≡
✏️ Blog 🌍 Travel ✅ Subscribe 👍 Donate
🔍
👤
18 july 2023
You Can't Get Moderation Right
But without it, your community will die. Guaranteed.
This post is available in other languages 👉
Русский

90% of any online community's success comes down to moderation. Not the platform, not the “algorithms,” not even the content. Moderation.

The problem is, techies are all convinced it's the other way around. They're forever looking for some kind of “engine,” building fediverses and social graphs on the blockchain, while some little Telegram group or email newsletter leaves their community in the dust.

I've been building communities in one form or another (not always by choice) for the past ten years or so, and I've wanted to write a post about the different kinds of moderation and how our approaches to it evolve for a while now.

I remember how surprised everyone was that the code for Vastrik.Club was up on GitHub from day one. “Wait, so anyone can just steal it, launch their own clone and rake in the cash?” Haha, you sweet summer kids. In 3 years we've had about ten forks, some by people way more “popular” than me, but not one of those forks ever got more popular than the person who made it.

See, content is always a product of moderation. Look at far-right Gab, full of American rednecks, then look at some random wokey-wokey Mastodon instance where you can get banned for using non-inclusive pronouns like “he/she.” Same technology (almost), completely different content.

People don't start writing long-form posts and sharing what they've learned just because you gave them yet another text box on the internet.

Besides, communities aren't about content. They're about people.

So, content another time. Today we're talking about how to moderate, how to ban insufferable bores without mercy, and why writing clear rules doesn't work in real communities.



It always surprises me how many people are convinced communities have to be run on the same principles as countries. Hate to disappoint you, but that's complete nonsense, for one simple reason:

👉 You don't get to choose what country you're born in. Joining a community is always a choice.

If you don't like a community or its rules, you just close the browser tab and go find one you like. Nobody says a word. Nobody gives a fuck. You don't give a fuck either. But if you decide you'd like to switch countries — YOU'RE FUCKED)0)))).

That's why these two systems — compulsory vs. voluntary — are so different and so obviously need different ways of running them. Nobody's stopping you from playing democratic institutions in a ten-person group chat, but odds are you'll get tangled up in no time and shit the bed. Because “democracy” has a fuckton of nuances. Democracy was designed to slow things down. Communities are there to get things moving.

Running a community is more like running a company than running a country

Small companies are run by one charismatic founder, and the approach is all upside: decisions get made fast, mistakes get fixed without a lengthy hunt for someone to blame, and the project moves at the speed of light. Same goes for communities.

As you grow, of course, you have to start adding moderation systems, decentralized decision-making and assorted other bureaucracy. But even for that particular growing pain, the “world of private enterprise” has already come up with a bunch of tried-and-tested solutions. ? Hi, libertarians! I wouldn't count myself among you, but I do keep an eye on your methods.

Rushing off to build a democracy just because it's a word you recognize is a recipe for failure.

Even countries, with all their power and resources, don't always manage to pull off democracy (lol). What makes you think you will in your little group chat for besties? Find some other tools, for fuck's sake.

A community is always a private party with its own rules and values. Use that to your advantage. If you're just getting into running communities, I've saved you a couple of years of trial and error right there.

By the way, this cuts both ways: if a community starts giving you a spiel about “democracy” and “free speech,” they're 100% about to fuck you over :)

Hi, Musk! Hi, Trump! How are TRUTH and Gab doing?



As they used to say at university, this problem has no analytical solution. We can get arbitrarily close to the answer, but we'll never reach it.

All moderation is a mix of fuzzy rules, personal ambitions and the inevitability of the tragedy of the commons from game theory.

That last bit says: in any community, there will always be conflicts between the interests of the individual and the survival of the community as a whole.

Say you've got a field where your whole village grows potatoes. Every villager's goal is to plant more and more potatoes to make more and more money. But after a while the soil will lose its fertility and the village will die.

Same thing in communities. When a couple of toxic old-timers fill 90% of the chat with their chatter, for the other 1000 members it turns into a stinking dump where they can't get a word in edgewise. And the chat dies.

If the village wants to survive, it needs someone to regulate the harvest or make the tough calls, like banning that toxic old-timer. That person will always be hated. And they're the only reason the community will survive.

I've seen it happen plenty of times myself.

I even have a favorite line about it (from some tweet, I forget which one):

When you get banned, it's CENSORSHIP. When you do the banning, it's moderation))0))

Musk even bought Twitter for that.

By definition, somebody's always going to be unhappy with moderation. You can have a pile of rules, decentralized consensus, a magic crystal ball that tells the future, whatever. Sooner or later you'll have to penalize one person to make things better for a thousand others.

