After finishing this book in September of 2023, I wrote,

 

"I don't normally indulge in spending $30 on a book, but after reading the free sample of this book, I decided that the investment fee was probably a small price to pay for the lifetime value I would get from this book. I am happy with that investment."

 

My clippings below collapse a 222-page book into six pages, measured by using 12-point type in Microsoft Word.

​

See all my book recommendations.  

​

Here are the selections I made: 

​

Most of the focus is on decisions about decisions, but (a confession) the lens will occasionally shift to decisions, period.

 

Chapter 3 deals with the decision whether to know. Knowledge is power, as they say, but ignorance is bliss, as they also say. They are right on both counts. The challenge is to know whether we have a knowledge-is-power situation or an ignorance-is-bliss situation.

 

Chapter 5 looks at beliefs in general. It suggests that beliefs are, in a sense, like goods, and we decide whether to “buy” them. We often make a rapid decision: If I believe this, will my life be better? How?

 

Chapter 6 turns to the problem of inconsistency in decisions. With respect to

 

Second-order decisions involve the strategies that people use in order to avoid getting into an ordinary decision-making situation in the first instance.

 

There are important issues here about cognitive burdens and also about responsibility, equality, and fairness.

 

People have diverse second-order strategies, and a main goal here is to identify them and to understand why one or another might be best.

 

As we shall see, these strategies differ in the extent to which they produce mistakes and also in the extent to which they impose informational, moral, and other burdens on the agent and on others, either before the process of ultimate decision or during the process of ultimate decision.

 

We shall see, for example, that a second-order decision in favor of firm rules (a form of High-Low) is appropriate when an agent faces a large number of decisions with similar features and when advance planning is especially important; in such cases, the crudeness of rules might be tolerated because of their overall advantages.

 

Sometimes ordinary people and public institutions rely not on a rule but instead on a presumption, which can be rebutted.

 

Rules are often contrasted with standards.5 A ban on “excessive” speed on the highway is a standard; so is a requirement that pilots of airplanes be “competent,” or that student behavior in the classroom be “reasonable.” These might be compared with rules specifying a 55-mph speed limit, or a ban on pilots who are over the age of seventy, or a requirement that students sit in assigned seats.

 

Sometimes a reasonable way to deal with a decisional burden is to adopt a routine.

 

A forgetful person might adopt a routine of locking his door every time he leaves his office, even though sometimes he knows he will return in a few minutes; a commuter might adopt a particular route and follow it every day, even though on some days another route would be better; an employee might arrive at the office by a specified time every morning, even though he does not always need to be in that early.

 

A possible way of simplifying a difficult situation at the time of choice is to make a small, incremental decision, and to leave other questions for another day. When a personal decision involves imponderable and apparently incommensurable elements, people often take small, reversible steps first.

 

Sometimes the difficulty of decision, or symmetry among the options, pushes people to decide on a random basis. They might, for example, flip a coin, decide in favor of the option they see first, or make some apparently irrelevant factor decisive

 

A familiar way of handling decisional burdens is to delegate the decision to someone else.

 

People often use heuristic devices, or mental shortcuts, as a way of bypassing the need for individualized choice.

 

Recall that second-order strategies differ in the extent to which they produce decisional burdens and mistakes. Those burdens might be emotional; it might be unpleasant to make decisions on the spot, and second-order strategies might reduce or eliminate that unpleasantness. Those burdens might be cognitive; it might take a lot of time and energy to make decisions on the spot, and second-order strategies might be a blessing.

 

Second-order strategies should be chosen by attempting to minimize the sum of the costs of making decisions and the costs of error, where the costs of making decisions are the costs of coming to closure on some action or set of actions, and where the costs of error are assessed by examining the number, the magnitude, and the kinds of mistakes.

 

As we have seen, delegations may require little advance thinking, at least on the substance of the issues to be decided; the burdens of decision will eventually be faced by the object of the delegation (who may be one’s future self).

 

Picking can even be said to operate as a kind of delegation, where the object of the delegation is “fate,” and the agent loses the sense of responsibility that might accompany an all-things-considered judgment.

 

Typical situations for small steps thus involve a serious risk of unintended bad consequences because a large decision looms when people lack sufficient information; hence reversibility is especially important.

 

Watching television, playing tennis, or going out to dinner with friends might be pleasurable, whether or not any of those activities is meaningful. Helping others, doing one’s job well, or parenting might be meaningful, whether or not any of them is pleasant.

