Sandboxing - story before, story now, story after

[James_Nostack]

In his recent essay "Setting and Emergent Stories" (PDF), Ron includes an provocative afterthought about the term "sandbox play," and I'd like to talk about it.

Here's Ron:

Problematic term: "sandbox"
Recent discussions among the self-described Old School Renaissance have revived and extended the term "sandbox play," but have failed to define it.

As far as I can tell, it can mean anything but railroading, but that means it can include the whole range of Story Before, Now, or After, and the whole range of setting use from barely-any to all-encompassing.  Which makes it pretty hard to talk about outside of a given group's actual play experience. The term is also completely unconstructed regarding the size of a sandbox or if it's supposed to have a size relative to the whole setting, regarding how changes to the setting procedurally occur.

In other words, the term means almost entirely nothing, and I think it's kind of a shibboleth based on romantic notions of "Gygaxian play" (another everyone-knows but can't-define term) and imagined notions of what it's like, or must have been like, to play The Keep on the Borderlands.

I've been pretty active in this particular scene, and I agree with Ron on one point, which is that "sandbox" could use some scrutiny.  So: a thread to do that! 

A Quick Conditional[/u]
I'm writing about my own play, and I encourage you to write about your own.  To the extent my wishes matter, what I absolutely do not want in this thread is (a) a denial of anyone's recitation of first-hand experience unless you were there, (b) saying that someone's play-preferences are wrong or irrational, (c) a first post that assumes a counter-factual and then draws an elaborate hypothetical from it. 

What I would like is for people to share their actual experiences, to ask (and be asked) questions about it, and to see if there's a way to identify things that work well or poorly, or discrete techniques or features of play that need more real-world research. 

(My own wishes are simply my personal preferences, of course, but I want to be up front about them.)

Actual Play "Certification"
For the past 3+ years, thirty of us have been playing in a self-identified "sandbox" style of play at the New York Red Box. 

We have been playing the hell out of early-edition Dungeons & Dragons.  We've logged about 142 sessions across two campaigns:

• Tavis's White Box Game, 42 sessions of 1974 D&D (0e) plus eclectic supplements, approximately monthly.  The setting is extremely gonzo, based on 1970's sword & sorcery and science-fantasy paperbacks.  I've played in about 20 of these sessions.
• Eric's Glantri Game, 100 sessions using the 1981 Basic Rules (B/X), approximately weekly.  The setting is fairly vanilla, lifting from Clark Ashton Smith's Averoigne in places.  I've played in about 20 of these sessions.
• When we first got started, I ran six sessions of B/X in a self-consciously generic "don't bother me, I'm taking the bar exam" D&D fantasy world.  I'll discuss its dissolution in a minute.

Beyond the D&D campaigns, we've got an aggressive program for one-shots and limited runs, including: Sorcerer, Trollbabe v1.0, Trollbabe v1.1, With Great Power..., Champions, Marvel Super Heroes, Mouse Guard, Burning Wheel, Apocalypse World, Starships & Spacemen, Tunnels & Trolls, RuneQuest, BrickQuest, Call of Cthulhu, and Traveller.  Lately we've been playing a lot of Pendragon, which has been hilariously mock-heroic.  I have played in virtually all of these "limited run" games.

Demographics: there are about 35 of us, maybe 15 of whom are regulars.  About 90% of us are dudes, and the overwhelming majority of them are mid-30's to mid-40's.  There's a decent amount of camaraderie among the regulars: movies, pizza, drinking, dinner parties, wedding invites, art shows, etc.  With a group this big and this long-running, there's some interpersonal friction from time to time, but generally we all get along.

Sandboxing as Used in NYRB: Early Days in the Black Peaks[/u]
The Red Box group was an outgrowth of the NerdNYC social scene and Ben Robbins's West Marches style of just-in-time, no-fixed-group gaming.  As Robbins explains it:

There was no regular plot: The players decided where to go and what to do. It was a sandbox game in the sense that's now used to describe video games like Grand Theft Auto, minus the missions. There was no mysterious old man sending them on quests. No overarching plot, just an overarching environment.

My motivation in setting things up this way was to overcome player apathy and mindless "plot following" by putting the players in charge of both scheduling and what they did in-game.

So the very first few sessions followed Robbins's model pretty closely. 

There was no particular story at work beyond the one loosely assumed by D&D play: dungeons + dragons + adventurers in approximately medieval drag.  I used the Grand Duchy of Karameikos, the default setting as presented in the Cook/Marsh Expert Rules, and seasoned it ever-so-lightly with Big Picture Situation (evil Dragon enslaved, its army regrouping, various forces of Law, Chaos and Neutrality trying to exploit the power vacuum).  My goal at this point in time was just to play maybe 3 casual sessions; I wasn't trying to do anything too fancy.

