Frequently asked questions on Apollo's Arrow/The Future of Everything, by David Orrell

Q: Reason for writing the book?
A: In 2000 I was doing my Ph.D. at Oxford University on the subject of weather forecasting. Everyone knows that weather forecasts tend to go wrong after a couple of days. At the time, the dominant theory was that the errors were caused by chaos - the so-called "butterfly effect." My work showed that actually the problem was errors in the model. The scientists were using chaos as a kind of fig leaf to explain bad forecasts. This raised for me a number of questions about the science and sociology of forecasting, which this book attempts to answer. Also, while there are many books about predictions, there are rather fewer by people with direct knowledge and experience of mathematical models.

Q: Is the book very mathematical?
A: No, it does not use equations, except a few in the appendices, and it is aimed at the general reader. The book is divided in three parts. The first is a history of prediction, from the oracles of ancient Greece on. The second part discusses how predictions are currently made in the areas of weather, genetics, and economics. The final part looks at how the three combine in long-term forecasts for the planet.

Q: What do areas like weather, health, and economics have in common?
A: The three areas of prediction share a similar history and use very similar methods. The systems being predicted - the atmosphere, the human body, and the economy - also have a similar level of complexity. Finally, the three are also frequently interrelated - Hurricane Katrina caused an economic crisis in New Orleans, as well as fears of a disease outbreak. A growing world economy is more likely to produce climate change which enhances the spread of disease-causing insects. So understanding one area of prediction helps understand the others.

Q: Why are we so fascinated by predictions?
A: Humans are hard-wired to be interested in predictions, because the ability to think about the future is vital for our survival. However we are not good at going back to check whether the predictions are accurate - which is why there are so many well-paid forecasters in areas like economics, despite their poor track record of success. There is a similar demand for predictions in fields such as genetics or health. In 2006 many forecasters believed that avain flu would cause the next deadly pandemic, but fortunately that has not yet happened. False alarms can be good if they motivate us to protect ourselves against future threats.

Q: Why is it so hard to make accurate forecasts?
A: The predictions are made using mathematical models, which suffer from two problems. The first is that they cannot capture the full detail of the underlying system, so rely on approximate equations. The second is that they are sensitive to small changes in the exact form of these equations. This is because complex systems like the economy or the climate consist of a delicate balance of opposing forces, so a slight imbalance in their representation has big effects. The models can be adjusted to fit past data, but still fail to predict the future. Increasing the size of the model doesn't necessarily help, because the number of unknown parameters just increases. This is why predictions of things like economic recessions are still highly inaccurate, despite the use of enormous models running on fast computers.

Q: What is Apollo's arrow?
A: According to legend, the Greek philosopher Pythagoras was sired by Apollo, the god of prophecy who provided the predictions for the Delphic oracle. At one point, Pythagoras was presented with an arrow said to have belonged to Apollo, which had magical powers that allowed him to dart across space and time, cure plagues, and so on. He founded a school - really a cult - that taught prediction using numbers, and worshipped Apollo. Many regard the Pythagoreans, as they were known, as the founders of Western science. So Apollo's arrow is a metaphor in the book for numerical prediction. Today we use mathematical models to dart into the future.

Q: Of the different kinds of prediction discussed in the book, climate change is the most contentious. Is there a connection between predictions of the short-term weather, and the long-term climate?
A: It is often said that predicting the climate is much easier than predicting the weather. However both types of prediction are based on similar models, and suffer from the same kinds of model errors. One of the largest sources of error is clouds. Cloud formation is a complex process that depends on myriad local interactions between water vapour, air, and microscopic particles that act as seeds. This cannot be precisely modelled - there is no equation for a cloud. Modellers therefore use approximate formulas. This is why predictions of rainfall and other precipitation are so unreliable. Because clouds also play a key role in regulating the climate, climate predictions are also sensitive to small changes in the way clouds are represented.

Q: Can scientists accurately predict climate change?
A: No. One indicator is that in recent decades there has been amazingly little progress in prediction accuracy. Back in 1979, some climate scientists met to estimate the likely effects of doubling the level of carbon dioxide in the atmosphere. They guess-timated a range of 1.5-4.5ºC, with an average of 3ºC. The Intergovernmental Panel on Climate Change (IPCC) was founded to refine the result using advanced mathematical models. Their most recent summary statement concludes that "It is likely to be in the range 2 to 4.5°C with a best estimate of about 3°C, and is very unlikely to be less than 1.5°C." So there has been no improvement in 28 years, despite huge increases in technology, computers, and the number of scientists. The range really represents a kind of fuzzy, social consensus in the climate community, which is heavily influenced by precedent (as the book shows, the uncertainty in the models themselves is even greater). When different economic scenarios are taken into account, the IPCC's final range is about 1-7°C, which barely qualifies as a prediction - either there will be little change, or it's the end of the world. Climate scientists have become very good at understanding the current climate; but their forecasts for the future are trying to be both incredibly vague and authoritative at the same time, which in the end just confuses people.

Q: Does the book argue then that climate change is not a problem?
A: No, I believe that we are having a dangerous effect on the climate, but I also believe that we can't predict its future. Most environmentalists gloss over the problems in the models. This is dangerous for two reasons. The first is that bad science will never convince skeptics. The second, more subtle reason is that the emphasis on mechanistic models and technology perpetuates the idea that we can predict and control the planet, which is what got us into trouble in the first place. So what we need is a different, more humble approach. As I discuss in the book, many environmentalists embrace Gaia theory (named for the Greek Earth goddess) which states that the Earth can be viewed as a living organism. If the Earth is alive, that makes it deserving of our protection, but at the same time rather unpredictable. So there's no contradiction in believing that climate change is likely to be a problem, but is beyond our computational abilities. Interestingly, according to legend, the prophecies at the Delphic oracle were initially read out by Gaia, before the oracle was taken over by Apollo.

Q: If it is so hard to predict the future, then how can we make a decision about issues like whether we should cut back on carbon emissions?
A: In most areas of life we don't rely on complex mathematical models to make decisions. For example, if you are choosing whether to vote for a particular politician, you can't predict exactly what they will do when they get in office, but you can still make an informed judgement and vote accordingly. Another example is diet. If a child is eating a lot of candy bars, and is putting on weight, then we might suggest they cut back - even if we can't mathematically prove that they will become dangerously overweight or develop diabetes. Carbon dioxide is like sugar for the planet - it boosts plant growth and makes the climate warmer - and observations show that our emissions are having a dangerous effect. In any case, our environmental problems extend far beyond global warming, and there are many reasons to reduce our impact on the planet. The realization that we are out of our depth may open the way to a more immediate and direct response. As I discuss in the book, mathematical models are powerful and essential tools in many areas of science, and they help us understand complex systems like the weather, the economy, or our own bodies. But we shouldn't rely on them to predict the future.