post by adamzerner
(Cross posted on my personal blog.)
Epistemic status: Exploratory. Trying to figure it out.
There is a feeling I often find myself feeling. It is an uncomfortable feeling, and the fact that I have trouble labeling it makes the discomfort even worse. I think I may have arrived at a solid label though: tentative anger.
Let me start by giving some examples of what I mean.
Yesterday was Thanksgiving, so let's talk about football. I am a Steelers fan. However, I'm not really a fan of their head coach, Mike Tomlin. Why? Because he does stupid things.
Here's one example. We always end up in these situations where linebackers are covering slot receivers. Linebackers are generally about 250lbs or so, whereas slot receivers are more like 200 pounds. Which means those receivers are almost always faster and more agile, making it really hard for linebackers to cover them. It's like asking a boxer to fight outside of his weight class. Usually you'd have ~200lb cornerbacks guarding those receivers instead.
No one, including Tomlin, actually thinks that it is a good strategy to put linebackers on receivers. It's just that... I'm not sure actually. For whatever reason, sometimes these mismatches happen. But Tomlin doesn't seem to mind very much. Compared to other coaches, he doesn't seem to try as hard to avoid them. And when they do happen, his response is something like:
Tough luck. The standard is the standard. You've gotta step up and meet the challenge. I don't care that he's 50 pounds lighter than you.
What makes more sense to me would be something like:
I got outcoached. My bad. I've got to put my players in a position to succeed, and I didn't do that. I'll hit the drawing board and try to prevent this from happening in the future.
But here's the thing: he's an ultra successful football coach, and I'm not. Maybe he knows better than me. Maybe there's a Bayesians vs Barbarians [LW · GW] angle to this.
That's plausible. But then why doesn't this happen to the other football coaches? And why do all the other fans get angry about this too? And why do journalists who seem to know football make the same points I make?
Ultimately, all things considered, I do suspect that Tomlin is being dumb here, and that I am justified in being angry. However, I'm only something like 80% sure. There is still a very reasonable chance that I am wrong and that my anger would be misplaced.
So what is the appropriate response? Hold off on feeling angry until I am more like 95% or 99% sure? Or am I justified in being angry right now?
Here's the deal with Paxlovoid. It's a drug for treating covid. They had clinical trials. The results of those trials were so good that they decided that it would be unethical to continue the studies, because people in the control group shouldn't be denied Paxlovoid. But, simultaneously, the FDA isn't ready to approve it yet.
Notice the hypocrasy? Zvi does [LW · GW]. So does Alex Tabarrok. And Scott Alexander. And Elizer Yudkowsky. And probably a bunch of other internet people who I trust.
It seems pretty fucked up. If the drug is so good that it is unethical to continue studying it, then shouldn't we let people take it? Shouldn't we at least leave that decision up to them?
On the other hand, I can think of reasons why it might make sense to act the way they are acting. Well, those bloggers I mentioned above addressed a lot of those reasons actually. But maybe there are things those bloggers missed. Maybe there are things those bloggers don't understand. Maybe they are being overconfident. Maybe the bloggers just don't have access to the same information the FDA does, and there is something the FDA knows that we don't that makes it a good idea to do what they're doing. Maybe we just are at some sort of inadequate equilibrium where no individual actor is doing something that would justify my being angry at them and instead it's just the situation and incentives that is to blame.
I don't know. It does seem pretty likely that individual people are blameworthy and that I should be angry at them. I just don't have a good enough understanding to know that to be true. I'm maybe like 90% confident I should be angry at someone. Vaguely, "the politicians and the FDA".
But like the case with Mike Tomlin, this isn't 100%. It's not even 99%. There's still a reasonable chance that my anger would be unjustified.
Here's what happened with Stallings (YouTube, news article). George Floyd was murdered by Derek Chauvin. Then there were protests in Minneapolis. Stallings showed up to those protests with a gun. Police were patrolling the streets, shooting rubber bullets at people in an attempt to enforce a curfew and get them to go home. Stallings got hit in the chest with one of those bullets and then shot back real bullets at the police. He didn't know it was the police. He was recently acquitted of various charges.
I have a lot of mixed feelings about this. One the one hand, what are you doing bringing a gun to a situation like that? On the other hand, aparently that's legal. On the other hand, so what? On the other hand, maybe there's a decent reason that I can't think of? Or maybe intent is what matters, and he had good intent? On the other hand, it seems unlikely that he had good intent. My money is on him having some sort of tough guy attitude. On the other hand, that isn't illegal. On the other hand, the legal system sucks and illegal is not a prerequisite for something being wrong. On the other hand, the topic here is whether it was right for them to acquit him. On the other hand, maybe that isn't actually the topic.