I drew a map of a mountain valley and placed some small dungeons around the area.  Some of the dungeons and local NPC's had connections that would maybe come to light, or maybe wouldn't, and a couple of the more feisty NPC's were busily pursuing their own agenda, occasionally meeting the PC's and usually regretting it.  Players were free to drive whatever agenda they wanted in the setting.

Over a couple of months, my interest in running a long-form D&D game dwindled to zero, for a couple reasons.

• 6 sessions is really too short for a D&D campaign to really take on much of an identity or for characters to take on a life of their own.  I didn't know that at the time.
• The setting was so generic that it didn't pique my interest sufficiently as DM.  I'd just latched onto it as a time-saving device, without any regard to long-form play.
• My Big Picture Situation all involved NPC's of approximately 6-9th level.  In early-edition D&D play, that is literally years off before you can deal with it as equals if you start at 0 XP as we did.
• Also, the geographic scope of the sandbox's Big Situation was too wide.  Everything in the mountain valley--the hometown, the dungeon, the Dragon Army's remnants and the Dwarf diaspora--was solid, functional, and close at hand.  But the Duke and his court politics was a good 200 miles away; the evil Bishop was even further away.  So we'd be dealing with their proxies, henchmen, and so on.  Bah, fuck that.  I'm a sorcerer, I only deal with principals.  (There was no reason for me to make things so far away, except that I just grabbed that published map as a way to save time prior to a session, without thinking of long-term effect.)
• I'd started a new job that prevented me from running both D&D and indie stuff, and so had to choose.

But also, and very importantly, early-edition D&D simply does not have very good procedures on stuff that doesn't involve dungeon-delving.  In particular it does not have good rules on how the campaign world changes as a response to player action.  As Ron puts it:

[N]ow for my real point: [mechanics equivalent to the elaborate rules for how characters change over time, i.e., advancement] for the consequences upon setting are rare. . . . Whereas for setting, the historical default is for there to be little if any such things, so that's what we see across the games today.

Now, the absence of those rules is pretty much universally regarded in the wider OSR as a design feature rather than a bug.  Like role-playing the demons in Sorcerer, the long-term evolution of the campaign world in Dungeons & Dragons is exclusively the DM's province--not even the rule books get in the way.  It's a very personal expression of the DM's creativity and outlook.

But it's also a pain in the ass if you're trying out a "new" game without a lot accumulated knowledge.  Dungeons & Dragons at low-levels is so unbelievably cruel, you don't want to make the world so adversarial that the PC's lose heart.  On the other hand, you want to always make the world . . . "interesting," in the sleep-with-one-eye-open sense.  That's a fine line to walk, and I would have appreciated a crutch.

So that was it for the ol' Black Peaks game.  Then the White Box and Glantri games got started, and I'll talk about them in the days ahead.

Value Added[/u]
Cutting to the chase a little bit, I think these procedures work pretty well, not necessarily in this order:

• Create an environment, not a plot.  A wilderness map, a dungeon, whatever.  It should be big enough that the PC's can, with sustained effort, exhaust it, but not much bigger.
• Find a couple of zones in the environment that interest you.  Who's there?  What's something interesting going on there?  It's probably best to think about these things individually, and then maybe lace them with some, but not a whole lot, of connective tissue.
• Give the players a home base.  Put some NPC's in or around the home base, who want things.
• Plug the players into the setting.  I.e., some Cleric belongs to a church that wants ________.  A Magic-User is an apprentice to a Wizard who fled from _________ which pursues him.  A Thieves Guild is enmeshed in a trade war with ______.  These are specifically D&D examples, but every game has something like this.  These things should be "colorful" in the Forge sense of the term, and act as "invitations to a situation."  You're doing this only so that the characters have some minimal roots in the setting.  The players may do this for you!  If so, don't get in their way!
• Don't be coy about sharing rumors/gossip/news.  Help the players get oriented on things to do here.  You want a list of like 3-7 loose ideas for them to work with.
• If the players go for some of your suggested content, awesome, flesh that one out ahead of time and run with it.  When you flesh it out, it should be like a house of cards, and which way it falls depends on the players' (witting or unwitting) actions.
• If the players start doing their own thing, cool.  Hopefully your game gives you enough tools to generate mostly-meaningful random content on the fly.  Just let them get deeper and deeper into a weird mess, and if they have an easy time of it, so be it.  By next session, figure out an evolution of the prior content that bites them in the ass (not as punishment for going off-reservation, but as, "Hey, I know that looked random and you're feeling chuffed--but guess what, payback's a bitch.")

