Signal & The Noise The Art and Science of Prediction

Less and less wrong

In our knowledge about the world, Newtonian mechanics is an exception. It allows you to make such precise predictions that you can send a man to the moon and bring him back safely. Outside physics however, unpredictability reigns: a plane flies into the World Trade Center one sunny morning, the prestigious Goldman Sachs bank goes tits up and Germany ends up as Weltmeister. Who would have guessed that?
Nate Silver, a 35-year old statistician at the New York Times, has written a remarkable and far-reaching book on our attempts to get to grips with the whimsical world we live in. Silver earned himself quite a reputation in the 2012 election by correctly predicting the results of all 50 states. He also likes betting on baseball games and playing poker. Here is a man who really loves to see the dice roll. His book The Signal and the Noise ranges over disciplines as diverse as economy, politics, sports, weather forecasting, finance and more. The central question is all this is 'the art and science of prediction'. Since making correct predictions is what practical knowledge is all about and what drives progress.
Silver starts out by analysing prediction failures such as the banking crisis and the frequently mistaking experts on television whose main quality seems to be their lack of recollection.
He then moves into the natural sciences with topics as weather forecasting (the major success story in this book), earthquakes and spread of infectious diseases. What makes predicting in one discipline different from another?
Silver then introduces his hero Thomas Bayes who devised a simple method to improve you predictions about the world. It applies to anything from the probability of breast cancer to the chance of being cheated upon when you spot a long blond hair on your boyfriend's jacket. Bayesian logic (a short mathematical formula) may seem to be tailored to our times of big data and the digitation and storage of everything and everybody, but it actually stems from an 18th century statistician and Presbyterian minister who followed the empiricism philosophy of David Hume.
Bayes' theorem forces you to make a prior estimate of a probability of an event taking place (you bicycle being nicked for example) which may be refined with every following experience. Even if you start out with a wild guess, you end up being less and less wrong about the chances.
This is an exceptional book. Not only for the scope of subjects, but also for the completely new view on knowledge theory that Silver presents. It's a timely book as well since we've never had so much information at our disposal and most of us are frankly a bit at loss with what to do with it. "We think we want information", Silver writes, "when we really want knowledge. The signal is the truth. The noise is what distracts us from the truth. This is a book about the signal and the noise." (JW)

Boek gekocht op aanraden van een vriend vanwege onze statistische achtergrond. Het is een fijn boek, hoewel ik sommige afkortingen en termen niet zo snel terug kon vinden. Verder leest het prettig door de gestructureerde uitleg van de onderwerpen die worden behandeld (zoals de housing bubble in Amerika) en is het behalve ontspannend om te lezen, ook een leerzaam boek. Ik raad het boek aan aan iedereen die regelmatig in aanraking komt met statistiek (al is het minimaal) of er interesse in heeft.

Geschreven bij The Signal and the Noise

Aanrader

Scherp!

Haalt persoonlijke blindspots weg door het gebruik van veel bekende wereld issues gebaseerd op data en kritisch omgaan met data. Is het ruis of net niet (signaal).