文章探讨了冲突解决为何成为组织者和社区管理者面临的难题。作者指出，对冲突解决的兴趣具有工具性趋同性，即无论是能提供帮助的人还是别有用心的人，都有强烈表达对冲突处理方式的看法的理由。文章通过银行安保和哈利波特中的例子，分析了不同类型的人在冲突中的表现，以及为什么恶意行为者会试图破坏解决冲突的系统。文章认为，即使存在完美的系统，恶意行为者也会试图破坏它，因此在处理冲突时需要保持警惕。

🕵️‍♂️ **冲突解决的工具性趋同性：** 无论是有益还是有害的人，都有表达对冲突处理方式的强烈意见的理由，使得冲突解决变得复杂。

👤 **不同类型的人在冲突中的表现：** 文章将人分为几类，包括专业人士、好奇者、怪人、“正常”人以及恶意行为者。恶意行为者会试图破坏解决冲突的系统。

🛡️ **恶意行为者的策略：** 恶意行为者会质疑系统的公正性，即使系统是完善的，他们也会试图破坏它，因此在处理冲突时需要保持警惕。

🏦 **银行安保案例的启示：** 通过银行安保的案例，说明了当有人对安全系统表现出异常兴趣时，需要保持警惕，因为他们可能试图利用系统漏洞。

Published on May 9, 2025 2:16 AM GMT

Why is conflict resolution hard?

I talk to a lot of event organizers and community managers. Handling conflict is consistently one of the things they find the most stressful, difficult, or time consuming. Why is that?

Or, to ask a related question: Why is the ACX Meetups Czar, tasked with collecting and dispersing best practices for meetups, spending so much time writing about social conflict? This essay is not the whole answer, but it is one part of why this is a hard problem.

Short answer: Because interest in conflict resolution is instrumentally convergent. Both helpful and unhelpful people have reason to express strong opinions on how conflict is handled.

Please take as a given that there exists (as a minimum) one bad actor with an interest in showing up to (as a minimum) one in-person group. 

I.

See, a funny thing about risk teams: the list of red flags you have? On it, one of them is “Customer evinces an odd level of interest or knowledge in the operations of internal bank policies.”

-Patrick McKenzie

Imagine you are a bank security guard. 

You are standing in your bank, sipping your coffee and watching people come in and out. It's a good gig, bank security, they have dental insurance and the coffee machine makes a surprisingly good cup. As you're standing there, someone in sunglasses and a baseball cap waiting in line strikes up a conversation with you.

"Nice day, isn't it?" they say.

"Sure is," you reply.

"Nice bank too," they say, "I love the architecture on these old buildings. Good floors. Do you know if it has a basement level, maybe with a vault?"

"Yeah, concrete flooring."

"Nice, nice," this stranger (who you start thinking of as Sunglasses) says, "You know, I'm also into videography. What kind of cameras do you have around these parts? Like, about how many, and covering what angles?" You notice Sunglasses has a notepad out, pen held expectantly.

". . . you know, I'm not sure I should tell you that," you say slowly. "It's not like it's a secret exactly, the cameras or at least their bubbles are pretty visible, but I don't think idle curiosity is a good reason to tell strangers how the bank security system works."

"Okay, I admit it's not just curiosity," Sunglasses says with a charming smile. "What if I'm just concerned whether my money is going to be safe in this bank? Isn't it reasonable to want to understand how it's kept safe from bank robbers, and how the teller will figure out if I'm actually me when I come to withdraw money again?"

"That would be reasonable," you answer, "and lots of people might be interested in knowing their money is safe and they'll be able to get it back. Some of that information is public, we have a newsletter about it."

"But not all of it's public," Sunglasses points out. "Every customer should care that their bank is secure. It seems like you're new at this whole bank security thing. Look, I'm willing to help you out, make some suggestions about vault locks and camera angles, maybe recommend a good security firm. Looks like you're using TS-53 cameras? Those were fine for ten years ago, but these days networking TLAs are faster and someone could technobabble their tachyons to break in." 

You stare at Sunglasses. "I admit I'm new at security. It would be nice to make the bank more secure, and you're right that customers have a legitimate interest in the bank's money being well defended. What you said sounds about the TS-53 sounds right at first pass, and you seem very confident. But I am also getting increasingly suspicious about your interests here."

Sunglasses shakes their head disarmingly. "I solemnly swear I have a lot of experience with bank security systems, and I think there may be a weakness in the anti-bank robber measures you have here. I've seen a lot of good banks get robbed, and that's why I have such strong opinions on how bank security should work. Just let me tell you how to arrange the cameras, what vault lock to install, and how to set up the night guard patrols."

"No," you say. "While some people do have a professional skillset around bank security, not everyone with that skillset is automatically on my team. I would not do better at keeping the customer's money safe if I accepted help from the people most insistent on giving me help. I'm going to ask you to leave now." 

"Fine, be that way," Sunglasses says. Then they cup their hands and yell to the other customers, "Hey everyone, this bank guard is throwing me out even though I haven't done anything! They're probably racist! Ya'll should get another bank guard!"