Don't hold me to any of that, because I'm tired of writing, but something akin to those elements has proven very workable.

I wanna talk about White Sandbox, Glantri, and failed attempts at Story Before Traveller and Story Now Marvel Super Heroes, but this post is too damn long.

[C. Edwards]

The procedure summary at the end of your post is pretty solid in my experience. Just adhering to an Apocalypse World like insistence to "make the world seem real" will get you through half the battle. Cause and effect, action and reaction, in the short and the long term is at the core of sandbox play in my opinion.

My introduction to rpgs was using the Moldvay Basic Set (pink box) to run module B2: The Keep on the Borderlands when I was thirteen. While not being explicit in many ways regarding technique and procedure, I think it made a pretty decent road map for learning how sandbox play operates. B2 contains many, if not most, of the elements you mention and there are brief guidelines concerning the way many of the factions and individuals may react to interference or other actions of the PCs and how the area may change over time.

I should probably note that no one taught my friends and I how to play nor was D&D anywhere on our radar as a cultural thing (except maybe for the Dungeons & Dragons cartoon, which I had no idea at the time had anything to do with an actual game). This was in 1988 and for us "gaming" could only mean playing Nintendo. Which is actually the reason I discovered D&D at all. While a friend and I were digging through every nook and cranny in his room looking for the RF adapter for his Nintendo I came across this odd pinkish-purple box in a drawer where it was buried beneath a bunch of junk. My friend had put it there, unopened, after his grandmother had given it to him and basically forgot about it. He let me take it home to check it out and a few days later we were playing.

As for "sandbox play" as a term and an idea I think Ron is mostly correct, but that's because I don't think the basic elements and techniques of sandbox play have much of anything to do with things like setting size, how or when story is created, or set and standard procedures for setting change. My experience is that the core of the whole idea is "make the world seem real" and "let the players decide what is important". Those are your primary priorities as a sandbox GM. The rest is mostly a combination of flavoring to taste and the proclivities of the players. Oh, and improvisation. Lots of improvisation.

I'm looking forward to your further posts, James.

[James_Nostack]

Thanks C.!  I hope to talk about my own experiences with Keep on the Borderlands in a couple of posts' time.  It's worth discussing, but I want to do a data dump too!

Phase Two: The Glantri Game
As my own Black Peaks game wound down, one of my players, Eric, began his own campaign, set in the Principalities of Glantri, a canonical piece of the Basic D&D sample setting.  It was introduced in a single paragraph in X1: The Isle of Dread, written by Cook and Moldvay, which shipped with the Blue Expert rules just as B2 shipped with most of the Basic rules.  As introduced here, Glantri is a bunch of minor princedoms ruled by a magocracy--sort of like the petty wizard kingdoms you'd expect in The Dying Earth.  (Curiously, Eric had no knowledge that Glantri had been turned into a 100-page setting book late in Basic D&D's product lifecycle, and I believe he disregarded all of it.)  Eric described the Principalities as decadent wizardlings recovering from a magical civil war that left much of the place in ruins, with a culture similar to medieval France.

As a player, I don't have a lot of insight into how Eric set up his sandbox, though I suspect he was thinking along the lines I outlined above.  From a player's point of view, at first we had a town, and a nearby ruined Wizard's Tower.  Eric hinted at some other options, but we knew the Wizard's Tower a little bit, so we wanted to leverage that knowledge for profit.  Sometimes the setting's evolution resulted in classic stories that became legendary in our group.

Sometime around Session 12, things got busy for me and I stopped coming to the weekly games, and there was almost a complete turnover of players.  The new guys have now played for 90 sessions--sometimes dropping out, sometimes dropping back in--and have built up impressive accomplishments that I don't totally understand.  Major locales in the game involve the Keep on the Borderlands, Quasqueton, and Eric's home-made "mega-dungeon" the Chateau d'Amberville (no relation to X2).  The players are free to move between these locations, often deciding their next moves based on how hot things are in town. 

One recent subplot involved a Throne of Gender-Bending and some kind of curse that shrank a Fighter into a Halfling.  This led to an epic overland and nautical journey to find a goddess who could grant a wish, while other allies stayed behind and got mixed up in political intrigues relating to the mega-dungeon.  These plotlines converged recently, and for Session 100 the party was finally reunited. 