And just like that: +1 enemy.

Accept it as a fact. Then let it go.

I can't think of a time a community died because somebody got banned. Seriously. ? I do remember the old drama when Lepra banned Artemy Lebedev and he produced that famous post about how it had “gone to fucking shit because of its eleetism.” Except Lepra carried on happily and productively for another ten years or so, and nobody gave a fuck about Artemy Lebedev. Communities that kicked out the mods in the name of FrEe SpEeCh and promptly sank into the swamp, on the other hand? Plenty.

The way I see it, there are roughly three approaches to moderation. In my ten years of working with communities, I've always run into either one of these or some combination of them.



🤔 Real-world examples: anonymous boards, any small group chat

Best case, the person who started the community keeps order. Worst case, nobody does.

“We'll just do nothing. People will sort themselves out somehow, they're not idiots!” That's the easiest and most popular option, especially among people who remember the “early internet,” when you could still get away with it because there weren't many people around and their average intelligence was pretty high.

It's cheap and effective in group chats with friends and on small sites. With a big crowd and in the long run, though, it's doomed. That old LessWrong piece I'm always pointing people to goes into it in detail: “Well-Kept Gardens Die by Pacifism.”

No matter how lovely your garden is, if you don't guard it, somebody's going to take a shit in it tomorrow. That's how people work. We get more pleasure out of destroying things than creating them.

No point crying about how imperfect the world is. Just know about that imperfection and plan for it. Be ready to get your hands dirty. You can't build community moderation on pacifism and staying out of it.

And out on the Big Internet, you also have laws from assorted random countries requiring you to remove hate speech from the comments or scrub copyrighted content. So at scale, moderation by anarchy just isn't possible anymore.

By the way, technocratic attempts to rely entirely on upvotes-downvotes-karma-ratings aren't all that different. Except now the resident psychopaths have cool tech guns to shoot everyone else with.



🤔 Real-world examples: Twitter, Habr, TikTok, other social networks and group blogs

The very next thought that pops into your head: so we need to write a list of rules and make everybody follow them!

If you've never tried writing rules for even 100–1000 people, this actually sounds workable and fair in theory.

Somebody recently posted a great little quiz called “No vehicles in the park” on Hacker News. It takes 3 minutes, give it a go. You're given one rule, “No vehicles allowed in the park,” and then you have to answer a series of questions like “If I'm walking through the park carrying a skateboard, am I breaking the park rule?”

At the end you get a chart showing how your answers stack up against everyone else's. I only got a 68% match, for instance.

We humans are incredibly efficient and inventive when it comes to getting around written rules. Maybe language is an imperfect tool for writing down rules, maybe mathematical or formal logic hasn't advanced far enough, hell if I know. Discuss in the comments.

Just give me a goal and write a rule that explicitly forbids me to achieve it. I'll find a way around it for you in a couple of minutes.

Still, for all their obvious flaws, you do need rules. They give newcomers a basic idea of how people are expected to behave in this place. We humans get uncomfortable when we don't know what to expect. Will we get called out for being rude or praised for being inventive?

Rules matter to the people who have no intention of breaking them. Ironic.

Plus, rules give the other members a “sense” that the moderators' actions are legitimate.

Even in the “shit on the rug” example above, the mods are still going to have to punish the culprit and make sure the rug doesn't get any filthier. In practice, they'll have to act outside the written rules, but most of the peaceable members of the community, who don't like having a filthy rug, will be fine with that call.

Lawyers have a line for cases like this: “It goes against the letter of the law, but not its spirit.”

If you help your mods hone that skill, or recruit people who have an instinct for it, a team like that can actually handle even a fairly large community.

Until they hit the next problem: over time, a system like this slides into either bureaucracy or abuse of power.

When mods routinely stray from the “letter of the law” and use their own judgment, sooner or later you're going to get mistakes and abuses.

But if the mods make an honest effort to “update” the rules every time someone finds a loophole, eventually the rules start to look like the German tax code, which barely fits in three cabinets. Enter “rule by bureaucrats”: the winner isn't whoever's right, it's whoever knows the most laws.

Both outcomes are shit.

So we end up needing some new system, one that might not even be directly involved in moderation but has authority over the moderators themselves and the calls they make.

“Moderating the moderators,” if you like wordplay.



🤔 Real-world examples: Wikipedia

This approach lets you strike a balance between bureaucracy with its endless rules and total “go with your gut” anarchy.

