I grew up playing video games like Morrowind or Baldur's Gate where all the content is already there and all you do is go out there and interact with it, with all the limitations of having a computer DM (e.g. lack of actions having ingame consequences to the rest of the game world, rewarding roleplay, etc.)

Now when GM-ing a Pathfinder game, I feel like I want to provide that kind of sand-box exploration freedom, where the characters could at any time attempt to do anything they chose to. For example when running "We be Goblins" they chased a monster to it's lair and I made checks to see if they got lost, they didn't. But I really didn't have anything to give them if they had gotten lost. Since then I've looked at the maps of the Inner Sea regions, and the Brinestump Marsh and surrounding hinterlands of Sand-point. And it's been using up a lot of my free time, and reminding me that games like Morrowind took years to build, and even if someone already built it familiarizing myself with the world enough that I wouldn't slow down the game too much having to react to the players choice of unexpectedly say exploring the neighbor town rather than go to the prepared forest or stay in the current local, leaving me wonder if it can even be done at all.


Has anyone here attempted to run such a game? Successfully?


I'm not interested in such a game working hypothetically or in theory, only actual experience. Whether that be your own, or with reference (e.g. link to a blog) to someone who reported on actual experience with that.


There seem to be three ways to provide material anywhere you go, one being prepping all areas, the second being having a way to quickly generate content as needed (improv or randomized tables), and the third "playing in other people's sandboxes" is looking up pre-generated content as needed.

For the first I'm worried about the amount of work and the second I'm worried about having to document everything I generated, so they could return to that location later. And the third seems like it would be too slow to do during the session.

Someone linked a question asking what sandbox means, and there @RSConley mentions "many PC groups feel rudderless and the game feels without direction. In fact, if you read through various forums posts, such as on ENWorld, you see these campaigns fail more than succeed." Assuming that these games failed despite having made it past the content generation problem (rather than failed because of these) there seems to be a risk that this kind of play isn't for everyone.

Computer analogy: In Morrowind if you found a good weapon (which is in that spot no matter your level, if you can find a way to get it), or leveled up early, that could make large portions of the game far too easy - so you either end up doing way too easy things or missing content just because your ueber hero is above that kind of quest. Which makes the world make sense. -- In Oblivion the rest of the world levels up when you do, but spends their points more optimized for fighting than you do (cause monsters don't need charisma). Suddenly all robbers are wearing armor that is so expensive that they could sell it and live in wealth for a long time to come rather than attack a known hero. It makes the game mechanically better (i.e. challenging) but the ingame world illogical.

The first world has creature ratings only descriptively as reference (dungeoneering check says you should probably run, this thing is dangerous), while in the second the player's level changes the world around them (independently of their actions), as though CR is prescriptive. Same goes for wealth by level. In a simulationist world you have the money that you got a hold of, however even if you luck out, magic item stores only sell what they have. If you get too rich money is no longer a motivator (happened very early for me in Morrowind), at which point roleplay matters.

In the video game I did it for the out-of-character reason that I'm a completionist who wants to see all the content. What are your character's motives for adventuring? Start bettering the world for free? Donate loads? Or would he just buy a castle and retire? This kind of role play requires having actual motives for your character, thinking about things, rather than just chasing plot for the entertainment of seeing it. Players as consumers and plot as product might be the standard, but I can theoretically imagine for some groups of role-play enthusiasts (or people who love the exploring/freedom aspect of the game) this could work, but it seems really difficult to pull off.

Now what I'm asking is are there any success stories of it actually being done, rather than fantasized about? How did you/they do it?

  • 8
    \$\begingroup\$ By "simulationist," do you mean "sandbox?" \$\endgroup\$ – Alex P Feb 18 '14 at 3:31
  • 1
    \$\begingroup\$ Sounds like you're looking to run Pathfinder as a "hexcrawl." There are a number of games (such as Traveller and the Adventurer Conqueror King System) that support and encourage this style of play, typically by using random content generation to quickly generate content; With a little effort, you might be able to adapt their methods to Pathfinder. \$\endgroup\$ – GMJoe Feb 18 '14 at 3:56
  • 1
    \$\begingroup\$ @AlexP: " one where the GM sets up a world (or at least a small section of one) and the PCs are free to wander where they will and find adventure where they will. It's about freedom of player choice." - Yes I do mean Sandbox. "Pure sandbox play is purely simulation driven. A super hardcore sandboxer places a dungeon (or whatever) in the game world and that's where it is, for the PCs to come across or not (and for NPCs to come across before them or not)." - My question is is that even possible? And if so satisfyingly or does it make GMing a fulltime job? \$\endgroup\$ – Julix Feb 18 '14 at 8:55
  • 1
    \$\begingroup\$ I'm increasingly unsure what your question is especially after the edits. If it's "can you run a sandbox", then yes, obviously, people do that. If it's "can you run a simulationist game" (though I'm not sure you know what that means) then yes... But that's not the problem you're having, is it? You want either content generation or some other thing... What is the problem you have exactly? \$\endgroup\$ – mxyzplk Feb 18 '14 at 12:36
  • 2
    \$\begingroup\$ @Julix The great thing about random content creation in a pencil-and-paper game is that you can freely combine your random content with content you've custom-designed; You can therefore add designed features to a randomly generated location as easily as you can use randomly generated content to fill in the blank sections of a designed dungeon map. If you get the mix right, players won't be able to tell which is which, and you'll be able to save a lot of effort. \$\endgroup\$ – GMJoe Feb 20 '14 at 3:40

While I haven't run a game like what you're asking, I do believe that the "West Marches" experiment is what you are looking for. In short, this is a sandbox campaign that draws inspiration from GTA but takes off the missions. It was run in D&D 3.5, which is quite close to Pathfinder.

