Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Model, Care, Execution, published by Ricki Heicklen on June 26, 2023 on LessWrong.

When Alice harms Bob, it is likely that one of the following three things went wrong:

Model: Alice’s mental picture of what Bob would want, or of what is going on in the situation (which we’ll call her model), was incorrect and led her to behave in ways that caused Bob harm. If she’d had a better model, she would have acted differently, resulting in a better outcome for Bob.

Care prioritization: Alice prioritized something else—herself, another person, some abstract value—over Bob. Even if she cares a lot about Bob’s happiness, other factors outweighed her care for him in this situation.

Execution: Alice knew what Bob would want and cared enough about him, but failed to execute on bringing about the best Bob-outcome. Her execution failure could be the result of physical limitations, impulsivity, depression, structural barriers, or any of a number of other constraints.

Often, the problem is a combination of two or even all three of these: if I had taken the time to think a little about what you would want (a care problem), I would not have made the mistake I made (a model problem) or would have made a plan that actually worked (an execution problem).

We have found the framework of Model, Care, and Execution (MCE) to be useful when postmorteming cases of interpersonal harm. What went wrong, and how can we do better next time? More generally, what failure modes are most common for each of us, and what do we need to do to help each other improve in those areas?

Postmortem: Your Roommate Sleeps Through His Interview

It’s 1pm and your roommate Choni has fallen asleep on the couch. You see him sleeping and leave him alone. When he wakes up at 4:30pm, he has missed his 2pm job interview and is annoyed. What went wrong? Here are three plausible stories:

There was a modeling error: You falsely believed that Choni wanted to nap all afternoon—you didn’t know he had an important interview, or thought he had rescheduled it—so you decided to let him sleep, acting reasonably based on your bad model.

There was a care mismatch: You knew Choni would want to be awakened, but decided against doing so, perhaps because Choni has a tendency to punch people in the face when they wake him up, or because you were helping a friend in crisis and decided that was more urgent, or because you thought there was some chance he actually wouldn’t want to be woken up and you didn’t want it to be your fault if he was tired later.

There was an execution failure: You knew Choni wanted to wake up, you wanted to help him wake him up, and you nevertheless failed to, perhaps because he slept through your many attempts, or because you came back to wake him for the interview at 1:55pm but had locked yourself out of the apartment by mistake, or because you were too depressed to get out of your bed and knock on his door.

Depending on which of these went wrong, your plan for the future should be different.

If it was a model problem, then Choni saying ahead of time “I have an important interview at 2pm today that I can’t miss” would have helped. You could also establish a default decision together—e.g., any time Choni is asleep in the middle of the day, you should wake him—or communicate other relevant information, like “it’s extremely easy for Choni to fall back asleep when awakened undesirably, so it’s better to err on the side of waking him” that will help you get a better idea of what he would want next time.

If it was due to care prioritization, then Choni has a few options. He might try to change his own behavior so that it’s more rewarding to do what he wants, e.g. by making waking him up more pleasant (and less punchy), or by profusely thanking you afterwards for doing so, or by offering you a cash bounty for every successful ...