Eric's game has a reputation.  Low-level D&D, by the book, is brutal as hell.  (Look up the stats for the Killer Bee!  Who the hell designs something like that?)  But Eric, as DM, is often very adversarial.  I suspect this is an instinct from Story Games, where for a while many games encouraged players to look for the conflict of interest in a scene and play toward that.  Frankly, when I'm kicking back in a D&D town, I don't want to take guff from the bartender, when I've just delivered the carcass of a 6 HD crocodile-demon that can be hung over the bar--that's not the time for the dude to deny me a free drink.  The upside, however, is that the dudes who have stuck it out are men.  They are shivering in the cold and rain, amid the howling of wolves, telling each other encouraging tales of the White Boxers zipping between dimensions to argue with Death.

Value Added
So here's a thing I've noticed in my own game, and in both of the longer-running D&D games.  Despite being ostensibly "go anywhere" games, players want to go to the same dungeon, over and over again.  In the early Glantri game, we could not get enough of that Wizard Tower.  The n00bs (who have been playing like 9 times longer than me by this point) usually can't get enough of the Chateau.  Same's true in Tavis's 0e game.  I think what happens is that the most dangerous thing in D&D is ignorance.  Once you know something, you can exploit that knowledge if you're clever enough, so there's this feedback loop that encourages repeated delving.  Once the looting begins to peter out, people get anxious to find richer plunder.

Theory Wrangling[/u]
C. Edwards wrote:

Just adhering to an Apocalypse World like insistence to "make the world seem real" will get you through half the battle. Cause and effect, action and reaction, in the short and the long term is at the core of sandbox play in my opinion. . . . My experience is that the core of the [sandbox] idea is "make the world seem real" and "let the players decide what is important". Those are your primary priorities as a sandbox GM. The rest is mostly a combination of flavoring to taste and the proclivities of the players. Oh, and improvisation. Lots of improvisation.

I am not sure that I would phrase it that broadly.  For me, "make the world seem real," free will, and fictional cause-and-effect are practically the definition of imagining yourself as another person in another place.  To use Forge jargon, I think it's the essence of Exploration itself.  I hope to talk about a Traveller game, railroaded to hell, where the inability to explore beyond the railroad pretty much prevented me from playing the game at all.  (I think in terms of Ron's essay, this type of Story Before play stomped all over my free will and complicated the act of playing.  I realize this can be finessed, but usually by getting the players to agree that free will isn't all it's cracked up to be, at least on important issues.) 

So I think sandboxing is more than just "Exploration" in the Forge sense.  In theory every RPG is delivering that if it's functioning at all.  I suspect sandboxing is a particular technique within that broader framework.  More at some later time.

[C. Edwards]

So here's a thing I've noticed in my own game, and in both of the longer-running D&D games.  Despite being ostensibly "go anywhere" games, players want to go to the same dungeon, over and over again.  In the early Glantri game, we could not get enough of that Wizard Tower.  The n00bs (who have been playing like 9 times longer than me by this point) usually can't get enough of the Chateau.  Same's true in Tavis's 0e game.  I think what happens is that the most dangerous thing in D&D is ignorance.  Once you know something, you can exploit that knowledge if you're clever enough, so there's this feedback loop that encourages repeated delving.  Once the looting begins to peter out, people get anxious to find richer plunder.

This is my experience as well, and I think it helps add some predictability and manageability (for the GM) to the otherwise free-roaming nature of sandbox play. Curiosity and the desire to overcome a particular challenge may also account for part of this behavior. I've always associated this incremental player knowledge gain with platformers like Megaman. For every time you lose a life you learn something new about your environment and how to interact with it successfully. This knowledge results in you being able to go farther each time before losing a life. Eventually you learn enough to make it to the big boss. Beating the boss may take several cycles of life loss and knowledge gain. That association is probably part of the reason that the lethality of the older versions of D&D doesn't bother me.

C. Edwards wrote:

Just adhering to an Apocalypse World like insistence to "make the world seem real" will get you through half the battle. Cause and effect, action and reaction, in the short and the long term is at the core of sandbox play in my opinion. . . . My experience is that the core of the [sandbox] idea is "make the world seem real" and "let the players decide what is important". Those are your primary priorities as a sandbox GM. The rest is mostly a combination of flavoring to taste and the proclivities of the players. Oh, and improvisation. Lots of improvisation.

I am not sure that I would phrase it that broadly.  For me, "make the world seem real," free will, and fictional cause-and-effect are practically the definition of imagining yourself as another person in another place.  To use Forge jargon, I think it's the essence of Exploration itself.  I hope to talk about a Traveller game, railroaded to hell, where the inability to explore beyond the railroad pretty much prevented me from playing the game at all.  (I think in terms of Ron's essay, this type of Story Before play stomped all over my free will and complicated the act of playing.  I realize this can be finessed, but usually by getting the players to agree that free will isn't all it's cracked up to be, at least on important issues.) 