The project itself is laid in 5 parts, with the fifth being about creating a "West Marshes" project of your own.

I'll finish with quoting 2 paragraphs from the first post, showcasing the most important elements that were used there.

The game was set in a frontier region on the edge of civilization (the eponymous West Marches). There’s a convenient fortified town that marked the farthest outpost of civilization and law, but beyond that is sketchy wilderness. All the PCs are would-be adventurers based in this town. Adventuring is not a common or safe profession, so the player characters are the only ones interested in risking their lives in the wilderness in hopes of making a fortune (NPCs adventurers are few and far between). Between sorties into the wilds PCs rest up, trade info and plan their next foray in the cheery taproom of the Axe & Thistle.

The whole territory is (by necessity) very detailed. The landscape is broken up into a variety of regions (Frog Marshes, Cradle Wood, Pike Hollow, etc.) each with its own particular tone, ecology and hazards. There are dungeons, ruins, and caves all over the place, some big and many small. Some are known landmarks (everbody knows where the Sunken Fort is), some are rumored but their exact location is unknown (the Hall of Kings is said to be somewhere in Cradle Wood) and others are completely unknown and only discovered by exploring (search the spider-infested woods and you find the Spider Mound nest).

Hope this one helped you a little bit.

  • \$\begingroup\$ Thanks, that's a great example of the kind of thing I'm looking for. +1 \$\endgroup\$ – Julix Feb 18 '14 at 8:58
  • 2
    \$\begingroup\$ Two points of order: 1) It was run with D&D 3.0 specifically. 2) GTA was only used to explain the type of game, and Ben never says that it was actually the source of the game style's inspiration – this is an old style of playing D&D that Ben could easily have been familiar with, and if not, being an accomplished designer, he could easily have re-derived the sandbox RPG structure just by trying to satisfy the GMing needs outlined in the first post. \$\endgroup\$ – SevenSidedDie Feb 18 '14 at 16:44

I wouldn't say that this is Pathfinder specific in any way. If you run a world-spanning game (i.e., not a single quest) then there are two options - you either run your game in an existing setting or make your own.

When I'm running a 'my own setting' game, then what you ask comes pretty much naturally by default - the details of surrounding cities and NPCS can and is generated as needed; and of course it can change very drastically due to impact of players. This has the advantage in that the world dynamics can easily [be made to] happen in a way that supports the narrative of the PCs story. The problem there is that it's quite much effort. You need to prepare content in excess, to have ready locations and encounters that might get visited, but 50%+ of them won't - but there are many articles on how to optimize this preparation (in essence, have a strong overview of the global things, but for the small details/encounters/NPCs have a store of 'fill-the-gap' premade data that you can use whenever/wherever appropriate).

If you're using an RPG-specific setting (such as the D&D Forgotten Realms / Ravenloft / Planescape / etc) with the setting books giving you the content you need - then you can look up the world data from there as needed, and note the deviations caused by your party if they change things that affect multiple locations; but I prefer creating my own, simply because it's easier to make everything fit that way - a premade setting would have an advantage if all the players knew most everything about the setting by reading it, but that doesn't seem to happen in practice.

For a specific example - to answer "ok, you got lost following a bear through these woods..." you need info:

  1. what woods usually contain in the setting - say, in this world, are you likely to encounter magical spiders or fairies, or a Robin Hood-style gang;
  2. are these particular woods special in any way - is there an elf village or a dragon's lair in the middle?
  3. where you will arrive if you go for two days north (for example).

Any general setting will give you answer to (1) + stats for enemies you might encounter; and a local map is all you need for (2) and (3). The rest is just GMs improvisation and roleplay skills. In a hands-off fair simulationist style, roll for the type/change of random encounters; and roll for direction that the party waders off (and look up where they arrive).