As you can see, I'm a little confused. And that is precisely the point. I have some sense that something unjust happened, so I want to be angry. But I'm not sure. And if something unjust did not happen, then I don't want [? · GW] to be angry.
At first I was going to call this "tentative indignation", but then I realized that it is more broad than that. There are situations where indignation isn't the right word. Like with Mike Tomlin. There I'm simply angry. I don't feel indignant. It's just football.
Anyway, what is the proper way to handle such situations? I think what happens in practice is that my anger just gets diluted. If someone does something that is an 8/10 on the badness scale, and I am only X% sure that they did it (or were wrong, or whatever) my anger gets brought down to something like a 6/10. But that feels uncomfortable to me. If they did it, I want to be 8/10 angry. If they didn't, I want to be 0/10. Either way, my 6/10 isn't "correct".
I guess we need to dive into what makes an emotion like anger "correct" in the first place. Well, it should be based in reality, for starters. Our understanding of reality is probablistic, so maybe this 6/10 anger thing is perfectly fine. Let's explore a little more.
Suppose hugs made me angry, I saw someone hug someone else, and I experienced anger. This feeling would be based in reality, but it still seems wrong. It seems wrong because, well, hugs shouldn't make me angry! Hugs generally just bring happiness to the world without having any harmful side-effects. They are a pareto improvement. So feelings should be based in reality... and... correspond with what is "good" for the world? We can dive further into the ethical philosophy another time. For now this seems to be roughly correct.
And with that said, my diluted anger does actually seem appropriate. Think about it: what are the alternatives? Going to the full blown 8/10 is the same response I'd have if I was 100% sure they did it, but here I'm only X% sure. Thinking about what is good for the world, from behind a veil of ignorance, if everyone went to full blown anger every time they were only X% sure of something, that just seems like it'd be too much anger. So my true level of anger should be less than an 8/10.
Let's explore it from the other side now. Should it be a 0/10? No. Asking the same question about what would be good for the world, if it took 100%, no, let's say 99% confidence before people got angry, then I think it wouldn't be enough. Anger is useful. It imposes social punishments on people who do bad things. It draws a spotlight on people who are suspicious. Stuff like that disincentivizes bad behavior, which we want.
So yeah, dilution is really seeming like it is in fact the correct response. Now let's revisit something I wrote earlier:
But that feels uncomfortable to me. If they did it, I want to be 8/10 angry. If they didn't, I want to be 0/10. Either way, my 6/10 isn't "correct".
Emotionally, I still feel this, even though logically I think dilution makes sense. Is that weird?
Maybe. Maybe not. Maybe I feel that way emotionally because of some sense of justice, where I don't want to wrongly point a finger at someone and chastise them, but I also don't want to pull my punches when someone truly is at fault. I suppose this sense of justice isn't such a bad thing though.
Comments sorted by top scores.
comment by localdeity ·
2021-11-28T11:10:21.241Z · LW(p) · GW(p)
My approach is to choose a strategy (i.e. what actions to take, given what I know and possibly in future situations where I learn more), based on an expected-value estimate, and let my emotions follow. Emotions that affect actions have a few use cases: (1) as a shortcut, (2) behaving "irrationally" for game theory purposes (e.g. gratitude, pride, loyalty, etc. making you trustworthy to others; anger and vengeance discouraging others from mistreating you), (3) motivating you to do something that you intellectually know is the right thing.
In the football coach case... Exactly what actions are you deciding between? If you imagine yourself as one of the team owners, then I imagine you'd want to consult more carefully with some experts, then confront the coach with some arguments (possibly bringing a trusted expert along), demand he explain himself, and see how satisfied you are with his explanation and planned future actions, and decide based on that (plus other considerations) whether you want to do something like get a new coach.
But if you imagine yourself as you... what are you going to do? Email the owners? Email a sports journalist to try to get them to research the issue and make a stronger case in public? Complain to other fans at parties? (Just for fun, or because you hope the idea will spread and eventually bubble up to the owners?) I suspect pondering the actions you might take would help you clarify your emotions: maybe "Eh, it annoys and disappoints me, and I will note it as a likely instance of high-level incompetence and will mention it in relevant discussions, but I'm not bothered enough to do more than that"; or "Yes, I am seriously offended by the idea that my team is being run by a bozo, and I will start contacting people about it, and maybe do some more research to strengthen my case."
