Writing Rules People Actually Read
Open the rules page of almost any community and you'll find some version of the same document: a numbered list running to fifteen or twenty items, several of them redundant with each other, most of them written in the defensive, slightly legalistic tone of a founder anticipating every conceivable bad-faith argument someone might make while appealing a ban. It's not that the concerns behind these rules are wrong. It's that the format is wrong for the actual job a rules page needs to do, which is to change behavior before it happens, not to win an argument after the fact. A document optimized for the second goal is almost always worse at the first one, and most rules pages are optimized entirely for the second.
This matters more than founders tend to assume, because the rules page is one of the only pieces of a community's culture that's written down and stable, as opposed to the much larger, much fuzzier body of unwritten norms that actually govern most day-to-day behavior. When it's written well, it does real work — it sets expectations before anyone has to learn them the hard way, it gives moderators a shared reference instead of relying on individual judgment call by call, and it gives members a legitimate way to push back if they think enforcement went wrong. When it's written badly, which is most of the time, it does almost none of that, and instead sits there as a wall of text nobody reads until the moment they're accused of breaking it, at which point they read it adversarially, looking for a technicality rather than absorbing the spirit of it.
Nobody reads a rules page the way you wrote it
Founders write rules pages the way they'd write a legal document, because the underlying fear is usually legal-adjacent — what if someone does something terrible and I didn't explicitly say it wasn't allowed? That fear produces documents that try to enumerate every possible bad behavior in advance, with careful qualifying language, exceptions to the exceptions, and a tone that reads more like a terms-of-service agreement than something a human being would actually absorb before deciding how to behave.
The problem is that almost nobody reads a rules page the way it was written. People skim. They scan for headers, they read the first sentence of each numbered item and assume they've got the gist, and they close the tab well before reaching item fourteen, which happens to be the one rule that would have actually applied to the thing they end up doing three weeks later. A twenty-item rules page isn't more protective than a five-item one; it's less protective, because the marginal fifteen items past the point where attention naturally drops off don't change anyone's behavior — they just make the document longer without making it more effective, and they bury the handful of rules that actually matter under a pile of ones that rarely come up.
There's a second, subtler problem with long rules pages: they signal distrust. A document that anticipates fifteen different bad-faith scenarios before anyone has done anything wrong tells a new member, implicitly, that the founder expects the worst from them. Most new members haven't done anything wrong and have no intention of doing anything wrong, and greeting them with a wall of preemptive prohibitions sets an oddly adversarial tone for a space they were hoping to feel welcome in. Short, clear rules read as confident. Long, hedge-everything rules read as anxious, and anxiety is not a tone that attracts people who want to stick around.
The five-to-seven rule
A rules page that actually gets read and actually gets followed almost always fits in five to seven core items, each expressed in a single clear sentence, without legal hedging. This isn't an arbitrary number — it roughly tracks how much a person can hold in working memory after a single skim, which means it's roughly the number of rules that can survive the transition from "written down somewhere" to "actually influencing behavior in the moment." Every rule past that point competes with the ones before it for the same limited attention, and past seven or so, that competition starts actively degrading recall of the rules that matter most.
Getting down to five to seven items requires real editing discipline, because the natural instinct is to keep adding items every time a new edge case comes up. A member does something mildly annoying that isn't covered anywhere, so the founder adds a rule specifically addressing it. Six months later there are eleven narrow, specific rules, each one a fossil of a single incident that will likely never recur in exactly that form again. The better move, almost always, is to ask whether the new situation is actually a new category of harm, or just a specific instance of a category you've already covered in broader language. Most of the time it's the latter, and the fix isn't a new rule — it's trusting the broad rule you already have to cover it, with moderators applying judgment to the specific case rather than the rules page trying to anticipate every case in advance.
The categories that tend to actually need their own dedicated line are remarkably consistent across different kinds of communities: no harassment or targeted abuse, no hate speech or content that attacks people for who they are, no spam or self-promotion outside designated spaces, staying on topic, respecting other members' privacy, and following the platform's own broader terms of service. That's six. Almost every specific bad behavior a community will actually encounter falls under one of those six broad umbrellas, and writing the umbrella clearly and trusting moderators to apply it is far more resilient than trying to enumerate every possible instance underneath it in advance.
Write for the 95% reading this in good faith
A subtle mistake in how rules get worded is writing every sentence as though it's addressed to the worst possible bad-faith actor a founder can imagine, rather than to the overwhelming majority of members who are reading in good faith and simply want to know what's expected of them. This produces language full of hedges and closed loopholes — "any content that could reasonably be construed as," "including but not limited to," "at the sole discretion of the moderation team" — that reads as hostile even to someone with no intention of ever breaking a rule.
Rules written for the 5% who'll look for loopholes end up alienating the 95% who never would have. And the uncomfortable truth is that the hedging rarely even works against the 5% it's aimed at — a determined bad-faith actor will find a technicality in even the most carefully lawyered rules document, because that's the nature of bad faith: it's not actually looking for the rule's true intent, it's looking for any gap it can exploit, and there is always some gap in natural language, no matter how many qualifying clauses you add. What the hedging actually does is degrade the reading experience for everyone who wasn't going to be a problem in the first place, in exchange for a false sense of protection against people who were never going to be stopped by careful wording anyway.
The better approach is to write in plain, direct language aimed at a reasonable, well-meaning reader, and to handle the bad-faith 5% not through exhaustive wording but through moderator discretion applied case by case, with an explicit line in the rules that says as much — something like "moderators use judgment; the spirit of a rule matters more than a technicality." That single sentence does more to close loopholes than fifty words of legal hedging, because it explicitly removes technicality-hunting as a viable strategy, without making the document read as adversarial to everyone else.
Examples do more work than definitions
Abstract language is where most rules pages lose their grip on actual behavior. "No harassment" is a fine header, but it's also vague enough that a huge range of behavior sits in ambiguous territory around its edges — is repeatedly disagreeing with someone harassment? Is a single sharp comment? Is a joke that lands badly? Members genuinely don't know, and in the absence of clarity, they either self-censor more than necessary or push right up against a line they can't see, both of which are worse outcomes than a rule that actually clarified where the line sits.
One or two concrete examples attached to each core rule do more to clarify intent than any amount of additional abstract wording. Under "no harassment," a line like "this includes repeatedly messaging someone after they've asked you to stop, or piling onto a single member across multiple threads" gives members something they can actually pattern-match against their own behavior, rather than an abstract noun they have to interpret alone. Examples also help moderators, who otherwise end up relitigating the exact meaning of "harassment" from scratch every time a new, slightly different case comes up. A couple of anchoring examples per rule, chosen to represent the clearest, least ambiguous version of a violation, gives everyone — members and moderators alike — a shared reference point that plain definitions rarely provide on their own.
Put consequences next to the rule, not in a separate section
Most rules pages separate the list of prohibited behaviors from the list of consequences, usually in two different sections, sometimes on two different pages entirely. This split makes intuitive organizational sense to a founder writing the document, but it actively works against a reader trying to understand what will actually happen if they cross a line, because now they have to hold a rule in mind, jump to an entirely different part of the document, and manually match it up with whatever consequence structure applies.
Putting a brief consequence note directly next to each rule, rather than in a separate section, closes that gap and does real work in shaping behavior, because people respond differently to a vague, distant threat of "enforcement action" than they do to a concrete, specific, nearby one. "No harassment. First offense is a warning and the offending content removed; repeated or severe cases result in a ban" sitting right under the rule it applies to is far more legible, and far more likely to actually be read in the moment someone is deciding whether to send a message, than the same information filed three sections away under a generic "Enforcement" heading that most readers never scroll down to in the first place.
Revisit the rules publicly, not just privately
Most founders update their rules privately and silently — a word changed here, a new line added there, with no announcement and no visible history. This is understandable; nobody wants to make a big production out of a minor wording tweak. But it has a real cost: members never develop a sense that the rules are a living document they have any stake in, because they never see them change in a way that's connected to anything they can observe.
A better practice, especially once a community has grown past its earliest days, is to occasionally post an update — even a short one — when a rule changes in a way that matters, explaining briefly what changed and why. This does two things. It keeps existing members' mental model of the rules current, rather than relying on people to periodically re-read a static page they long ago stopped checking. And it demonstrates, concretely, that the rules respond to real events in the community rather than existing as a fixed document handed down once at launch and never revisited. Members who see rules actually respond to lived experience tend to trust the enforcement of those rules more, not less, because it's evidence the rules were written by someone paying attention rather than someone copying a template once and moving on.
The rules page is not the only place rules live
A mistake that compounds the length problem is treating the rules page as the sole place a community's norms are ever communicated, when in practice the vast majority of what actually governs day-to-day behavior in a healthy community is unwritten, absorbed from observing how other people act and how moderators respond to edge cases, not read off a static page. This is fine, and even desirable — no rules page could ever fully specify every norm a functioning community needs — but it means the rules page's job isn't to be exhaustive. It's to cover the handful of hard lines that need to be explicit and unambiguous, while trusting the softer, more contextual norms to develop the way they always do, through modeling and gentle correction rather than through ever-expanding documentation.
Founders who understand this distinction write shorter, clearer rules pages, because they're not trying to make the document do a job it was never suited for. The rules page handles hard lines: harassment, hate speech, spam, illegal content, doxxing. Everything softer — how formal or casual the tone should be, how much off-topic chat is tolerated, what level of disagreement is normal versus too heated — gets handled the way it always has been in every functioning community throughout history, through example, through gentle moderator nudges in the moment, through the accumulated texture of how people actually treat each other, not through an ever-lengthening list of numbered items nobody's going to fully read anyway.
What this looks like assembled
Putting all of this together, a rules page that people actually read and actually follow tends to share a specific shape: five to seven core rules, each a single plain sentence, each followed immediately by one or two concrete examples and a brief note on what happens if it's broken, written in a tone aimed at the reasonable majority rather than lawyered against the bad-faith minority, with an explicit line trusting moderator judgment over technicalities, updated visibly and occasionally rather than silently and constantly, and treated as a foundation for the hard lines rather than an attempt to exhaustively specify every norm the community will ever need.
None of this is complicated in principle. It's mostly a matter of resisting a set of natural but counterproductive instincts — the instinct to add a new rule every time something new goes wrong, the instinct to hedge every sentence against a hypothetical bad-faith reading, the instinct to treat the rules page as a legal shield rather than a piece of writing meant to actually be read. Fix those instincts and the rest tends to follow. The result is a document members can actually carry in their heads, which is the only kind of rules page that ever really changes what people do.