You no longer start with formal rules. In their place you have higher-level “principles.” The rules either follow automatically from those principles or are worked out by consensus (agreement) among the mods.

Wikipedia has its “Five pillars,” Vastrik.Club has its “Club values.”

Unlike “rules,” which people love to add to every time somebody sneezes, “principles” are always few in number and might not change once in the entire life of the community.

That goes a long way toward keeping bureaucracy from taking root. Now when a case comes up, the question isn't whether it complies with some formal law that says “you can do this, you can't do that.” It's whether it's in line with the community's principles.

For example, Wikipedia's fifth pillar explicitly says it has no hard-and-fast rules, and the fourth explicitly says not to try to “test them to breaking point.” With that setup, the answer to our shit-on-the-rug problem is obvious and follows automatically.

Principles aren't rules. They tell you how to make the rules. Consensus takes it from there.

It's a little more complicated than writing down a formal list of dos and don'ts, but it lets you stay flexible, scale up and adapt to a changing world and to crises (like the one that hit Russian-speaking communities after 2022).

That doesn't mean you can't have “policies” for specific situations. Wikipedia has plenty of them (though you're better off looking at the English version, it's better laid out).

Take policies on using images, copyright, or editing articles about yourself: clearly you need specific guidelines for those. You can also have local or temporary policies for issues people are having a lot of flame wars over and can't agree on yet, like which Russian preposition to use with “Ukraine”: “na” or “v.”

And notice how clearly they're written and how well they make their case. Hats off. You don't see that on the internet anymore.

Unlike rules, policies always have a limited scope and always answer to those same higher-level “principles” (they don't get to invent their own). And when a policy stops serving a purpose, you can just toss it.

For example, here are some policies I've found myself reaching for pretty often in all sorts of communities:

  • A policy on swearing in posts or chats
  • A policy on post titles (banning clickbait is pretty useful)
  • Formatting policies (for instance, in our Buy/Sell/Trade chat, every message has to follow the same format; anything else gets deleted as chat clutter)
  • A policy on using photos of people and personal data in general
  • A policy on reposting content, both from the community to the outside world and the other way around
  • A policy on avatars and using real names or pseudonyms

Now let's talk about the problems with this approach.

First problem: finding thoughtful moderators. Without clear-cut rules, you need the people making the calls to put a lot more thought into what they're doing. You can't just hire people like that off the street ? much as startups trying to outsource moderation to Kenya love doing that . You have to let them grow into the role and recruit from among your active members.

Moderation is already mostly thankless, unpaid work. Now the bar for getting the job is suddenly a lot higher, too. You need people who can do more than “ban by the book”: they have to be able to use their words, talk problems through and reach a consensus. Good luck finding people like that even in everyday life. That skill is worth a lot.

Second problem: the complexity of the system and how it's organized. Just look at the hierarchy of editors on Wikipedia. And they couldn't even fit all the different roles (user rights) in there.

For simplicity, I cropped off the right side, which has a whole other chart like this for the Wikimedia Foundation hierarchy and so on. This is just the community
For simplicity, I cropped off the right side, which has a whole other chart like this for the Wikimedia Foundation hierarchy and so on. This is just the community

To organize something like that from scratch, you need either an IQ of 200 or five community launches under your belt. Systems like these don't just spring up on their own. There's always a dictator with a vision behind them, someone like Jimmy Wales, who spends years sticking everyone's nose in the principles and picking out the people who can actually live by them.

Those are the two reasons attempts to make “a clone of X” almost always fail.

Tech bros can replicate Reddit's “algorithms” or Twitter's, no problem. But only a handful ever stop to think that organizing people into functioning communities has “algorithms” of its own. And if those aren't the first thing you think about, they'd damn well better be the second.



The good news is, if you're running a Telegram chat for your friends, you probably don't need any of this shit. Good old-fashioned dictatorship and bans for “insufficient enthusiasm” (c) (tm) will do just fine.

The main thing is not to buy the myth of “self-regulation.” Have a clear idea of what you're building and who you're building it for. Beautiful gardens don't grow themselves. The rest you can figure out as you go.

One last hot take from me:

Moderation isn't about bans and flame wars. It's about the values and principles you say you stand for.

That's exactly what sets you apart from the other 10000000000 communities out there. Which is why it's worth thinking about BEFORE you launch your community, not after.

I hope this post means we get a few more well-moderated communities on the internet where it's actually nice to hang out. Feels like we've kinda fucked the whole thing up lately.

Too real. I feel this one personally
Too real. I feel this one personally


Comments 👇
More? Here you go