Effective criticism
Responding to Scott Alexander’s excellent post, Criticism of Criticism of Criticism.
Giving effective criticism is hard. There’s always the danger that people will get offended or reject the criticism, of course. But there’s also the subtler failure mode, in which people will accept the criticism and then not do anything about it.
Some of my friends, for instance, have recently been complaining to me about how they’re unhappy with their weight. Now, if these were just acquaintances, I’d smile and nod and then change the subject, because agreeing with someone they have a weight problem can be a recipe for an awkward silence. But, given that these are my friends, I like to be honest with them, and say, “Yeah, I understand. You’d probably feel better if you lost some weight. I feel that way too.”
Phrased that way, my friends usually agree with me, and we have a brief bonding moment where we agree that we should both lose some weight. However, if I then try to take it a step further, and suggest specific ways for them to lose weight, suddenly I get a lot of disagreement. For some reason, we can agree on the general problem, but the specifics of why they’ve gained weight and how they could lose it seem to be impossible to agree on.
I have one friend, for instance, who's 15-20 pounds overweight because of his love of good drinks and good food. He’ll readily admit that he’s overweight (and even bemoan the fact), and frequently comes up with exercise plans for how to lose the weight. When I suggest that he cut out some of the fast food and sugary cocktails, though, he’ll outright disagree with me. He loves his unhealthy habits too much, and pointing to them as a source of weight gain is too painful.
Meanwhile, I have another friend who’s much more overweight. It causes him serious health issues. Pretty much everyone in his life has pointed out that he’s gained way too much weight in the past decade or so, and he’ll say it, too. His concern about his weight has led him to try pretty much every weight loss strategy you’ve heard of, and some you haven’t.
His biggest issue is that his work schedule is absolutely insane, causing him to work very late and get up early. This causes all sorts of knock-on effects in his eating and exercise habits. He’ll never change jobs, though, and I’ve given up trying to suggest alternatives to him.
When I read other people have trouble with effective criticism, I wonder how similar their situations are to mine. For instance, Scott Alexander recently wrote a post on how he dislikes most criticism of the effective altruism (EA) movement, as it never seems to have any effect, despite the fact that the EA movement is extraordinarily open to hearing criticism. Scott claims that the criticism that does seem to be both more effective and more annoying to EA movement leaders isn’t the broad, sweeping kind, but the nitpicky kind, in which people ask for clarification on specific facts or claims.
While in the post, Scott divides broad and nitpicky criticisms into paradigmatic vs. specific criticisms, I can’t help but wonder if his situation is more similar to my situation with my friends. In other words, if the issue isn’t paradigmatic vs. specific critiques, but non actionable vs. actionable critiques. Or, to be even more precise, critiques that seem actionable but aren’t.
If my friends tell me, “Man, I really feel like I should lose some weight,” and I agree, it might seem like that’s an actionable critique: they should take the action of losing weight. But, it’s easier and more comfortable for them to take it as a non actionable critique, something like, “It’d be best if you lost weight, but don’t trouble yourself about it right now”.
Meanwhile, if I say, “You shouldn’t order the margarita,” or “you should apply to this 9-5 job,” it’s impossible to take that as anything other than an actionable critique. There’s no way they can take it as, “It’d be best if you didn’t order the margarita” or “In an ideal situation, you’d apply to this alternative job”. Either they order the margarita or they don’t in the next 30 seconds when the waitress comes over.
Philosophically, this comes to a question of definitions in a pragmatic sense. The great (and racist) American philosopher Charles Sanders Peirce said that two definitions were different if they had different effects. My friend calling himself overweight is different from him calling himself in good shape only if he changes his actions when he calls himself one or the other. If I call my friend overweight and expect him to take all practical efforts to lose weight, but he does not expect to take those efforts, then he and I are working off of different definitions of overweight.
When it comes to organizations like EA (or psychiatry, in Scott’s other example), this difference in definitions is a serious problem. Two people can both agree that EA is ineffective or psychiatry is racist, but mean different things in terms of what they expect the consequences of that definition will be. Then it becomes a coordination problem, and nothing gets done.
I think the solution to this is ultimately to start with the general definition of what people agree on, and then outline specific consequences of that definition. So, it might be something like “Given that we both agree EA is ineffective, I expect to find the following results in the next report. Following that, we should take these specific actions to rectify those problems.” Either people will agree with the consequences (in which case they just need to do the thing), or they won’t, in which case there’s more room for discussion.