Writing Server Rules People Actually Read

Published 11 September 2026 · JoinGG · Guides

Cut your rules to five to seven. State each one as an observable behavior rather than a virtue, and attach the consequence to the rule itself. A forty-rule wall nobody reads makes every ban look arbitrary. Short, specific and enforced beats long, complete and ignored.

A rulebook is not a list of everything a player could do wrong. Its only job is to let a stranger predict what will get them banned. Predictability is the whole product, and it is the thing a long rulebook destroys, because nobody finishes reading it.

That is a numbers problem before it is a writing problem. Across the 11,561 servers listed on JoinGG, the pattern in the description field is the same one you see in MOTDs: either two lines that say nothing, or a wall that says everything. Both produce the same outcome at the moment a player is deciding whether to connect, which is that they learn nothing about how this server is run.

This guide is about the middle option. Seven rules, each written as a behavior with a consequence attached, living in four places at four lengths, enforced the same way by every admin you have, and changed on a schedule people can see coming.

Why a 40-rule wall makes every ban look arbitrary

Start with the reading budget. A Source engine MOTD is on screen for a handful of seconds before the player clicks through to spawn. A Minecraft server-list entry gives you two short lines. Someone joining mid-round reads perhaps thirty words before something more interesting happens. Against that budget, a 900-word rules page has an effective readership close to zero among new joins, which is exactly the population the rules exist to inform.

The damage is not that the rules go unread. It is what unread rules do to enforcement. When there are forty of them and nobody has read any, a ban works backwards: an admin decides someone has to go, then reaches into the pile for the rule that fits. From the outside, that looks like the decision came first and the justification second, because that is what happened. Do it twice in public chat and your staff have a reputation for banning on mood.

The second failure is worse. Rule 27 says no spawn camping, and nobody has ever been punished for spawn camping. The moment one written rule goes unenforced, every other rule becomes negotiable, and every argument in your appeals channel becomes an argument about rule 27. A rulebook that lies about one thing is not trusted about anything.

There is a cheap diagnostic. Pull six months of ban and mute records and count actions per rule. Almost every forty-rule set has a handful of rules producing nearly all of the enforcement. Those are your real rules. The rest are decoration, copied from a server that ran a different game mode, or norms that got promoted by accident. Cutting them changes nothing about how your server actually runs, and it removes the ammunition people use to argue that enforcement is inconsistent. If your population is drifting down at the same time, the rulebook is rarely the only cause, and the piece on why nobody joins your server covers the rest of the list.

A rule is not a norm, and mixing them breaks enforcement

A rule is enforced by staff, against evidence, with a consequence you administer. A norm is enforced by the group, through social pressure, with no consequence at all. Both are real and both matter. Writing them in the same list is the single most common cause of a forty-rule wall.

Norms are things like calling your rotations, not taking the last vehicle, letting the new player have the good gun, queuing for the workbench, and not building something hideous next to spawn. Every healthy community has dozens. None of them survive contact with an admin panel, because enforcing them requires you to judge intent and taste, and your staff will not agree with each other about either.

Promoting a norm to a rule creates a debt you have to pay in enforcement. If you cannot pay it, you have not gained a rule, you have degraded the rulebook. Demoting a rule to a norm is the legitimate move, and it is how you get from forty down to seven without losing anything you were actually doing.

Your tooling decides the category, not your preference. On the 4,230 Minecraft servers listed on JoinGG, no building within 100 blocks of spawn is a rule if you run a claim plugin that draws the boundary and a rollback tool that proves who placed what. It is a norm if enforcement means an admin squinting at coordinates and deciding whether it felt too close. Same sentence, two categories, and the difference is whether the evidence exists. The same logic sorts your whole list, and it shifts by game mode: an anarchy server and a curated SMP need almost disjoint rulebooks, which the breakdown of Minecraft server types gets into.

Keep the norms. Write them down somewhere else, under a heading like "how we play here", explicitly labeled as things staff will not enforce. That label is the release valve that lets the enforced list stay at seven.

The three tests every rule has to pass

Run every candidate rule through three questions. A rule that fails any of them is a norm, a heading, or a rewrite waiting to happen.

Is it observable?

Could a stranger reading a chat log or watching a thirty-second clip reach the same verdict you would, without knowing what anyone intended? "Do not be toxic" fails, because toxic is a conclusion. "No slurs or targeted insults" passes, because you can point at the line. "No griefing" fails, because your players and your staff draw that boundary in different places. "Do not break, place or take blocks inside another player's claim" passes.

Intent is the tell. Any rule requiring you to establish what someone meant is a rule two admins will apply two ways, and it is the rule that generates your appeals.

Is it attributable?

Can you tie the act to one account with evidence that survives an appeal a week later? A chat log line, a block-history entry, a demo tick range, an RCON log, a txAdmin action record. If your tooling cannot produce that evidence, you cannot enforce the rule, and you should write a rule you can prove instead of one you would prefer.