So I think sandboxing is more than just "Exploration" in the Forge sense.  In theory every RPG is delivering that if it's functioning at all.  I suspect sandboxing is a particular technique within that broader framework.  More at some later time.

You're right, that's some pretty broad phrasing. Let me see if I can narrow down what I mean. In my mind it's like the difference between filming a movie on location or filming it on a set. On location: during play you know there is a wider, functioning world out there even if you never make an effort to learn about it. On set: during play you know that straying too far outside the frame of what is relevant to "the adventure" (even if you don't exactly know what "the adventure" entails) will reveal the limits of the facade that confine your character. It's not so much about exercising the freedom of going anywhere or doing anything as it is about simply having that freedom. It's the freedom of all your choices as a player being relevant. Maybe not exciting, maybe not smart, maybe not dramatic, but relevant. That can only happen if the GM is committed to portraying a (hopefully) complex world no matter where the PCs go or what they do, instead of creating an adventure path for the PCs to trundle down like a blind man with a cane.

I think the idea of sandbox play sounds overwhelming to a lot of people but, as you've pointed out, there are emergent properties that tend to confine the bounds of play. Throw in some useful procedures or techniques and away you go.

[Abkajud]

I always took "Make the world seem real" to mean "The world must have its own logic; when the GM declares that X occurs, it must appear that there is good reason, or at least a reason, for X to occur, rather than X coming out of the blue or feeling contrived."
It goes back to being able to trust the GM to faithfully referee your game - is zhe just cribbing from some hidden notes and forcing plot to happen because zhe came up with it on hir lunch break on Friday and is now invested in seeing it realized in play? Or is zhe just taking notes on play as it happens, and thinking about how to introduce these elements again later?

This also goes back to issues of whether a pre-made plot is acceptable to the group, or not. Pre-made stuff can definitely make the world feel more complex, but in my experience only a really skillful GM can take something devised in isolation from actual play and make it feel alive. I'm not talking about having blurbs or writeups for locations along the course of the adventure, or even having ideas and stat-blocks for encounters in the adventure. I'm talking about being able to use such elements independent of the passage of in-game time, of the course of in-game events, etc.

In Apocalypse World, "Make the world seem real" is about gaining credibility with the players so that when you pull a Hard Move on them, they aren't upset and confused at this turn of events, but rather they say, "Yeah, haha, I should have seen that one coming!" Best of all, you don't have to be a master MC to pull it off - the Moves help you become good at MCing.

[James_Nostack]

Phase 3: White Boxin'
In early 2009 as Eric's Glantri game began picking up steam, a new player named Tavis began running a monthly "0e" game.  The major framework of play is the 3 Little Brown Books of 1974, but we've got a smattering of 1e, Arduin, Basic, and the occasional bit of 2e or 3e thrown in for good measure.  In keeping with the eclectic rules, Tavis's setting is a kitchen-sink 1970's DAW paperback world, which I've taken to calling "acid fantasy."  Our map, with 40 sessions of work, is here; much of the action for the first 30 sessions concerned the Lost City, based on Paul Jaquays's classic 1978 module, The Caverns of Thracia.  (The fussy notations on the north of the map reflect recent stuff I've missed out on.)

Tavis's game is by far the most open-ended campaign world I've ever been in.  The Black Peaks game and the Glantri game were open-ended, in the sense that Eric and I were committed to following along with player initiative.  But I, and possibly Eric too, don't handle improv especially well: I can do it but it's work sometimes.  Tavis, on the other hand, has mastered and internalized about a zillion random tables, and knows his inspirations so well that even if some detail isn't prepped he can infer it.  On the other hand, Tavis tries to save effort by incorporating published modules, typically old Judges Guild stuff which tends to adopt the same acid fantasy aesthetic, as the main focus of play.  (I'm not sure we've ever played in a Tavis-designed dungeon.) 

This has been a lot of fun to play in, but we don't always do as much playing as I would like.  Taking Gygax's suggested number of players from Men & Magic into account ("At least one referee and from four to fifty players can be handled in any single campaign, but the referee to player ratio should be about 1:20 or thereabouts"), Tavis plays with a very large group, often 10-15 people.  This makes it almost impossible to reach consensus, particularly when the more risk-averse long-timers start to fret over the optimal way to handle a situation.  Typically this group plays for 6 hours a month, the first 2 hours of which are spent deciding what to do, and then up to 1 hour prepping to do it.  (I would purposefully arrive 3 hours late, ready to rock.)

