Skip to content

What a West Marches campaign is, and how to run one without it running you

West Marches is a way of organising a campaign, not a setting or a rule set: a pool of players bigger than one table, a dangerous map, and sessions that only happen when the players arrange one. This is what the original campaign did, which of its rules are load-bearing, and how to run one on a Discord server instead of a mailing list.

· 11 min read

What is a West Marches campaign?

A West Marches campaign is a sandbox game run for a pool of players much larger than one table, with no fixed group and no fixed night. The players decide where they want to explore, gather whoever is free, and book the GM; the GM runs the part of the map they chose. Between expeditions every character returns to the same safe town, and what each party found is shared with everyone else. The thing that persists is the world and the roster, not the party. If you have one group and one night that suits it, a fixed campaign is simpler to run; how to start a D&D campaign covers that setup.

The name comes from one campaign. Ben Robbins ran West Marches from the start of 2001 into 2003, in what was then the new third edition of D&D, and wrote it up on his blog Ars Ludi in October 2007. He described it as the opposite of the normal weekly game in three ways. No regular time: every session was scheduled by the players. No regular party: each game drew different players from a pool of around ten to fourteen people. No regular plot: the players decided where to go and what to do. Everything now called a West Marches style campaign is a variation on those three sentences.

His reasons were ordinary ones. Players who followed the plot without caring about it, adults whose calendars never lined up, and his own habit of putting off a session until it was perfectly prepared, which a party naming its destination cured by giving him a deadline. That is worth knowing because it is the test of whether the format suits you. If your table turns up every week and enjoys following the story, you do not have the problem West Marches was built to solve.

The West Marches rules: what makes it one

There is no West Marches rulebook. There is a handful of structural rules from the original campaign, and a game that keeps them is a West Marches game in any system, D&D or otherwise.

  • Players start every expedition. Nothing happens until a player proposes a destination and finds a party. The GM does not send quests, post a schedule or chase attendance.
  • Scheduling happens in public. Robbins’s advice to anyone running their own is to require every game to be announced on the campaign’s list, even when the players agreed it in person, so the campaign does not splinter into private groups.
  • Town is safe, and there is no adventure in it. Everything worth finding is in the wilderness; in the original, "town game" was a dirty word. A safe home base is what lets a different party form every time.
  • Danger rises with distance. Each region had a set level of threat and the next region out was a little worse, so players could judge a risk before taking it. That is what makes it fair to let first-level characters walk anywhere.
  • The world is the same for everyone. One map, one set of facts. What the Tuesday party broke is still broken when the Saturday party arrives.
  • The players keep the record. Write-ups and the shared map were their job, on purpose.

What the original did that you can change

The original also built characters from the Player’s Handbook alone and rolled every die in the open. Both served one idea: players should never need the GM’s permission or the GM’s mercy, because the point of the format is that their own decisions get them rich or killed. The idea matters more than either rule, and you can honour it in whatever system you run.

Modern games add things the original posts never mention: several GMs, a Discord server with a hundred members, downtime rules, a sign-up bot. Those change the workload, not the definition. If the players choose the expedition, the roster floats and the world persists, it is West Marches.

How to run a West Marches campaign: build the map, not a plot

Prep for this format is front-loaded and geographic. You are not writing a story. You are stocking a region so that whichever direction a party walks, something is there. Start smaller than you want to: a town, four or five named regions around it, and one or two sites in each.

  1. Draw the town and declare it safe. Walls, law, nothing to fight. Give it one place where characters meet between expeditions; in the original it was a taproom with a map carved into a table.
  2. Ring it with regions and give each a danger level. The nearest are survivable at first level, and each step outward is a notch harder. A player should be able to guess how dangerous a place is from how far away it is.
  3. Write one encounter table per region. Robbins called these tables the definition of each territory: what lives there, what the terrain does to you, and a result or two that sends you to the neighbouring region’s table instead. A table is reused every time anyone crosses the region, which is what makes on-demand sessions possible.
  4. Place sites that outlast one visit. A dungeon whose floor plan stays put can be occupied by something new after a party clears it. Put one room in each that the first party cannot crack, so they come home with a reason for someone to go back.
  5. Seed connections, not a plot. An inscription in one ruin that explains a sealed door in another. Nobody is told to follow them; a roster that shares its notes will.
  6. Give every new player one easy lead. The original handed each batch of newcomers a crude treasure map, which got them out of town without the GM proposing a session.