I guess the inchoate anger you describe may be taken as a signal that you need to think out the situation and your strategic options.
(Incidentally, I think that, for many news items, the chain of questions "What are you going to do about it?" → "What emotions does it make sense to have about it?" → "What is the point of your even knowing about it?" is a good one to follow, with a conclusion many news purveyors do not want you to reach. (It is possible for random news items to be useful in unexpected ways, but, well, should it bother you that you haven't read the majority of the 13 million articles from the NYT alone? Of course not. So you should only bother collecting news items that seem likely to be useful, or if you have some purpose in mind.))
For the FDA case: For the record, my position is that the FDA appears to be a bunch of bureaucratic jerks who would rather delay a lifesaving drug and watch 10,000 people die, than approve a bad drug and watch 10 people die; that they've been like this for decades, and that this is presumably due to incentives they are for some reason faced with. Historically this hasn't ranked very highly among my list of complaints about destructive U.S. government interventions, but now it's more relevant and important and obvious, and I have some hopes that the issue may get addressed.
Now, in terms of actions? I'm not a politician, nor do I know any, nor do I think they'd listen to me; nor do I have a communications platform better than arguing on internet forums; among people I know, generally they either agree with me already or seem unlikely to respond well to my injecting my politics into a conversation. So I think what I'll do is make the point about this most obvious failure of the FDA when COVID policy comes up in conversation, and otherwise just remember it. (If I were in charge of a political organization, then I guess I could do something.)
For the Stallings case: Ok, so, if you personally had any prospects of running into the guy, then you might or might not decide he has bad judgment. The legal case itself is decided, unless you think you can influence an appeal. What is the point in having emotions about it, or even in knowing about it? What actions does it help you with? You could use it to judge the people who decide to talk about it—although if you yourself are confused, then what conclusions could you draw from the positions others take? (I guess it's possible to say "This argument contradicts itself/makes non-sequiturs and appeals to emotion, so I can note that its author is untrustworthy" independently of the underlying issue, but you could do that without even knowing the facts. Knowing the facts helps only with detecting people who make well-formed arguments from false premises.)
To me, the main application is the legal-theory and rules-of-engagement angle—essentially the exercise of figuring out for yourself what the rules should be, and then also judging our current legal system and society based on how they agree/disagree with you, and seeing if there's anything you want to do about any difference you find.
Object-level look at Stallings: If someone is being aggressively shot at, and gets hit, is it reasonable to expect him to stop and take the time to think "Hmm, I think that hit less hard than a bullet is supposed to hit—maybe these are rubber bullets", before deciding to self-defensively shoot back? I'm not sure if people think that's supposed to be a reasonable expectation, but one thing the article mentions is that it's an official policy that rubber bullets should be fired from orange-barrelled guns to mark their non-lethality, and that the officers violated this policy. I think I'm inclined toward the position "If you want to aggressively shoot at people with nonlethal bullets and expect (in a legal sense) them to not shoot back with real bullets, then the burden is on you to do everything to signal that your bullets are nonlethal; if you miss one of the signals that a reasonable person is supposed to check for, then they are within their rights to shoot back and kill you."
It seems there might be another position: "If the police are shooting at you, you're never justified in shooting back in self-defense, no matter how aggressive and unjustified their actions might be, because they're police." I dislike this position—I prefer the position that the police have no special rights or immunity that ordinary citizens lack. Nevertheless, even if we accepted it: since you do have the right to self-defense against non-police, it follows that there must be rules about whether the citizen can be reasonably expected to know that those shooting at him are police. I would be inclined toward a similar position: "If the police want to aggressively shoot at people and expect (in a legal sense) them to not shoot back, then the burden is on the police to do everything to signal that they are in fact police; if they miss one of the reasonable-person signals, then they can't complain about being shot." According to the article, the police wore SWAT uniforms but came from an unmarked van—I don't know if anyone has a list of reasonable-person "We're police" signals, but I'd accept that one—and Stallings says he in fact didn't know they were police, and dropped his gun and lay down when he realized.
I suspect that at least one of these legal-theory angles would have been discussed in the trial, if I checked it. ... Hmm...
