Initially, only oracles possessed the ability to forecast into the future. With the introduction of numbers and philosophy on risk, to forecast the future became divorced from the oracles. From knowing the odds in games of chance to practical applications of the statistical methods, risk was able to be measured and managed. Word of caution about the use of statistics is spread throughout the book. That even the best statistics can create gargantuan failures.
The evolution of ideas in statistics changed the way everyone thought about risk. Starting with Pascal’s Triangle showing the probability that a certain combination can show up. Bernoulli gave measure to the measureless by added utility to value making the ideas of risk to be perceived differently per individual. Bayes introduced the revision of prior probabilities to new information, an updating mechanism. Gaussian regression to the mean showed how to obtain generalities from similarities. Quetelet’s normal distribution curve gave the probabilities which can occur outside the mean. Von Neumamm’s and Morgenstern’s development of game theory changed the perspective to look at the behavior and intentions of others in response to another’s and vice versa.
With each new idea came many caveats. Certainty could never be found, as statistics would be useless if there was certainty. The ideas showed the potential probability, the potential risk, not to eliminate risk. When individuals do not fear risk, their behavior changes to taking more risk. The statistical ideas are also given warning, providing the difference between their use and what the ideas actually measure. For instance, the law of large numbers indicates that with more data, the average is more likely to be closer to the true average than a smaller data set. The law of large numbers does not proclaim that errors will diminish as the data set becomes large enough, while the law is usually described as decreasing errors with more data. The data itself contains problems for it may be close to representing what needs to be analyzed, but does not represent it perfectly, creating a problem with relevance.
Access to all data is too expense but the right sampling techniques can show a good representation of the whole. Bernstein shows practical uses which inspired better statistics. From the need to find how to take care of a city, the statistic offices were set up to collect data on population and mortality rates. Coffee houses were data gathering places which started the insurance of shipping. When institutions used statistics incorrectly, their failures provided lessons for other which include that unlikely events should not be taken as impossible events. Risk never disappears and is usually only seen as obvious post hoc.
Written pretty well in most chapter, but transitions from topic to topic are not the most direct. This book serves as a survey of statistic ideas and does not explore topics in depth. Although problems of the ideas are expressed, they are set aside quickly to make room for the benefits. The only problem in statics (and economics) that is explored in detail is the concept of perfect knowledge (the rational being).
How decision making changed when humans took forecasting away from oracles is the core of this book. Explaining what the ideas are and how they changed the story of risk are wonderfully portrayed by Bernstein. Emphasizing the lack of certainty in the probabilities helps elucidate the proper use of statistics. Depending on the situation, a different statistic idea can be better than another. Even with the best understanding of risk, the statistics should be used as a tool to help aid in decision making and not supplant decision making.
Pages to read: 337
Ratings out of 5: