bodes for the future is the theme of this study. A clearly organised and well-written effort, this book should have significant impact.
In Greek mythology, "Apollo's Arrow" [the book's original Canadian title] was a form of time machine. Those fortunate enough to seize the passing arrow could be conveyed over obstructions and help perceive events. Orrell uses this as a metaphor to examine the forecasting of three significant elements: weather, health and wealth. Although at first glance these seem wildly disparate, he explains how the methods applied to them are very similar. And with similar results.
Orrell opens with a discussion of the ancient "forecasters" of Delphi where the prophecies of the Pythian Apollo were expressed. Although these prophecies were obscure and possibly misleading, people made decisions based on what they believed was said. From the beginning, some petitioners to the Oracle were favoured over others, allowing them to dictate conditions. The mathematician Pythagoras added a new dimension to the forecasts by applying the power of numbers to them - although the method by which this was accomplished remains unknown. Nevertheless, today's forecasting is based on numerical analytical methods. Are they a real improvement over Apollo's expressions?
Orrell begins answering that question with everybody's favourite topic - the weather. Although the ancients made many attempts to understand the workings of weather, it wasn't until the Renaissance that real gains were achieved. The invention of the thermometer and barometer initiated measurements that could be recorded and analysed over time. Further technological leaps seem to bring better forecasts. Robert FitzRoy, "evolution's captain" initiated the first forecasting service across the British Isles. Although complex mathematical models have since ensued, Orrell argues that the systems under investigation are too vulnerable to small perturbations to allow truly reliable forecasts of weather systems.
The same inhibition holds true for the other fields of Orrell's presentation, health and wealth. No matter how well refined and tested the model, little incidents or influences can skew the final pictures. Small, almost undetectable factors have the capacity to set in train a cascade of unforeseeable outcomes, rendering the most carefully conceived model ineffective. Trying to fit the model into the real world's events as they unfold results in the designers engaging in hand-wringing and often weak excuses. Orrell is mildly scornful that failures of models predicting events don't seem to discourage the modellers from making strong assertions about how well they are doing. In his mind, this can be amusing in some cases, but disastrous in others. He would not abandon numerical models or forecasting, but insists that these techniques be approached and used realistically. Those affected by the models will also have more realistic expectations. That is a message that needs wide exposure and it's hoped this book will help provide that. Uncritical acceptance of forecasts, no matter how authoritative they may appear, can lead to serious consequences. Understanding the limitations and shortcomings is vital [stephen a. haines - Ottawa, Canada]
on February 16, 2007
This book concentrates on three main, seemingly unrelated, subjects: the weather, human health and the economy. In particular, the focus is on the shortcomings of the computer models that are used to make predictions in each of these disciplines. The author's argument is that "model error", more than any other reason such as chaos theory, is to blame for the models' inability to accurately predict future (especially longer range) developments in these areas. The writing style is clear and friendly, as well as quite engaging. I did find a few passages a bit heavy going, requiring a couple of re-reads so that I could better grasp what the author was saying. This is a fascinating book from which I learned quite a bit about the ways in which forecasts are made and why they are so often wrong. This book should be of interest to everyone with inkling towards the above disciplines, but especially science buffs.
on July 26, 2007
For years most of us have been hearing about "models" of just about everything from weather and hurricane predication to the stock market. Since the advent of the desktop computer, almost everyone seems to have a model of something or other. For those of us who are not modelers, it all sees so cut and dried. Orrell, however, definitely pops the bubble on the works.
The author points out that many of the models, contrary to what most of us understand, are designed to "predict" the past. The presumed variables involved in a particular phenomenon are put into the form of an equation, numbers are estimated for each, and the equation fed to a computer. The problem is that both the variables chosen and the numbers selected to define them are arbitrary and subject to the estimates made by the modeler. I first ran into this when I was taking a structural geology course for my BS in geology. The text went into an elaborate description of a river's output and presented an equation that was said to help estimate the watershed that fed into the river. After looking over all the letters in the equation and what they represented, I realized that most if not all of the variable were almost impossible to measure directly in any accurate way!
Orrell indicates that this is the case with nearly all models. Furthermore he notes that in order to test the results of the model, it is compared to the past to see how well it "predicts" what has occurred. Then the model is tweaked to make it come closer to what was actually seen in the past to "fix" it, on the assumption that the future will be the same as the past.