Reading the document at the bottom (it's labeled "body camera transcript", but is actually a judge's order), going through the judge's "findings of fact", the following things strike me as important: (a) the officers were inside the van when they were shooting and when Stallings shot back; (b) it was dark (past 10 pm), and it says "no interior lights are activated, except for those on the dashboard"; (c) I'll just quote this whole section: "None of the officers in Unit 1281 announce their presence as law enforcement nor did the van have its lights or sirens activated. Neither Officer Stetson nor Officer Cushenbery wait to determine whether Mr. Stallings posed a threat before shooting at him. Officers Stetson and Cushenbery testified it was dark and Mr. Stallings was approximately 70 feet away when they shoot from the moving van; that testimony is credible and is supported by the video evidence"; (d) "The officers immediately yell “shots fired, shots fired” and rush out of the van. Mr. Stallings says it was only then he realized the occupants were law enforcement. This is credible since he is seen immediately placing his rifle on the ground, away from his body[...]".
The document has some irritating stuff where it seems there is indeed a law providing special protection to police officers, and that the application of that law doesn't require that the defendant know that the officer was an officer (but that different jurisdictions have different rules)... But it also says this, in its explanation of why the judge is allowing Stallings to claim self-defense:
Minnesota law authorizes the use of reasonable force toward another when a person reasonably believes they are resisting an offense against that person. The Minnesota Supreme Court has recognized defendants have the right to resist an “unjustified bodily attack” from law enforcement. This has since been referred to as self-defense. Deadly force, however, is not permissible against police officers “who have announced their presence and are performing official duties." [...] It would seem, therefore, defendants may use deadly force in self-defense against an unjustified bodily attack from law enforcement who have failed to announce their presence."
Here, Unit 1281 failed to announce its presence. Indeed, it appears the team deliberately wanted to not be identified in advance as law enforcement. The unmarked van had no distinctive exterior (law enforcement) lights or signage. The van did not use its siren. The interior lights, except the dashboard, were off. And the officers wore all-black uniforms at night. Additionally, against MPD policy, all officers in Unit 1281 were equipped with black-barreled 40mm launchers. The white, unmarked conversion van was followed by marked squad cars flanking its rear right and left sides. However, the flanking squad cars were not immediately behind Unit 1281’s unmarked van. Finally, Unit 1281 never verbally warned or announced their presence before Officers Stetson and Cushenbery fired their 40mm launchers at Mr. Stallings. Because Unit 1281 did not announce its presence, Mr. Stallings may claim self-defense if he satisfies the elements for the defense[...]
Ok, this seems like a relatively reasonable rule—not exactly the position I stated above, but acceptable—and in this case leads to a good outcome.
The article also says the police kicked and punched Stallings when he was on the ground, then lied that he resisted arrest and ran away (body camera footage proving the lie). The judge's writeup confirms this. ... I think any officer who does that should be charged with a crime and/or fired. Article says "As of Thursday, there are no updates on whether any discipline will be handed down to officers involved in this incident."
By the way, it's often the case that, if P is true, then actions A+B+C are best; and if P is false, then actions D+E+F are best; but if P seems to have e.g. a 50% chance of being true, the best strategy is neither A+B+C nor D+E+F, but rather A+C+F+G. It seems like this gets lost in some mainstream discussion.
For example, if someone is suspected, possibly accused, of some kind of misconduct, but strong evidence has not been put forth and an investigation won't yield results for a few weeks, then perhaps your estimate of his guilt is somewhere in the 20-80% range. In that case, punishment is probably inappropriate (there is a common legal principle that "It is better that ten guilty persons escape than that one innocent suffer", which may be interpreted to mean "Punishing an innocent is at least 10x worse than letting a criminal escape, and therefore you shouldn't punish if your guiltiness likelihood estimate is less than 90%"), but doing nothing may be suboptimal. There are usually precautions and remedies you could take in the meantime, where the "benefit if you're right" is much higher than the "cost if you're wrong".
If the misconduct is some kind of harassment or other mistreatment of those around the suspect, then an obvious precaution is "make him take a vacation for a few weeks while the investigation proceeds" (i.e. administrative leave), while remedies would include offering time off and free counseling and other support to those who claim to have been mistreated; the costs are fixed and low, and the benefits if guilty seem high. If it's academic fraud, then I imagine worthwhile precautions include bringing in a competent third party to double-check his work on any current and not-yet-published studies (delaying their publication if necessary), and remedies include requesting replication of past studies whose results anyone cared about (this should happen more often in any case).
