New: Book Report: The Wisdom of Crowds

Ron and Sua were in town on Friday. That's why I was stuck on a train. I'd had dinner with them on the peninsula, caught the train back, blammo. But it was good to see Ron, good to see Sua. This blogging software lets me "tag" my posts, assign them to categories. A while back I named one of my tags "wisdom of ron". This was a joke on the book "The Wisdom of Crowds". I guess if I'm going to make fun of books, I should read them. I finally got around to reading this one.

I'd heard a lot about this book before I read it. I thought it was going to be a "Casablanca" book--you read it, you say "there's nothing original here", and then you realize that's because the ideas have already leaked out into the world so much. Some of that had happened. I'd heard plenty of people talking about the ideas in the book. But the book is more interesting than what I'd heard; some distortions had crept in. So I'll assume that you've heard the same things that I have, but that you haven't read the book (or else why would you be reading this book report?), and I'll try to point out some things I'd heard that weren't quite echoing what the book had to say.

Word-of-mouth says that this book says that a group of 100 randomly-chosen people can make better decisions than an expert. But that's not quite right. This book says that 100 people who are at least kinda interested in some topic can make better decisions than an expert in that topic (assuming those decisions are in the relevant topic). It also says that if you grab 100 randomly-chosen people, not necessarily interested in your topic, they'll make better decisions than you would probably expect.

Of course, it's not so easy to get information from 100 people. Do you poll them? Ask all questions such that they have numerical answers so that you can find the average answer? Markets do this--they want to find out where supply and demand meet. The answer emerges in the form of a price. A market can tell you how much a chair is worth. But if you're wondering: I have two ideas for totally new chairs, one with rockets, one with flowers... you can't poll the market to tell you which you should build. The market isn't a polling service; it can't answer questions about future developments. The only way you can get the market to tell you the answer is to build both chairs, attempt to sell them, and see which one generates more demand. Our political elections can only tell us "yes" or "no". They don't tell us "We'd like another candidate very similar to this one but with more free speech and less blaming violence on video games."

You can get small groups together and ask them to come to a decision. That's a way you can get an information that's more rich in detail than "The price is 12" or "We voted yes". However, you need to be careful of group dynamics. A persuasive ignoramus might sway the answers of some gullible folks, throwing off the results. You want to find a way to let people learn facts from neutral sources, then negotiate with each other to reach a decision. But you don't want to let persuaive people sway other people about what the facts are, what the priorities are. It's not clear how to prevent this, however--to encourage people to be stubborn about some things but pliable on others.

This book was better than I thought it would be based on people's reports. I learned from it; it made me think. I recommend reading it.

Labels: , ,

Posted 2007-05-02