II.

"If you once tell a lie, the truth is ever after your enemy; and there's a lot of people out there telling lies -" Harry's voice stopped.

"What does that have to do with Fawkes?" she said.

Harry withdrew his spoon from his cereal, and pointed in the direction of the Head Table. "The Headmaster has a phoenix, right? And he's Chief Warlock of the Wizengamot? So he's got political opponents, like Lucius. Now, d'you think that opposition is going to just roll over and surrender, because Dumbledore has a phoenix and they don't? Do you think they'll admit that Fawkes is even evidence that Dumbledore's a good person? Of course not. They've got to invent something to say that makes Fawkes... not important. Like, phoenixes only follow people who charge straight at anyone they think is evil, so having a phoenix just means you're an idiot or a dangerous fanatic."

-Harry Potter and the Methods of Rationality, Eliezer Yudkowsky

Let's be reductive and say there are five kinds of people in the world.

The Professionals: People with a legitimate, professional or semi-professional interest in social conflict. Divorce lawyers, social workers, therapists, the chair of the sci-fi convention complaint department.The Curious: People with a noticeable curiosity or special interest in social conflict. These people took up reading Difficult Conversations or Non-Violent Communication or the like the way other folks decided to study Spanish opera, or U.S. Navy ships, or knitting.The Oddballs: People with weird, outlier behaviors that trip a lot of false positives and as a result have lot of experience with conflict resolution systems. Autistics and other neuroatypicals, ethnic and religious minorities, a fair number of homeschoolers."Normal" people who don't really care about social conflict as long as it's not bothering them, and it usually doesn't.Bad actors: People who cause problems and try to get away with it, which here includes people who just legitimately want to pursue their hobby of punching everyone who disagrees with them in the face.^[1]

(These categories absolutely overlap and intersect some of the time. Some autistics set off a bunch of yellow or even red flags, then developed a special interest in human social norms and find them interesting. Some therapists are abusive bad actors, leveraging the privacy and power of their position to do harm. This list of overlaps is not exhaustive.)

Now let's say that the bad actor is not an idiot. They have considered what and who might stop them, and what they might do about it. 

It is an obvious, straightforward move to accuse whatever system is responsible for catching them of being corrupt and the people running that system to be horrible or incompetent. 

Yes, there are other reasons that someone might say the system is flawed. Yes, sometimes the people in charge of it make mistakes. Sometimes, yeah, it is actually the case that there's a glaring problem with the way hypothetical bad actors are identified and treated. The KGB and Soviet Russia is a famous historical example, but there are many more and many smaller examples of misrun HR departments and convention safety chairs in over their heads. No, I do not think I or humanity at large have found the One True Way to correctly handle complaints and conflict. Yes, I want to improve the setups around me and to get more skilled at handling things like this. Yes, sometimes I think it is correct to try and dismantle the thing and put something better in its place.

But. 

Even if you had a system that was perfectly accurate, universally applicable and flexible, whose agents were unfailingly correct in how they carried out its orders, you would have some portion of people who have an obvious motive to say the system is broken and the agents are horrible. If the bad actors had a little forethought, they wouldn't say "the system is horrible, they won't let me punch people in the face." They'd say things like "the system is horrible, it thinks that innocent person punched someone in the face even though they didn't."

And when you don't have a perfect system, but only a decent system with reasonable people in the position of its agents and where it didn't quite match the local social norm but was making an honest effort, then you will wind up with a lot of things an antagonist can point at to argue that nobody should trust it.

Yes, due process and rights for the imprisoned. Yes, the system also has an incentive to smear and put away anyone who threatens to rebel. And yes, we can always try to do better. But maybe be a little suspicious when the person in sunglasses, already being dragged into the cop car, complains that the cop is just being racist and decries the legitimacy of the justice system?

(Though also remember I mostly deal with the kind of complaints you get about ACX meetups. It's considerably less dramatic than that sentence might sound.)

III.

Firewalls [CBR03], packet filters, intrusion detection systems, and the like often have difficulty distinguishing between packets that have malicious intent and those that are merely unusual. The problem is that making such determinations is hard.  To solve this problem, we define a security flag, known as the "evil" bit, in the IPv4   [RFC791] header. Benign packets have this bit set to 0; those that are used for an attack will have the bit set to 1.

-RFC 3514, "The Evil Bit"

When you are trying to set up your disciplinary process, justice system, network security permissions, or other system by which you will identify and handle bad actors, you should be aware that some of the people who appear to be trying to help you might have ulterior motives. 

If you do not have some reason to expect that you are already good at this — if you're one of the normal people in the bullet points above, who just wants to have a nice society or meetup group and is wondering why we can't just do something simple and reasonable — then the bad actors probably have more experience with this than you do. Consider that you may have never interacted with a complaint department at all, while they may have been through the ban committee process from multiple different groups. 

(For that matter, being the person in charge of banning others is a position with obvious appeal to someone who suspects they may come to the attention of The System sooner or later. If there's a ninety eight normal people, one honest professional, one bad actor, and you have no way to distinguish, then you may prefer choosing your overseer via random lot rather than taking a 50/50 chance between your bad actor and your honest professional, even though the honest professional is really really useful if you have one.)