So that's one peril of extreme laid-back sandboxing "you tell me what you're doing" style of play: if the Dungeon Master takes no social control over the deliberation process, you can end up with analysis-paralysis situations.  Tavis doesn't see it as his role to ramrod people.  Rather, the group will organically evolve its own procedures for reaching decisions that are optimized for the preferences of the regular players.  People who don't like it will self-select out, as I largely have.  (I don't agree with this, but it seems to be philosophy.)

Part of the reason decision-making takes so long is that there are so many choices, so many character-driven priorities, and so many setting-ramifications to think through.   Taking the planning process on-line didn't help (warning: 8 pages of hand-wringing bullshit.)  There can be a point where the setting is too complex, or at least, where the "adventuring party" is so heterogeneous in its desires that it's hard to figure out how to interact with a highly responsive setting.

However!  Once play begins, and everyone knows what they're supposed to be doing, it's a very solid, entertaining time.  I am not sure if this qualifies as "Story After" play as Ron's defined it.  For me, there are about 3 hours of greatness in a 6 hour social event, but really, only about half of those 6 hours are spent in play; the rest is more like lightly-RP'd fussing over the menu.  I don't find that entertaining, so I'm not considering it part of the greatness; but it keeps happening so I presume it serves some purpose for others.  The delving is fun, action-packed, and almost always entertaining in the moment; we're not revising stuff, other than to chop out the tedious "tune up the orchestra" phase of play. 

"Fuck the Adventurer"?  Fuck You!
In his essay, Ron writes:

I will now provide a set of concepts and practices to bring out what seem to me to be these games' best
features for setting-centric Story Now play. The idea is to embrace the setting as a genuine, central
source of the colorful thematic dilemmas explicit in the games' introductory text, and to resist the
retraction and retreat to comparatively tame Story Before which are explicit in the later GM-advice and
scenario-preparation text. . . .

[snip]

Make player-characters in it. In doing so, drive this into your brain: fuck "the adventurer."
• Not all types of characters described in the character creation options are OK. They need to be
characters who would definitely be at that location, not just someone who could be there. They
have something they ordinarily do there, and are engaged in doing it.
• All characters, player-characters too, have lives, jobs, families, acquaintances, homes, and
everything of that sort. Even if not native to that location, they have equivalents there.
• Player-characters do not comprise a "team." They are who they are, individually. Each of them
carries a few NPCs along, implied by various details, and those NPCs should be identified. It is
helpful for at least one, preferably more of them to be small walking soap operas.

I see what Ron is saying here: in order to really dig into the setting the way cleats dig into a soccer pitch, the adventurer can't be some nameless, rootless wanderer with no interests, friendships, or whatever.  This may be true in games aiming for Narrativist goals, and I can certainly see an argument that even when it's not strictly necessary it should be best practice.  But I think it overlooks some stuff, at least specifically in the contest of historically revisionist OSR D&D play. (Of course, that style of play isn't explicitly aiming at Narrativism, so maybe it's not a counter-example after all.)

Judging from our own extensive play, and about a zillion blogs recounting the same experience, D&D characters enter the world as six attribute scores, a class, an alignment, and a name.  That's not a lot to go on!  Maybe the player comes up with a half-ass backstory that the DM integrates into the setting, or not.  But D&D characters are passionately, tragically, achingly plugged into the setting: they are all drug-addicts, fiending for gold like Bogart in Treasure of the Sierra Madre. 

On first arriving anywhere:
* Who's got the most gold?
* How well is it guarded?
* What are the reprisals?
* Who will benefit from us robbing this guy, and therefore might shelter us?

For example, in Tavis's White Box game, we've got a pretty good idea that the Necromancer has a ton of treasure.  But he also keeps level-draining Wraiths trapped in crystals to throw as grenades against intruders.  We're terrified of level-drains, and don't have enough high-level Clerics to feel confident against a large number of Wraiths, so we have left the Necromancer alone until very recently.  In the Glantri game, people quickly realized that B1: In Search of the Unknown had really shitty treasure, so we kicked that place to the curb and only go back when we're
You can even draw up a list of expected treasure values, and then if you're really bored, try to weight that against the expected difficulty to figure out the juiciest targets.  The point is that "D&D adventuring" (i.e., drug addiction) is an immediate hook into the setting.  (When "adventuring" isn't synonymous with drug addiction, as in Alternity or, I don't know, GURPS, I think Ron's points acquire greater force.)