Scheduling a West Marches game: how a session actually happens

This is the part that makes or breaks the format, and the original ran it on two hard rules: the GM has to be free that day, and the party has to say where it is going far enough ahead for the GM to prepare it. Everything else is the players’ problem, deliberately. Someone names a day and a destination and asks who is in; the replies haggle over both; when there are enough people, it is a session. You prepare the place they named and nothing else.

Run it in one public channel and nowhere else. Private arrangements are how an open table turns into three closed ones. If the same four people keep going out together, Robbins’s suggested fix is a rule that parties must mix after two expeditions. If people keep getting left behind by whoever could not wait, he suggests the opposite rule: a party stays together for two. Pick whichever problem you actually have.

Expedition call (pin this format in the scheduling channel)

EXPEDITION CALL
CALLED BY:
WHERE: (region or site, as named on the shared map)
WHY: (the lead we are following, and who found it)
WHEN: (two options, with the time zone)
SEATS: __ of __ filled
  1.
  2.
  3.
WE NEED: (a healer, someone who was at the mill last time)
GM CONFIRMED: yes / not yet
REPORT WRITER: (decided before we leave town)

The shared map and the session report: what holds it together

In the original campaign the average player was at about a third of the sessions, and with a different mix of people each time. Robbins is blunt about what that means: everyone sees a different fraction of the game, and without shared information they are not playing the same game at all. Two things carried it. Game summaries written by the players, and one map that any party could add to.

The summaries began as a few lines of warning and bragging and grew into full stories that people enjoyed writing. That did two jobs at once. It kept absent players current, and it made them want to go and see for themselves. The map worked the same way: a region somebody else had filled in was a reason to book a session.

Both were the players’ work, and his advice to anyone running their own is to keep it that way: do not write the summaries and do not tidy the map, because doing it trains the players not to. Take that seriously. In this format a report is not paperwork. It is the reward for going.

Expedition report (copy this)

EXPEDITION REPORT
SESSION DATE:              WRITTEN BY:
PARTY:
WENT TO:                   BY WAY OF:

WHAT WE FOUND (add these to the map)
  -
WHO AND WHAT WE MET
  - (name, where, friendly / hostile / unknown)
WHAT WE CHANGED (cleared, broke, angered, opened)
  -
WHAT WE LEFT BEHIND (the door, the room, what we ran from)
  -
LEADS FOR ANYONE ELSE
  -
WARNINGS
  -
LOOT, AND WHO HOLDS IT
  -

What a player’s report cannot do

A report is written by one player, from one seat, a few days later. It is the right record for the roster and the wrong one for the GM, because it leaves out exactly what you need: the name the party gave the hermit, the price they agreed, the door they left open. Keep your own state-of-the-world notes beside the players’ reports, one line per site on what changed, or record the sessions so the detail exists somewhere.

That second layer is the one piece of this a tool can take. What DM Scribe does for an open table covers it, including when a volunteer archivist is the better answer.

What breaks first

Robbins saves his strongest warning for last, and it is not about prep: fear the social monster. In a weekly game everyone has a seat. In an open one, people wonder why nobody answered their call or why they were not asked along, and the format amplifies whatever insecurity the group already has. Watch the scheduling channel for the same names being left out, and use the mixing rule before it becomes a grievance.

The second failure is the GM. On-demand play means preparing on someone else’s timetable, and a popular game produces more requests than one person can run. Set a ceiling and say it out loud: one expedition a week, or two a month. Past that, the answer is a second GM, not less sleep. DM burnout is where the other road ends.

The third is the record going stale. It decays the first month the keenest writer gets busy, and nothing announces it. You find out when a party walks into a ruin that another party emptied, and nobody at the table knows.

The fourth is too few players. With six people and one night that suits them all, you have an ordinary campaign with extra scheduling. The rivalry between parties, which Robbins calls what the format is all about, needs enough people that nobody can be at everything.

West Marches on Discord, and with more than one GM

Many West Marches games now live on a Discord server, and the mapping from the original is direct. The mailing list becomes a scheduling channel, the taproom becomes an in-character tavern channel, the web copy of the map becomes a pinned image, and the summaries become a reports channel. How to run D&D on Discord has a channel layout to copy: its scheduling channel takes the expedition calls, its recap channel becomes the reports channel, and you add a tavern channel for talk in character.