There's been rhetoric around issues of this kind that makes me think the discussants can't effectively work with any probability estimates other than 0% and 100%; either the alleged perpetrators should be locked up, or the alleged victims should be ridiculed, and everyone needs to make up their minds right now; a notion like "The probability is high enough that we should offer comfort to the alleged victims, but low enough that we shouldn't end the alleged perpetrators' careers, and should wait for more evidence from a serious investigation" seems outside the conversation. Duncan's Split and Commit [LW · GW] post made a related point.
comment by cata ·
2021-11-27T21:40:23.343Z · LW(p) · GW(p)
One thing which I see often which is present in the Paxlovid example is a case that goes like this:
- Someone does something which obviously looks bad to a naive onlooker, but might be justifiable for some reason I don't know.
- Although they could easily attempt to explain why they did it, they make no such explanation.
- Even when people complain, no explanation that makes sense is ever given.
In these cases, if my neighbor or my coworker was doing this, I would reason: Hey, if this was justifiable, this guy could gain a lot of goodwill with me trivially, and he's not doing so. Therefore it's probably not. As a result, I basically notch the probability of "this guy did a bad thing" up to 100%, and I think it is right to do so.
It's tempting to extend this intuition to "the FDA" or "my congressperson" or "a celebrity". But I think it's frequently epistemically questionable to do so. In these cases, there are a lot of reasons that could get in the way of the reasoning:
- Maybe the actor is different enough from me and the complainers that they don't understand that their action looks bad.
- Maybe the actor never heard the complaints, so they didn't know anyone noticed the problem.
- Maybe the "reward" of goodwill is very small compared to other concerns, like drawing more attention to the action.
- Maybe the actor is made of multiple people and there is a communications breakdown between "the person who made the decision" and "the person who can communicate justifications."
In the Paxlovid example, I think several of those reasons apply. The FDA probably thinks it doesn't look bad, because they have a reference class in mind of even slower approvals, and an inside view that highlights all the important things that need to get done in between now and approval. The reward from satisfying a lot of people like us by providing explanations is probably low -- most people don't seem to really notice that there is a problem or that anything could be better. The communications breakdown is probably real, too. So even if there were a pretty good explanation for the delay, I am not confident we would hear it.
Replies from: adamzerner
↑ comment by adamzerner ·
2021-11-27T22:08:59.701Z · LW(p) · GW(p)
That sounds like solid reasoning for the Paxlovid case. However, for the more personal situations, notching the probability up to ~100% doesn't seem right to me. I get the sense that it isn't too implausible for the actor to have good reason to "keep quiet". Or, rather, for the actor to think that they have good reason to keep quiet.
An example of this that comes to mind is the last time I visited my grandparents. They are 90+ years old and should not be driving. Plus, I am particularly [LW · GW] crazy [LW · GW] about the dangers of driving. Before the trip we agreed that we would all hang out at home and avoid driving. But then during the trip, they wanted to go out. I would say no. They would ask why. I would say I don't want to talk about it. They would say I'm being unreasonable by restricting us and not giving a reason why. But if you think a few moves ahead, what happens when I give my explanation? I don't want them to know I have this anxiety about death (and I don't want to impose that anxiety on them given that they are close to death, although I think that would be unlikely). And I don't want to start an argument about how capable they are of driving. We've tried that in the past, and it always balloons into a bigger argument. So I sense that the "keep quiet" route is best for all parties.
Replies from: cata
↑ comment by cata ·
2021-11-27T22:59:26.337Z · LW(p) · GW(p)
I like your example. Perhaps I am a little too inclined to update, either due to typical-minding (I am usually very frank and expect others to be) or due to a kind of misplaced irritation (I want others to justify themselves to me, so if they don't I ascribe bad things to them.)
comment by Viliam ·
2021-11-27T21:05:52.303Z · LW(p) · GW(p)
Yes, this is a difficult situation. If there is a 50% chance someone did a bad thing, from certain perspective feeling 50% angry sounds like the best you can do.
From another perspective, either the person did the bad thing and then you just forgave them 50% for no reason, or the person didn't do the thing and then you are 50% angry at an innocent person, wrong outcome in both cases. (Imagine a judge saying: "so, with probability 50% this person is a serial killer, and with probability 50% they are completely innocent... therefore, 10 years in prison sound quite fair to me".)
Intellectually, we should distinguish between "a certainly of minor crime" and "a small probability of a big crime". What are the proper emotional equivalents, I don't know. Perhaps there is no good solution, because our emotions were not designed to be fair.Replies from: adamzerner