Additionally, Tavis's game features "carousing rules" based on Jeff Rients's house rules (which in turn were based off concepts from an old Dragon article, "Orgies, Inc.").  The idea is that you're a dude with a lot of gold burning a hole in your pocket, and you don't earn XP until you piss it away--and the process of doing so might get you into trouble leading to further adventures.  This can be seen as a mechanical way to generate the friends/enemies/relationships that Ron mentions in his essay, plowing the reward system into generating new predicaments.  Eric has adopted similar rules, though I think the mechanics differ.

But I also think that this stuff happens over time, even with PC's who don't have roots.  It takes a while--probably 5 sessions or so--but in that time, rivalries and so on will naturally spring up.  This assumes, of course, a long period of play. 

[Eero Tuovinen]

The topic is very dense and I've got real world stuff to focus on, but I'm reading. Also, this:

So that's one peril of extreme laid-back sandboxing "you tell me what you're doing" style of play: if the Dungeon Master takes no social control over the deliberation process, you can end up with analysis-paralysis situations.  Tavis doesn't see it as his role to ramrod people.  Rather, the group will organically evolve its own procedures for reaching decisions that are optimized for the preferences of the regular players.  People who don't like it will self-select out, as I largely have.  (I don't agree with this, but it seems to be philosophy.)

Our campaign is just like this, except it's only been going for around 40 sessions so far. Great fun, and as the GM I can underwrite the notion of letting the players coordinate themselves. The planning bit is quality entertainment in itself, we often have similar 20-50% shares of the session dedicated to doing party logistics, sorting information and deciding on the next adventure and its parameters. I read that planning thread just now, would love to play in that group!

My advice to players who think that we're getting stuck on minutiae has been boldness and initiative: seize leadership, make some snap judgments and get the party off into the dungeon - the GM doesn't need your planning, he's quite happy to run the game from the moment we sit down at the table. Dying might be losing, but it's not failing in the ultimate agenda of having fun.

That wiki inspires me, I should see about moving the planning process for our campaign to the Internet at some point. Something light-weight to begin with, the teenagers aren't often too sorted with web community tools.

[James_Nostack]

Eero wrote:

we often have similar 20-50% shares of the session dedicated to doing party logistics, sorting information and deciding on the next adventure and its parameters. I read that planning thread just now, would love to play in that group!

Wow.  That thread was killing me.  I think there's this trick of parliamentary procedure where, if you just go around in circles long enough, weaker-willed participants will eventually give up: "Just make a decision, I don't care so long as this meeting ends!"  Sometimes it feels like that to me, but obviously other players in that same thread loved it.  But I think it's all sorted itself out, and everyone is happy.

Thinking about this a little more, about how to define a sandbox.  Perhaps a sandbox is a medium for play?  You've got a world, and within that world there are these dungeons where a lot of any particular session occurs.  But when the session is not occurring (both literally, as in you're away from the table, but also when you're just bullshitting about what to do next), you've stepped back one level and are inhabiting the sandbox. 

So the sandbox has some descriptive qualities: "The Lost City of Thracia contains Ontussa the Sphinx, who was enslaved by the Necromancer Ashur Ram."  Right?  Stuff to see and things that can be done.  Backstory.  Ancient history.  Connective tissue between the dungeons, and between a dungeon and a town.

But it also has a procedural or prescriptive component: the sandbox is what the players choose to do, in the order they choose to do it.  The descriptive stuff in the setting (the map of Scather's Hoard, the politics of the Nameless City) largely comes into existence in response to players' meanderings, both as preparation and response. 

I'm setting up to discuss B2: Keep on the Borderlands a bit, but that also segues into historical discussion.

[C. Edwards]

The planning aspect of play I tend to think of as "The Hannibal Factor", as in the A-Team, as in "I love it when a plan comes together." It's not something I mind, and I think that the creativity and imagination that often goes into planning and a positive return on that planning are part of the draw of play. Part of the implicit reward cycle I guess. But then I wouldn't run a game for ten of fifteen players at one time either. That's just not something I have any desire to do. Either tweaking the rules, or using rules that are less likely to hand your ass to you would probably cut down quite a bit on how much planning players feel is necessary.

James wrote:

Thinking about this a little more, about how to define a sandbox.  Perhaps a sandbox is a medium for play?

That makes a lot of sense to me, and would explain why at a higher level (including the factors Ron mentions in his essay) individual examples of sandbox play can vary so much.

[Abkajud]

Wasn't "sandbox" a direct response to the old way of having the players on a rail straight for the dungeon?

In my experience, people sign up to play a game of D&D on the basis that they're going to be led straight to the adventure and walked through it in a sort of slightly-detached "tour guide" kind of way. This is not as much the case with games of 0e, as it seems the OSR crowd has encouraged people to play early editions with more of a daring, explorative flair.