Roleplay servers have the hardest version of this problem, because so much of the disputed behavior is in character and the evidence is a twenty-minute voice clip. That is why serious FiveM roleplay servers lean on recorded rulings and named precedents rather than rule text alone, a habit worth borrowing whatever you run. The guide to choosing a FiveM roleplay server is a useful read from the other side of the desk: it is essentially a list of the rule-quality signals players use to pick between communities.

Does it carry its own consequence?

The rule states what happens the first time and what happens on repeat. Without that, an admin invents a punishment at two in the morning, and the next admin invents a different one for the same act on Thursday. Use four tiers: a logged warning, a mute or timeout, a kick, and a ban with a stated duration. Escalation is mechanical, not discretionary, and the simplest workable version is that the same rule broken twice within 30 days moves one tier up. That turns most admin work from judgment into lookup, which is the entire point.

Before and after: rewriting bad rules into enforceable ones

Here are the six rules almost every server has, why each one fails the tests above, and what they look like rewritten. The pattern is consistent: replace a virtue with a verb, add a number, and attach the consequence.

Three things changed in every row. The subject became something a camera could record. A number appeared, so the boundary is a fact rather than a mood. And the consequence moved into the rule, so the player knows the price before they pay it and the admin does not have to price it live.

The last row is the important one, and it is the hardest to give up. Sandbox communities feel this most: on the 351 Garry's Mod servers listed, sprays, props and prop-kills produce endless situations nobody wrote a rule for, and the temptation is to keep a catch-all. Resist it. Write the three prop rules you actually enforce and let the rest be a norm.

A seven-rule set you can adapt

This is a working skeleton, not a template to paste. Swap the game-specific terms, keep the shape: numbered, observable, consequence attached.

1. No slurs or targeted harassment in text or voice. First offense is a 24-hour mute; a second within 30 days is a 7-day ban.

1. No client modifications that reveal hidden information or automate aiming. Permanent ban, appealable with evidence.

1. Do not take, break or alter another player's property in a claimed area. Rollback plus a warning; a second time is a 7-day ban.

1. No advertising other servers in chat, names or direct messages to our players. Permanent ban, no warning.

1. Do not impersonate staff or use a name that implies you are staff. 7-day ban and a forced name change.

1. Appeals go in the appeals channel. Arguing a decision in public chat is a 30-minute mute and does not affect the appeal.

1. Staff log every action. If a mute or ban is not in the log with a rule number, it is void on request. This rule binds us, not you.

Rule seven is the one people skip, and it is the one that buys you the benefit of the doubt for the other six. A community that can see its staff bound by something forgives mistakes; one that cannot assumes the worst about every decision.

Notice what is missing. No "respect others", because rule one already covers the part you can prove. No "no camping", because you almost certainly do not enforce it. No "English only" unless you genuinely will act on it, in which case name the channel it applies to rather than the whole server. If you find yourself wanting a rule for who is allowed in at all, that is not a rule, it is an entry gate, and the write-up on running a whitelist people actually apply to is the better tool. A whitelist enforces at the door instead of after the damage.

Where should server rules physically live?

One ruleset, four surfaces, four lengths. Only the longest is allowed to be long.

• MOTD or join screen, about 40 words. Not the rules. The two rules people break in their first hour, plus where the full list lives. On Minecraft you get two short lines in the server list, so treat it as a headline, not a document.

• In-game /rules, the canonical seven. Numbered and paginated, in the same order everywhere, so an admin can type "rule 3" in a kick reason and it means one specific thing.

• Discord rules channel, the full version. The consequence ladder, the appeal process, the norms section. Discord's membership screening gate is the only surface where you get a recorded acknowledgment click, which is worth having when someone claims they were never told.

• The kick and ban message, your highest-read surface. Nobody reads the rules; everybody reads the reason they were removed. "Kicked" teaches nothing. "Kicked: rule 1, slurs. 24h mute. Appeal at example.com/appeal" teaches the entire system in one line to the person most motivated to learn it.

• Your listing description, one line. What the server is, and what gets you banned. Filtering someone before they connect is cheaper than moderating them afterwards, and it is part of what the piece on what makes a server listing trustworthy is measuring. If you have not claimed a listing yet, you can add your server for free and write that line properly.

• The long version on your site or wiki. Two thousand words is fine here, because the only people who reach it are staff and appellants, and both of them need the nuance.

The failure mode is drift. Four surfaces means four copies, and within a quarter the Discord version says 24 hours while /rules still says 12. Pick one source of truth, put the version number on every copy, and make updating all four a single checklist item rather than four separate jobs someone will half finish.

Training admins so two of them reach the same answer

Consistency across staff is not a personality problem, it is a documentation problem. Write the admin handbook as a decision table rather than prose: situation, rule number, action, log format. Nobody reads three paragraphs about tone while a fight is happening in chat.

Then calibrate, because a table alone does not produce agreement. Take ten real incidents from the past month, strip the names, and have every admin independently write down the action they would take. Compare. If one clip gets three different answers, the rule is broken and not the admins, and you rewrite it that week. Run this when you onboard someone and once a quarter after that. It takes an hour and it catches the ambiguity you stopped seeing years ago.

Four structural habits do the rest:

• One log format, always. Timestamp, admin, target account, rule number, action, duration, evidence link. SourceBans++, LiteBans and txAdmin all record this if you configure them to. An action without a log line is an action you cannot defend.

• Probation for new admins. Thirty days of mute and kick only, no bans. In my experience, bad permanent bans come from someone in their first fortnight who has not yet learned what normal looks like.

• Two keys for permanent bans. A permanent ban needs a second admin to sign off within 24 hours or it downgrades automatically to seven days. This one policy removes most of the damage from decisions made at two in the morning.

• Staff your actual peak. Coverage matters more than headcount. FiveM alone carries roughly 72,600 players across 1,854 servers, and you can see the shape of the evening in the FiveM population statistics. Across every game JoinGG tracks, the busiest measured hour is Saturday at 18:00 UTC with 242,719 players connected and the quietest is Wednesday at 06:00 UTC with 73,158. Put your admins on the first one.

Publish the aggregate monthly: actions taken, appeals received, appeals upheld. An overturn rate of zero does not mean your staff are perfect, it means nobody believes the appeal route is real. Somewhere between 5 and 15 percent is the sign of a process people trust enough to use.

How do you change the rules without a revolt?

The revolt is never about the rule. It is about being surprised by it, and surprise is the thing you are actually removing.

Lead with the number that caused the change. "We logged 41 spawn-camping reports in August against 6 in July" is an argument. "We are updating the rules" is an announcement that invites people to guess your motives, and they will guess badly. Then run a fixed 14-day cycle so the community can see the whole shape of it in advance: announce with the reasoning on day zero, take comments for seven days, enforce warnings only for the next seven, and go live on day fourteen. The comment window is not a vote, and you should say so plainly.

Three rules about the rules themselves:

1. Never retroactive. Bans issued under the old wording stay under the old wording. If you delete a rule entirely, review the bans it produced and lift the ones that would not stand today.

1. Version everything. "Rules v4, effective 25 September 2026", with a changelog. Almost nobody reads the changelog. Its existence still ends the argument about whether you changed something quietly.

1. Trim as often as you add. At every quarterly review, any rule with zero enforcement actions in six months goes back to being a norm. This is the mechanism that keeps you at seven instead of drifting to forty again, and it is the step everyone skips.

Be careful with votes. Putting a rule change to the community reads as democratic and behaves as a trap, because you will not accept the result if it goes the wrong way, and everyone can tell. Save voting for genuinely reversible cosmetic decisions like map rotation or a wipe date, where the mechanics described in the explainer on how server voting works earn their keep. Rules are your responsibility, and pretending otherwise costs more trust than deciding openly ever does.

Then watch the population for a fortnight afterwards. Concurrency is the honest referendum: if evening numbers drop 15 percent and stay down, something in the change is landing badly, and you will see it in your own curve before anyone says it out loud. Comparing your shape against the Minecraft population statistics or whichever game you run separates a rule problem from an ordinary seasonal dip.

Related reading

• How players actually find servers, which is what your listing line is competing in.

• Discord changes affecting gaming communities, relevant if your rules channel is doing the heavy lifting.

• How many slots your server should run, since crowding drives a surprising share of rule breaches.

Frequently asked questions

How many rules should a game server have?

Five to seven enforced rules, plus a separate unenforced list of norms. The limit is not aesthetic. It is the number a new player can hold in their head after ten seconds on a join screen, and the number your admins can apply the same way at two in the morning.

Where should the rules live if players only read one place?

The kick and ban message. It is the single highest-read surface you control, because a player who ignored every other surface reads the reason they were removed. Put the rule number, the plain-language behavior, the duration and the appeal link in that one line.

Is do not be toxic ever acceptable as a rule?

It is fine as a heading and useless as a rule, because toxic is a judgment rather than an observable act. Keep it as the section title if you like the tone, then list the behaviors underneath: slurs, targeted insults, following a player between rounds to abuse them.

What do I do when two admins enforce the same rule differently?

Assume the rule is broken before you assume the admin is. Run the same incident past three admins independently. If you get three answers, the wording is ambiguous or the consequence is missing. Rewrite it that week, and tell the admins which reading is now correct.

Should I copy another server's rules?

Copy the structure, never the list. Copied rulebooks are how servers end up with rules for game modes they do not run and tooling they do not have. Every rule you keep must be one your logs can prove and your staff will actually act on.

How often should I update the rules?

Review quarterly, change rarely. At each review, count enforcement actions per rule over the past six months. Rules with zero actions are norms in disguise and should be deleted. Ship any real change on a 14-day cycle with a comment window and a warning-only period.

Rules are the cheapest infrastructure you own. Seven of them, written as behaviors, priced with consequences, mirrored across four surfaces and enforced identically by everyone with the ban command, cost you an afternoon to write and remove most of the arguments you would otherwise have every week for years. Once yours are in shape, the next job is making sure people can find the server they now describe accurately: JoinGG polls 11,561 servers hourly across 147 games in the catalogue, 31 of which currently have servers listed. Browse the communities directory to see how established groups present their rules, check the cross-game population statistics to find where the players actually are, and use the full game catalogue to see whether the game you want to run has an audience waiting for it.