 

We might think that our lives are full of pleasure without being full of meaning; we might think that our lives are full of meaning without being full of pleasure. Of course, the idea of eudaimonia might be taken to include both pleasure and meaning.

 

Existing evidence is supportive of a simple and striking conclusion: People benefit from making large life changes, and they are significantly more likely to show regret, and to be unhappy, if they stick with the status quo.

 

For ending a relationship or quitting a job, for example, the clear implication is that people are more likely to err if they decide to be cautious than if they decide to take some kind of plunge.10 The challenge for individuals, of course, is that statistical generalizations cannot resolve hard questions.

 

They also find that spendthrifts tend to agree with this statement: “When I engage in an enjoyable activity, I prefer not to think of the cost of that activity.”

 

Or suppose that you are fighting a serious illness. Your chances of success, and of a long life, might depend on not knowing the odds.

 

People might want to put off good news simply because it will adversely affect their performance.

 

Some of the most illuminating work on information-seeking emphasizes “strategic self-ignorance,” understood as “the use of ignorance as an excuse to over-indulge in pleasurable activities that may be harmful to one’s future self.”

 

My own evidence attests to the importance of both hedonic and instrumental value.

 

About 57 percent would like to know whether their partner or spouse ever cheats on them. Only 42 percent would like to know what their friends and family members really think about them!

 

Strikingly, 71 percent want to know if there is life on other planets. Perhaps surprisingly, only a bare majority (53 percent) want to know if heaven really exists. Those who did not want to know probably fell within various categories: (1) those who are sure that heaven does exist, so the information would be worthless; (2) those who are sure that heaven does not exist, so the information would be worthless; (3) those who think that they will not get into heaven, so learning of its existence could only make them sad or upset; or (4) those who think that it is best to have a degree of uncertainty. A smaller number (44 percent) want to know if hell exists – which is probably testimony to the fact that if hell exists, a lot of people think that they will be in big trouble.

 

The good news–bad news effect can be seen as a reflection of a form of motivated reasoning known as “desirability bias”:8 People are more likely to shift their beliefs in a direction that pleases them or that makes them feel better.

 

These dimensions can be roughly categorized into two groups: the external outcomes of holding a belief and the internal outcomes of holding a belief. The outcomes of holding a belief can be accuracy-dependent or accuracy-independent.

 

Restaurants and movie theaters are well aware of this phenomenon. A movie theater might offer Option A, which is regular popcorn, and also Option B, which is large popcorn. Knowing that people will choose Option A, it might introduce Option C, which is jumbo popcorn. Maybe few people will choose Option C, but its existence leads more people to choose Option B.

 

But because of comparison friction – defined as people’s unwillingness to obtain comparative information, even when it is available – a problem of evaluability in separate evaluation might persist.

 

People are intuitive retributivists.37 They do not naturally think in terms of optimal deterrence, and they will be reluctant to do so even if they are asked.

 

The goods should be popular, but not too popular. For demi-solidarity goods, the number of users matters and may be crucial to choice. But value neither increases nor decreases continuously as a function of that number.

 

In the case of “pure” solidarity goods, people care about the sheer number of consumers, and in the case of partnership goods, people care about who the consumers are.

 

But the preceding points suggest an additional class of goods, call them fraternity goods, where people care about fellow consumers falling under a certain description or belonging to a particular category (e.g., students, Catholics, athletes).

 

I offer a simple claim here, one that is meant to offer two-and-a-half cheers for deciding by algorithm. The claim is that algorithms can overcome the harmful effects of cognitive biases, which often have a strong hold on people whose job it is to avoid them, and whose training and experience might be expected to allow them to do so.

 

To see why noise can be a problem, return to the medical context. Suppose that doctors order a large number of tests in the morning, but that in the afternoon, they ask patients to go home and take aspirin. Or suppose that when doctors are in a good mood, they make very different decisions from those they make when they are grumpy. If so, doctors might not show a systemic bias of any kind. But they will be noisy, and the noise will be produce plenty of mistaken decisions.

 

Indeed, many people want to retain agency even if they know that if they delegated the decision to another (including an algorithm), they would end up with better outcomes.

 

Some evidence identifies a particular source of algorithm aversion: People are far more willing to forgive mistakes by human beings than they are willing to forgive mistakes by algorithms.

 

People are especially averse to algorithmic forecasters after seeing them err, even if they do better than human forecasters.27

 

There is another factor. People have been found not to trust algorithms, and not to want to use them, in part because they do not know how they work.

 

But research finds that people are more likely to trust algorithms, and to be willing to rely on them, if they are given a simple account of why they work.