In Pareto Best and the Curse of Doom, I talked about how finding people with the overlaps of multiple skills is hard. To use the example of a community organizer, there's selection pressure to have one who is a good marketer, good with handling logistics like the venue and food, and charismatic in person. 

Over a long enough run and a large enough community, there's eventually some pressure for them to be good at conflict resolution, but a group can get surprisingly big and last surprisingly long before this becomes important — and if they're bad at it, or just normal amounts of competent, there are many ways for a group to keep growing despite constant arguments until the organizer steps away and even after.

No other part of organizing has this problem. If you don't know what activities to run, you can ask, and people will tell you what they like. If you don't know how to advertise the event, you can ask, and people might have helpful suggestions. If you don't know how to book a venue, you can basically just ask, and it's pretty unlikely anyone has a motive to sabotage your venue selection. Maybe they own the venue and they're trying to sell it to you, but that's a bit more straightforward. Not so with conflict resolution.

IV.

"If I could predict exactly where Stockfish 15 would move, I could defeat Magnus Carlsen just by making the moves I'd predict Stockfish would make. Maybe if your moves have sufficiently straightforward vulnerabilities there'll be an obvious way to exploit those, as seen by a human grandmaster; but Stockfish can search more moves than any human can, better than we can, and it might find an even better way to defeat you."

. . .

"How about if I move my rook over here?" the kid says, a few moves into the game. "Then the AI will try to take it with its queen, and I'll come in and grab the queen. How will the AI win after I've got its queen?" 

"Okay, now you're failing at putting yourself in the AI's shoes and asking how to win from its position, even using your own intelligence," you tell the kid.

-Eliezer Yudkowsky

I don't have a solution to this. 

I keep encountering people with very strong opinions on the correct way to handle complaints and conflict. I don't have an omniscient view of who is good at it, who is right and who is wrong. But uh. I notice that something like half of the people who have expressed very strong opinions on this to me, it turns out there's a bunch of complaints about them and if I used the system or rules they're advocating for they'd be in the clear^[2]. 

(Which makes sense! If I heard lots of the people dragged away in the night by the KGB had strong opinions on how great jury trials were and that they'd have been cleared by a jury trial, that doesn't surprise me. And yet I also wouldn't be surprised to hear lots of the losing defendants of a healthy jury trial system have strong opinions on how the judge and the cops and the whole system are out to get them.)

If you are a good and virtuous person, you maybe are interested in how conflict resolution is done and having a part in it. If you're a harmless nonconformist, you're a bit more likely interested in how conflict resolution is done and having a part in it. If you are a nefarious person who wants to rob banks or punch faces, you have an obvious interest in how conflict resolution is done and having a position of trust or authority in it. 

If you just want the thing to work and not be a big deal, you should be at least somewhat suspicious of the people offering to help. Not a lot suspicious! Most people are basically well meaning, I'm not advocating pervasive paranoia here. Maybe less suspicious, if you have a firmer explanation for why they know this information and why they're interested, but remember that bad actors can lie or mislead about why they're interested.

And this generalizes all the way upstream of the conflict. If you have some part of the system that doesn't make or carry out the decisions, it's just the part that's supposed to investigate and report the truth of what happened, obviously that's a super useful part of the system to get control of. If there's a verification setup or a vote counting role in deciding who is supposed to investigate and report the truth then the vote counting role is a super useful part of the system to get control of, or if it can't be controlled than discredited.

Thus the answer. Why is the ACX Meetups Czar, tasked with collecting and dispersing best practices for meetups, spending so much time writing about social conflict? Why is this the topic that creates so much stress for so many otherwise skilled organizers?

Because this is the topic that is adversarial, not just during an incident, but in every step leading up to it. If you take everyone's advice on how to build your bank security system, you may well be doomed before the alarm sounds — if it ever does.

(Okay, but why should you trust me? Professional interest since complaint handling is part of my role, but good question and don't be satisfied by that. CONSTANT VIGILANCE.)

1. ^{^}

   There's a tangent I plan to talk about in a future post here, but I tend to use examples which are obviously bad to do and I expect everyone to agree are bad to do. These examples tend to be unusually bad, because I'm trying to meet that standard. I could have put "will imply everyone who disagrees with them is stupid" or "will awkwardly hit on every woman attendee" instead of the face punching thing. I could have put "will get drunk and stand so close to others people can smell the alcohol on their breath" or "will loudly bring up how Chairman Mao was a great leader every single meetup, even if the event is about ice skating."

   There is an issue of distinguishing what side of a line an edge case falls on, or how hard to come down on something that's kinda bad but not seriously bad, or how to carve out spaces where a thing that's bad in most places is accepted here. It's an important issue. I'm ignoring it in this essay.

2. ^{^}

   Or at least more in the clear. One in a while someone will advocate for rules they're pretty plainly breaking, but they tend to assert some interpretation where what they're doing is fine actually.

Discuss