  • 3
    \$\begingroup\$ I don’t think this answers the question, because the question is really whether or not there are any settings that are so highly detailed that you can say “ok, you got lost following the bear through these woods, so that means you would be...” and have a definitive answer about what is in that direction and what that’s like. \$\endgroup\$ – KRyan Feb 18 '14 at 4:03
  • 4
    \$\begingroup\$ The whole point is that any setting can provide that for players. The long running settings have extremely large volume of published content including maps of specific cities and the names of NPCs there which would do what you say; but my point is that the simulationist game + definitive answers should be provided for the players - not for the GM, but by the GM+setting. There will always be gaps to fill. The game setting won't roleplay the NPC conversations for you. If you have a regional map - what more do you need to have a definitive answer on what'd happen if you get lost in woods X? \$\endgroup\$ – Peteris Feb 18 '14 at 4:10
  • 1
    \$\begingroup\$ I've done this with D&D 3.5 (which is the same as pathfinder); but as I say I can't imagine what would be different in any other system with published settings. Perhaps there is a confusion about what we mean by "done this". I'm mainly answering about "sand-box exploration freedom, where the characters could at any time attempt to do anything they chose to" in the original question. \$\endgroup\$ – Peteris Feb 18 '14 at 4:35
  • 3
    \$\begingroup\$ In addition, a definitive, simulationist answer of 'what is there' doesn't need any premade content - a statement of '1d4 villages per global map square' + rules for generating random villages (won't quote, but I've seen them in D&D core books) is sufficient and goes well with the simulationist concept. \$\endgroup\$ – Peteris Feb 18 '14 at 4:42
  • 2
    \$\begingroup\$ In v3.5, I had named dots for settlement, a list of NPC names, and a list of NPC types by class and level for quickly assessing the kinds of encounters and resources available to the PCs. It did of course rely on me improvising quickly with this incomplete source, but IMO is the way to go for a practical sandbox game. A little computer scripting helped as well, to speed up generating and organising the source data. \$\endgroup\$ – Neil Slater Feb 18 '14 at 8:11

I see that West Marches has already been mentioned. Good. I find it to be an excellent source of inspiration for my own sandboxing efforts.

As to personal experience, I've previously run a very West Marches-inspired fantasy campaign (even named "Savage Marches" - it used the Savage Worlds rules - and based in the town of Eastmere) and I am currently running another similar sandbox on the fringes of the Warhammer 40k setting.

My main technique for handling the issues you appear to be asking about has been heavy use of random content generation, both up-front and at the table. I have a substantial collection of published products with various random generators, an extensive collection of bookmarked websites with more generators, and, as a software developer by trade, I've written up my own software to assist with this as well.

For the current campaign, as an example, I have a database of 650 star systems with a little over 20,000 total planets and moons, 845 of which are currently inhabited (and another 1,900 have ruins of prior habitation). Of course, the vast majority of them haven't been detailed yet beyond the physical characteristics of the various bodies and broad-stroke details of the populations of those which are inhabited, but there's enough there for me to improvise additional details on the fly if the PCs show up unexpectedly at any one of them.

  • \$\begingroup\$ Slightly mindblown. Here I am worrying about a little area of a slightly larger area and you got it down to >.5k star systems :D \$\endgroup\$ – Julix Feb 19 '14 at 6:39

My approach to this problem (running a simulationist AD&D world called "Thear" for over 10 years), is to work in expanding rings of detail. Yes, it is a tremendous amount of work to detail an Earth-sized planet, but the task is made easier by (a) breaking it up into regions, and (b) using all 3 of the above approaches you named. At any given time, the region the PCs are in is highly detailed with lots of original material, the ring around that region has some pre-written modules and rough notes, while the outermost ring contains little more than vague notions. The world gets fleshed out region by region, as the PCs move around in it. All you need to do is stay a little bit ahead of them. (Granted: This approach won't work in a world with air travel.)

Early Sessions: You totally detail ONE continent or island (this will be where the PCs start), writing lots of original stuff and peppering the map with random elements (option 2) and plugging in prebuilt modules or dungeons (option 3) wherever they make sense. Let's call this "Region 1". You also generate a rough world map containing just the names and major races (no more detail than that for now) of all the continents that would be known to an educated person from Region 1 (which is not necessarily all the races in your world). The exact layout of coastlines etc doesn't even matter at this point; you only need to know which well-known lands are "far east", "just to the north" etc. This will suggest other ideas, like "That must be who we get our lumber from, so there's probably a trade route around here," etc. Note those ideas down as they come, but don't knock yourself out detailing any other Regions yet. Your players can romp around in Region 1 for weeks or months, hearing little pieces of information about other regions but never having any real reason to go there (you can always keep them busy right here).

Later Sessions: While your players have been playing in the sandbox of Region 1, you have been building up the surrounding regions, plugging in pre-written modules wherever they make sense, and making up the basic details of each neighboring culture. But here's the trick: These notes don't have to be exhaustive. They only need to be enough for you to run one rangen (or prebuilt) adventure with a touch of local color in that region, if your players forced you to do it. So by the time your players are ready to explore a neighboring region, you already have a little bit of detail to start them off with. If/when they wander into one of these regions, you refer to your notes, use some rangen tables or a prebuilt for that night's session, and then "make it flesh" after the session (tying in local NPCs, power struggles, etc). Now this Region is fully detailed and the PCs can spend weeks or months in it. Repeat until your world map is complete. It's gonna be a long, long time.


Not the answer you're looking for? Browse other questions tagged or ask your own question.