Most of us would probably accept this as a plausible method of approach, but the author notes that, while events may seem similar, history is perforce not repeatable. He also notes that because life and the earth itself are complex systems in a state of dynamic equilibrium, they are inherently changeable. The estimates of the variables might be anything, and the outcome will change as the estimates do. Because of the complexity of systems in a constant state of change, the interactions of all the variables in the system are inherently beyond our calculation. Thus the more detailed the model, the more subject to error is becomes.
When I first read the author's comments on the weather and other phenomena, I was certain that he was among the nay-sayers over global warming and waited for his objections to the current trends aimed at correcting human impact on the environment. Not so. The author states emphatically that global warming, while the specific outcome is not predictable, is obviously occurring. I suppose it's difficult to ignore a missing glacier. What he does note is that our romance with science and data manipulation has encouraged us to be over optimistic about our ability to control nature. We place too great a trust in models and what their designer's tell us they mean and take mistaken courses of action with respect to climate change, epidemic diseases, and the economy. The entire book is intended to alert and caution the general public about grand claims.
A superb book.
on November 24, 2007
This book makes exceptional reading for a young scientist-to-be, or for an adult with a broad range of interests. Orrell's writing is light and enjoyable while still presenting an accurate history of philosophy, mathematics, and science. For me, the book provoked a great deal of thought, leaving me feeling like I could have a much greater interest in science and math than I would have thought. Stories of philosophers, mathematicians and scientists make for great reading when combined with Orrell's off-hand remarks and fluid writing style. Some of his enjoyable off-sides will age quickly--relating his historical tales to current times with references to Microsoft and George W. Bush for example. [I predict that these will become dated sooner rather than later. ;)] However, these slight flaws do not hinder Orrell's achievement of a delightful and easy-to-read look at matters that normally would seem a daunting study. If you have enjoyed books like Metamagical Themas, you will love this book!
on March 11, 2007
How can we predict the future, and can past discoveries help interpret tomorrow's events, from weather to finances? It's time for another probe of prediction science and THE FUTURE OF EVERYTHING: THE SCIENCE OF PREDICTION offers the latest research and methods, examining how past scientists predicted the future and how modern scientists forecast events. The author received his doctorate in mathematics from Oxford: his background provides a grounded, rational examination which considers the pros, cons and uncertainties of prediction science.
Diane C. Donovan
The book in general is easy to read, but some sections require some basic knowledge in math, economics and biology, though you may skip the excessively technical paragraphs, jump to the end of argument and still understand the general idea. The first three chapters present a very good summary of the history, philosophy and development of science, starting with the Greeks. A real crash course for those not familiar with these subjects and a necessary background to better understand the main topics being discussed in the book: The science of prediction in the fields of climate, health, and economics, and what I consider a very objective critic of simulation models and other techniques used for predictions on these fields. The book includes technical appendices, notes, a glossary and a full bibliography, so you can do follow-up or check the facts by yourself.
The explanation on Chapter 3 on the subject of complex systems is short but outstanding, allowing the layman to understand the basics without the confusion of the math involved. This explanation is fundamental for understanding the limitations of the science of prediction in non-linear systems, such as climate and economics, particularly because it makes clear that these models are not based simple on mathematical relationships reflecting cause-and-effect explanations, like Newton's laws. In astronomy you can calculate where the moon is going to be tomorrow at 5 am, this book makes crystal clear that complex systems are not like that, they are incomputable. And even if existing models can be twisted to fit past data, they cannot predict the future, as it is the case with many models in Economics and the climate simulation models supporting the consensus theory of Global Warming. It is well known that economists have developed models that can explain past external shocks, recessions, commodity booms, Dutch Disease, etc, so they understand conceptually what happened then, but they not predict the future, because neither the economy nor the climate are constrained to follow past behavior. The conditions of the initial variables are not the same, history does not repeat. Technically speaking, positive and negative feedbacks, and multiple feedbacks between the variable result in the inherent unpredictability of complex systems. For example, all 18 models used by the UN's IPCC 2007 Report cannot account for the clouds feedbacks, which may result in much lower temperatures than predicted or much higher (look for in the web for Chapter 8 of the UN's Climate Report, to check by yourself this and other important simulation limitations). As Mr. Orrell explained after showing how badly the OECD predicted GDP growth for the G7 countries from 1986 to 1998, "...Consensus between an ensemble of different models is no guarantor of accuracy: economic models agree with one another far more often than they do with the real economy" (pp 243). This statement is valid for climate models too.
