It would be wonderful never to make a mistake, nor ever to have a wrong idea. But perfection will always remain out of reach; we’ll always makes errors and oversights.
Joan’s sore knee has been getting worse. Today it hurts her all the time, even when it isn’t touched. She thinks, “I shouldn’t have turned while I lifted that box. And I should have put ice on my knee at once.”
We like to think in positive terms: “An Expert is someone who knows what to do.” And you know how to do most things so well that you scarcely need to think at all; you recognize most of the things you see, and converse without wondering how to speak. However, expertise also has an opposite side: “An Expert is one who rarely fails—because of knowing what not to do.” Thus we usually do not walk into walls. We rarely stick things in our eyes. We never tell strangers how ugly they are.
How much of a person’s competence is based on knowing which actions not to take—that is having ways to avoid mistakes? We don’t know much about such “negative expertise” because this was rarely discussed in Psychology, except in the writings of Sigmund Freud.
Perhaps that neglect was inevitable because we cannot observe, from outside, the things that people do not do. But it is almost as hard to study such things by observing from inside the mind, for example, what keeps you from having absurd ideas. To account for this, we’ll conjecture that our minds accumulate resources that we shall call Critics—each of which learns to recognize a certain particular kind of mistake. Here are a few of those types of Critics; we’ll list more of them in Chapter §7.
A Corrector Critic warns you that you have started to do something dangerous. “You must stop right now, because you’re moving your hand toward a flame.” But such a warning may come too late.
A Suppressor can warn you of a danger you face, and can veto an action that’s being considered, to stop you from acting before it’s too late—for example, by telling you, “No, do not move in that direction! Or it could tell you to use a debugging technique.
A Censor works early enough to keep you from having that dangerous thought—so it never even occurs to you to put your finger into that flame. A Censor can work so effectively that you don’t even know that it’s working for you.
A Self-Controller recognizes that you have been failing to carry out a plan because, you instead of staying with it, you have kept on “changing your mind” about it.
Suppressors are safer than Correctors are, but both of them tend to slow you down, while you think of something else to do. However, Censors waste no time at all, because they deflect you from risky alternatives without interrupting your other thoughts, and therefore can actually speed you up. This could be one reason why some experts can do things so quickly: they don’t even think of the wrong things to do.
Student: How could a censor ward off a bad thought—unless it already knows what you’re likely to think? Isn’t there some sort of paradox there?
AI Programmer: No problem. Just design each Censor to be a learning machine that records which decisions have led to mistakes. Then when it next sees a similar choice, it just steers your thoughts in the other direction, so that you won’t make the same decision.
Student: Then wouldn’t that Censor still take some time to have enough effect on your mind? Besides, what if both choices were equally bad? Then that Censor must work even earlier, to keep you from getting into that bad situation in the first place.
AI Programmer: We could do that by giving each Censor enough memory to record several of the previous steps that led to such situation.
Student: Might not that cure be worse than its disease? If your Correctors could save you from every mistake, this might make you so conservative that you’d scarcely ever get new ideas.
Indeed, some experts have learned so many ways for any project to go wrong that, now, they find it hard to explore any new ideas at all.
