Keith Akers's Reviews > The Drunkard's Walk: How Randomness Rules Our Lives

The Drunkard's Walk by Leonard Mlodinow
Rate this book
Clear rating

by
1394552
's review
Apr 10, 11

Read in April, 2011

This book is not really about randomness as such, but about what I would call statistics and probability. (O. K., a semantic distinction, but to me randomness has a connotation of "why was I born?" type of stuff.) It's actually a fairly sophisticated book with quite a bit of difficult concepts that I still don't quite understand. It's well written; you can keep reading even if you don't quite get one point or another. Also, there's essentially no math in it at all. Well, there's some math: you need to know about exponents, about fractions, that kind of thing, and there's even one reference to the "chi squared" test of statistical correlation. But he never explains what this chi squared test is, nor does he expect you to understand it, it's just a landmark along the way. I'm unfamiliar with statistics and so, quite honestly, I don't quite know how to rate the book. An expert on statistics would be a better choice, because they could tell you what the author covered and what they left out.

So the first test that this book passes is that it is well written, and it is not sequential -- you can go to the next chapter, generally, even if you didn't completely understand the previous chapter. However, the weakness of the book is that you're never completely sure what you've covered. But if the author had addressed this weakness, it would be more of a textbook, and probably not nearly as interesting, so this "weakness" might be a "strength" depending on how you look at it. To really understand this book and its strengths and weaknesses, I'd need to read it a second time, looking for basic themes and making sure I understood everything before proceeding further.

On the other hand, it was not just an episodic book with no underlying themes at all. There was a connection, and that connection was different people trying to understand what different results actually mean for the underlying reality. I liked the history of statistics in understanding errors in scientific measurement. A lot of statistics, evidently, just arose in the attempt to understand experimental error, but then it turned out to be a tool for understanding the unknown as well. In this case, the differing results were all the results of experimental error, there was a single underlying constant (and non-random) reality underneath.

One concept I still don't understand is "sample space" and involves the celebrated case of choosing a prize from three doors which hide either something worthless (say, a goat) or very valuable (say, a new car). You choose randomly, but then the master of ceremonies then opens one of the unchosen doors, revealing a goat, and on top of that offers to let you change your choice to the third door if you wish. Acting strictly on probabilities, should you switch or not? I would have been among the many who would have said "it doesn't matter, your chances are fifty-fifty either way." Well, if that's what you think, you need to read this book. While I now understand the logic that it is actually a better strategy, on average, to switch, I still don't quite understand the concept of "sample space." (This particular example is easier to understand if you envision a million doors. You choose one door, figuring one in a million chance might still be possible, but then the master of ceremonies opens 999,998 other doors, revealing all goats. Should you switch? In this case, definitely, it's the difference between a one in a million chance and near-certainty of getting the car. The "sample space" has dramatically changed.)

So, basically, I recommend this book for amateurs such as myself. If I ever get more deeply into statistics (as appears likely at some point) it will be a useful guide down the path. But if you need to know what the "chi squared" test is, you'll need to look elsewhere.
Likeflag

Sign into Goodreads to see if any of your friends have read The Drunkard's Walk.
Sign In »

No comments have been added yet.