Text play fills the gaps well. Downtime, haggling in town and the argument about where to go next can all happen as play-by-post between sessions, which keeps the people who cannot make a voice session in the game.

Several GMs change the bookkeeping more than the play. Each needs to know what the others’ parties did before running the same region, so agree on one place where the state of the world is written down and one rule about who may change a site another GM built. The players’ reports are not enough for this. They were not written for a GM.

West Marches, sandbox, hexcrawl, open table: the difference

The four terms get used as if they meant the same thing. A sandbox is any campaign without a set plot: the players choose the goals. A hexcrawl is a procedure for exploring a map hex by hex, with rules for travel and discovery. An open table is a campaign anyone on the roster can drop into, session by session. West Marches is the specific combination: a sandbox, run as an open table, where the players also do the scheduling, usually across wilderness you could run as a hexcrawl but do not have to.

The combination is what creates the pressure. A sandbox for one fixed party has no rivalry and no information gap. An open table where the GM sets the schedule and the plot is a drop-in game. Take away the player scheduling and you have taken away the thing the format was invented for.

It has become easier to pitch, too. Critical Role’s fourth campaign has been covered as a West Marches style game, with thirteen players split across three tables in one shared world, so the name, at least, no longer needs a paragraph of explanation.

Is a West Marches campaign right for your group?

It fits a specific situation, and outside that situation it is more work for a worse game.

  • Yes, if you have more would-be players than seats, and schedules that never line up.
  • Yes, if your players like planning, exploring and comparing notes, and you enjoy building places more than plots.
  • No, if the table wants a long character-driven story with the same companions. There is no regular party for an arc to happen to.
  • No, if the roster is not much bigger than one table, or you cannot be flexible about when you run.
  • No, if nobody wants to write anything down. The format runs on shared information, and somebody has to produce it.

The original West Marches, with sources

Every third-party figure quoted above, with where it came from and when it was checked. Prices move — if a date here is old, verify before relying on it.

WhatValueSource
Where the format comes from"Grand Experiments: West Marches", Ben Robbins, Ars Ludi, 17 October 2007: "West Marches was a game I ran for a little over two years"sourceas of 2026-11-13
The three defining rules"There was no regular time: every session was scheduled by the players on the fly"; "There was no regular party: each game had different players drawn from a pool of around 10-14 people"; "There was no regular plot: The players decided where to go and what to do"sourceas of 2026-11-13
When it ran, and in what system"We started West Marches at the very beginning of 2001 and ended in 2003. 3rd Edition D&D had just come out and we used it for the entire campaign"; characters were built "using the original Players Handbook" onlysourceas of 2026-11-13
The two scheduling rules"The GM has to be available that day"; "The players have to tell the GM where they plan on going well in advance"sourceas of 2026-11-13
How much of the game one player saw"the average person was in about a third of the games"sourceas of 2026-11-13
How information was shared"There were two main ways information got shared: game summaries and the shared map"sourceas of 2026-11-13
Encounter tables by regionOne wandering monster table per wilderness area: "They were effectively the definition for each territory"sourceas of 2026-11-13
Danger and distance"the farther you get from the safety of town, the more dangerous the land became"; "I rolled all dice in the open, not behind the screen"sourceas of 2026-11-13
Who keeps the record"Let the players take over — Don’t write game summaries, don’t clean up the shared map. You want the players to do all those things."sourceas of 2026-11-13
Public scheduling, and the mixing rule"Require scheduling on the mailing list"; "you can make a rule requiring parties to mix after two adventures", or "require parties to stay together for two adventures"sourceas of 2026-11-13
Town, new players and the main warning"“Town game” was a dirty word in West Marches"; new players were given "a very basic treasure map"; "Competition is what it’s all about"; "Fear the social monster"sourceas of 2026-11-13
Critical Role Campaign 4Covered as West Marches style: 13 players split across three groups at separate tables in one shared world; began 2 October 2025sourceas of 2026-08-15

How we sourced this: Every row but the last quotes Ben Robbins’s own posts on Ars Ludi, read on the date shown. The posts are dated October 2007 to July 2015 and describe a campaign that ended in 2003, so their content will not change, though their addresses could. Nothing here comes from a later retelling of the format. The Critical Role row is repeated from our West Marches page and carries that page’s date; it reports press coverage, not anything we watched or measured.

Related guides

All guides