Particularly Chapter 6, on economic predictions, is very interesting, but I will only comment on climate predictions, because so much is being discussed in the media and echoed by famous politicians and even Nobel laureates, with total disregard of basic scientific principles, and an absolute absence of scientific criticism or critical rationalism, as Karl Popper called it. Chapter 4, on climate forecast or prediction, begins with a brief but comprehensive summary of the history of meteorology and climate forecasting, which is important to understand the limitations of modern long-term predictions. As remembered to us by Mr. Orrell, Copernicus and Darwin hold publication of their works because they were afraid of the consequences, since their theories were against the scientific consensus of their times. Unfortunately, most of the environmental movement is blocking any serious discussion of the science behind the theory explaining the causes of Global Warming (the incomplete science has to do with the cause and effect relationship, not with the indisputable fact that most of the world is getting warm), which is based mainly on climate simulation models and assumptions about the state of affairs of the world for the next 100 years. And if you dare to contradict them, you become a heretic, since most environmentalists are acting as if defending a dogma. Not to mention that science must be politically neutral, as quite rightly cited by Mr. Orrell (pp. 107).
Since I do not want to spoil the contents of the book, let me just say that this book is a welcomed light of hope in the middle of the media and political frenzy regarding the real causes of Global Warming. We should be doing real science instead of politicizing science, and as Mr. Orrell recommends at the end of the book, "Apollo's arrow cannot fly to the future or protect us from plague, but it may serve as a compass, point out dangers, and help us navigate an unpredictable world", of course this is possible, if climate scientist stop playing politics and doing the science as they should, objectively and apolitically. Finally, someone has the courage to clearly explain what's wrong with the science behind the consensus theory explaining the Global Warming, as well as in other scientific and social fields.
The final chapter, "Consulting the Cristal Ball" is a must reading. Mr. Orrell presents quite a collection of ideas, scenarios, predictions, and concerns regarding how things will look in the year 2100, together with a box with some great predictions from the past. Just try to image how anyone would have made a reliable prediction of today in the year 1900 (such as cars, airplanes, television, CDs, iPods, computers, atomic bombs, you name it).
If you are serious about understanding the science behind Global Warming, this book is a must. Read it and as previously mentioned, it is worthwhile to search the web for the IPCC 2007 Report, Chapter 8, which presents the evaluation of the simulation models used by the U.N. and their present limitations. You will see that Mr. Orrell is right on the money, there are plenty of positive and negative feedbacks that these models can not replicate and other "anomalies" pending sound explanation. A highly recommended reading for follow-up is Marcel Leroux's "Global Warming - Myth or Reality?: The Erring Ways of Climatology" (too bad this is an expensive book!). We are in need of good old objective science.
Finally, after reading this book it becomes clear that climate simulation models lack any real explanatory power and are incapable to make any reliable predictions, so it seems appropriate to close with a quotation from the best known (Nobel Prize) advocate of the manmade GW based on climate simulation models:
"I have learned that, beyond death and taxes, there is at least one absolutely indisputable fact. Not only does human-caused global warming exist, but it is also growing more and more dangerous, and at a pace that has now made it a planetary emergency". Al Gore, "An Inconvenient Truth: The Planetary Emergency of Global Warming and What We Can Do About It", 2006.
on February 9, 2009
Orrell has written a spot-on truthful book about prediction. It is therefore somewhat dull. Sorry, but actual truth is often dull.
I enjoyed the book. He talks about prediction in general but uses the examples of weather prediction and financial market prediction. [MINOR SPOILER] He describes what he believes are the three main impediments to prediction: chaos (butterfly effect), measurement error, and inaccurate modeling.
There are a lot of parallels to the financial markets. Very interesting.
He devotes a little at the end of the book to the issue of global warming. He believes the global warming deniers are wrong. But he believes that Al Gore is also wrong. It is quite possible his position is the correct one. But its vague and boring. I doubt it makes the news.
All in all, a bit dry but enjoyable if you push through it. He is not going to be well liked by people who make predictions for a living.
on October 10, 2009
This was the first book I read on my new Kindle and I must say it was an electronic page-turner. It is unique in the way the author provides an extensive history of prognostication and highlighted the realms of human activity where prediction themes have common threads; weather, health, and economics. What I liked most of about the book was the way the author was able to cogently present the counter intuitive concept that the more tools we develop to attempt to predict the future, the more muddled prediction becomes. More granularity required to make finer predictions results in simply more sources of error.
Despite the consistently terrible track record of prognosticators from ancient times until the present, we are our leaders continue to base decisions impacting many people on what now is the contemporary equivalent of Taro Cards. As Orrell writes "There is a curious disconnect between the consistent inaccuracy of the forecasts and the confidence with which politicians, banks, and business leaders regularly use to make important decisions".
