Thursday, June 6, 2019

Thinking in Bets

The world is an uncertain place, and according to Annie Duke, author of Thinking in Bets, we’re better off accepting that we don’t know for certain what will happen than to keep trying to guess the future. It’s better to assign probabilities to different outcomes and adjust them as necessary rather than pick an outcome and doggedly defend that choice. Individually her ideas are familiar, but taken collectively, they amount to a rational truthseeking lifestyle.

Duke is best known as a highly successful poker player, and although this isn’t a poker book, she uses a number of vivid poker stories (and many non-poker stories) to illustrate her points. She says that “all decisions are bets” because your choices lead to gains or losses of money, time, and other things that matter.

One of Duke’s early points is that we need to avoid “resulting,” which is a poker term for judging a decision by how things worked out. Walking across a highway blindfolded isn’t a good idea, even if you don’t get hit by a car.

One interesting point is about how we form beliefs. We think we hear an idea, “think about it and vet it, determining whether it is true of false,” and then “form our belief.” In reality, when we hear something we tend to believe it to be true, and “Only sometimes, later, if we have the time or the inclination, we think about it and vet it, determining whether it is, in fact, true or false.”

“Our default is to believe that what we hear and read is true.” “It doesn’t take much for any of us to believe something. And once we believe it, protecting that belief guides how we treat further information relevant to that belief.”

“Fake news isn’t meant to change minds. As we know beliefs are hard to change. The potency of fake news is that it entrenches beliefs its intended audience already has, and then amplifies them.”

So what can we do about the way we form beliefs and harden them? Among poker players, an antidote is “wanna bet?” When someone challenges us to a bet based on our beliefs, it leads us to ask ourselves many good questions: “How I know this? Where did I get this information? Who did I get it from?”

Rather than relying on others to challenge us to bets to help weed out wrong beliefs, Duke suggests “Incorporating uncertainty into the way we think about our beliefs.” Instead of believing something is 100% true, we might admit we’re only 70% sure. This makes it easier to reduce this probability in the face of new evidence instead of rejecting the evidence and doggedly sticking to the original belief. “This shifts us away from treating information that disagrees with us as a threat, as something we have to defend against, making us better able to truthseek.”

We have a tendency to believe our good outcomes are the result of our skill, and bad outcomes are the result of bad luck. This interferes with learning to improve our skills. Duke’s “learning loop” depends critically on correctly identifying both good and bad luck. To develop skill we start with a belief, take some action that amounts to a bet on that belief, view the outcome, throw out the part of the outcome due to luck, and update our belief. If all bad outcomes are bad luck, then we won’t develop skill. Failing to recognize good luck impedes learning as well. I had extraordinary good luck investing in 1999, and it took me about a decade to see that I was better off with index investing.

Our thinking about luck can slow learning in another way. Our habit of being competitive with others and thinking them lucky when they succeed impedes our ability to learn from their successes and failures.

Duke suggests forming a group of friends or colleagues committed to truthseeking and “thinking in bets.” This helped her learn to play poker at a very high level, and she sees benefits in business and other activities. She sees the failure to truthseek as a reason why some groups fail. In my experience, another big reason for business failure is self-interest. If you need to keep your job, you agree with the boss, even when you think he or she is wrong. It’s a rare employee at any level who feels free to disagree with a company’s top management.

A young Duke was told “It’s all just one long poker game.” This is “a reminder to take the long view, especially when something big happened in the last half hour.” I make a similar point when I talk about “the 1000-foot view.” I get a lot of benefit from trying to keep things in perspective. Few outcomes that seem terrible in the moment are very important in the long run.

Another important idea in the book is that when we make a choice, there are several possible outcomes. In our financial lives, this way of thinking is critical. Just because a bet has a positive expectation doesn’t mean it’s a good idea. If one of the reasonably-likely outcomes would be devastating, we should make a smaller bet or not bet at all.

Overall, I enjoyed this book for its vivid explanations and the way it tied together many familiar ideas into an action plan. Duke offers a set of tools for overcoming some of our natural tendencies to keep believing things that are wrong. Even a small improvement in this area can pay large dividends in life.

2 comments:

  1. Loved the book.

    The "wanna bet?" version in the talk you posted of her last week was more impactful for me. "How much of your networth are you willing to bet on it?"
    More relatable.

    ReplyDelete
    Replies
    1. @aB: I've used the "wanna bet?" idea before, but only in highly adversarial contexts. One time when I was a student, a senior engineer was talking about the Monty Hall problem, and he declared very publicly that the odds were 50/50 and that I was a fool for thinking they were 2:1. I offered to split the difference between probabilities 2/3 and 1/2 at 7/12, and we'd iterate the game 100 times at 7:5 odds. The idea was that he'd pay me $5 if I got the right door, and I'd pay him $7 if I didn't. My expectation would have been $1 per play, or $100 total. He got quiet. Later he approached me privately to tell me he'd figured out he was wrong.

      Delete