The book share some of the same themes with "Useless Arithmetic: Why Environmental Scientists Can't Predict the Future" by Orrin Pilkey which is also worth reading. The take home message is just because you can reduce things to numbers does not make them objective and true. Mathematical models can in some sense imitate the past but that does not mean they can predict the future.
My only disappointment is that the author appears to be genuinely conflicted by his conclusions. The final chapters of the book deal with the mother of all predictions, that being global warming. All the limitations identified in all other predictive models clearly apply to climate models. However, Orrell appears to be hesitant in being as pointedly critical of these models as he is of models of economic and short term weather forecasting. Perhaps I can understand his hesitancy given that well meaning critics of the science of climate change have been accused of being the equivalent of Holocaust deniers.
David Orrell tackles a hard subject that matters a lot -- the science of prediction.
Unfortunately, as Orrell tells us, where accurate prediction would help the most (say, with climate change and the stock market), there is more art to prediction than science. The billions spent on developing climate models do not help us predict the climatic future. No one has come up with any better global warming predictions than Svante Arrhenius made in 1896.
Orrell argues that we should see the world feelingly, rather than be blinded by our mental models. We need to know that we do not know. Grasping for illusory knowledge by over-modelling our environment is part of the problem, not part of the solution.
Orrell does not argue against mathematical models. They help us understand complex systems. In fact, we need a model to understand a complex system. While models do not let us predict the future with certainty, if at all, they do help us understand what is happening now. And that makes models indispensable.
Where does Orrell come out on climate change? He thinks we should have fewer children, pollute less, and tread more lightly on the earth. But not because of what any model says. Because of what Orrell feels. As he says, life is not a predictable machine. Life is a surprise.
Some other reviewers found parts of the book hard to read. I did not. For me, this book was a pleasure to read from start to finish. A fairly informal tone. Very thoughtful. Great examples (for example, "if Bill Gates attends a baseball game at Safeco Field in Seattle, the average net worth of those also in attendance increases by a factor of four"). And a pulling together of facts and theories not just from mathematics, but from a broad range of human thought.
(I must admit, though, that my enjoyment of the entire book may come from another reason than just my ability to comprehend more difficult passages than most readers. My guilty secret? I skip over the hard parts. I learned to do that in law school.)
Another good book in a similar vein is The Fortune Sellers: The Big Business of Buying and Selling Predictions, by William Sherden. Its take on the subject is less the science and more the popular culture. Some may like that approach better. (Although that book, published in 1998, is a little dated.)
With my interest in peak oil and climate change, this book helped me greatly. Even those without that interest will, I think, enjoy David Orrell's book. It's good. Very good.
on June 30, 2007
Orrell(O)does an excellent job in demonstrating why predictions and forecasts in a number of fields fail.I will concentrate my review on the roughly 100 pages that deal with economics,primarily macroeconomics,financial economics,and portfolio analysis.O simply demonstates that the current near universal belief in economics that you can model all markets either as or "as if" they were normally distributed is false.Again and again and again near catastrophic economic events occur or happen in the financial and world markets which are statistically impossible if all markets were normally distributed.O demonstrates that the Efficient Market Hypothesis(EMH)is not supported by either the historical or statistical record.The time series data is, in fact, easily shown to be correctly represented by the Cauchy distribution or by Power Laws[ Power laws are of the form y=k times(x raised to the rth power)where k and r are constants.An example of a power law is the formula for the area of a circle:Area=pi times r squared) .The major problems with the assumption of normality are the assumptions of homogeneity, continuity and independence.None of these assumptions hold for time series data.
O correctly gives John Maynard Keynes,Frank Knight,and Benoit Mandelbrot their just recognition for demonstrating that an economics profession that bases its policy analysis and prescriptions on the assumption that market mechanisms generate outcomes that are normally distributed ,and hence stable over time , leads to disasterous results for society as a whole as the necessary institutions that are needed to deal with the inherent,endogenous or internally generated destabilizing shocks are dismantled.
I have a few criticisms of the book.First,Orrell fails to mention that it was the wise Adam Smith who first recognized explicitly that the wealth generating process of free market labor specialization and division of labor internally created massive negative feedbacks on the work force which could only be mitigated by direct government interventions in the economy through massive spending on education and religious instruction,which was needed to counteract the purely economic misrepresentation of "self interest"(WN,1776,Modern Library(Cannan) edition,pp.734-735;also see pp.716-768).Second,Orrell has failed to give proper credit or mention to the work of J Schumpeter(indeterminancy and instability of an economc system where needed investment would occur with a regular irregularity and result in the creative destruction of many businesses over time) and Daniel Ellsberg(ambiguity).