As I recall it, and as it's been used in MMORPGs, a "sandbox" is a game setting that you're allowed to explore, without being told what the plot is, by the GM. The level of prep can be high or low; the only necessity is that there is a means of prepping or improvising the details of the setting as they come under scrutiny through play. It doesn't strongly correlate to Creative Agenda per se; the motivations for playing a sandbox game can vary greatly.

[David Berg]

I feel that the groups with whom I've used the term "sandbox" have meant the same thing by it (which may be my fault):

Player choice about which large-scale endeavors the character will undertake out of multiple viable options.

My fellow players who have expressed enthusiasm for "sandbox" have mostly been enthusiastic about (a) choosing the situations of play, and (b) doing so in character (or at least guided by their characters).  "Here's what my character's like and what he wants, and he's going to assess his environment and decide what's the best way to pursue that.  But, uh, I don't want to spend forever on this, so, GM, give me some coherent options, would ya?"  In my experience, this attitude extends throughout play -- the same players who want to pick their dungeon also want to pick whether to complete their chosen dungeon now or after leaving to fetch more supplies. 

I believe this narrows down the field somewhat in terms of Story When and dictates a few details of what constitutes an appropriate setting.  Sandbox Story Before is an extremely tough challenge for the GM.  Story Now usually seems to come from a different player mindset than what I've described; not that it can't be sandboxy, but, well, I haven't seen it.  Story After, on the other hand, is an easy and natural fit in the sandbox.  As for setting, the ideal is a quest-generator keyed into the characters' desires.

So, there's my data point. 

James, I intend to read everything you've written here and respond at some point!  I dig the topic.  Just kinda busy right now.

[David Berg]

...and, I can't resist chiming in on the planning meetings.  I've found the line between feature and bug to be paper thin.  I love an energetic review of past achievements, present options, future goals, and a collaborative creation of an exciting way forward.  I hate aimless speculation and accounting.

Two systems I was about to test in Delve before my move disbanded our game:

1) First person to get tired debating what to do plops a token on the table and gives a proposal for what to do right now.  Anyone else who wants to propose anything else (including continuing the discussion) must match the ante.  This goes around the table in rounds until everyone gives but one, and that one person gets there way.  Tokens would have been accrued by some related performance -- e.g. coming up with good ideas of stuff for the group to do.  This method applies most to strategy/tactics sessions with minimal but non-zero time pressure.

2) Keep a to-do list 6 adventures long.  At a meeting, each player rates each adventure in order of enthusiasm and urgency from 1 (min) to 6 (max).  From these, 3 adventures are considered: highest Enthusiasm, highest Urgency and highest Total.  Each of these 3 is then rated as "actionable right away" or "needs more information".  A simple vote is taken on whether to (a) pursue more info (usually in a fast-forward or with some divination power) before deciding or (b) cross off the "need more info" options.  Vote on the remaining options.  This method applies most to late-campaign situations where there are dozens of possibilities for the party.

I'd link the actual worksheets I made for these, but I can't find them right now...

[James_Nostack]

Wasn't "sandbox" a direct response to the old way of having the players on a rail straight for the dungeon?

The way the question is worded, I can't tell the context you're talking about.  Certainly in 2011 this style of play is a rejection of the last 25 years or so of mainstream adventure/scenario design.  Whether this style of play circa 1975-1985 was a rejection to similar railroading, I don't know--I wasn't gaming back then--but I would tend to doubt it. 

If people want to discuss the early history of D&D with respect to this type of play, preferably from actual experience, I'm cool with that.  But it's not my primary focus.  If it turns out that nobody played this way in 1974 ("sandbox," as a term, appears to come from relatively recent video games, so maybe the idea behind it is brand new), it wouldn't change the fact that in 2011 these procedures work reasonably well. 

B2: Keep on the Borderlands in Actual Play
Because Ron singles this one out, I thought I'd talk about it a bit.  Players in our gang like to compose session summaries.  They're listed here, mostly in the Ruined Vale section.  More concise summaries are here, around Sessions 8-14 + 21.  In a single sentence: we found the inhabitants of the Keep insufficiently deferential, befriended an evil wizard, and suffered mightily in the Caves of Chaos before giving up.

I participated in most of these sessions.  As experienced, Keep on the Borderlands is a pretty good model for sandbox play.  You've got:

• a base of operations, the Keep, swarming with NPCs.  (Curiously, the authors give them hit points but not names.  Stop giving us bad ideas, Gary!)  By default the Keep is a safe, friendly place, but we briefly discussed whether it made more sense to sack the place rather than risk our lives in the dungeons.  (Answer: no.  The problem with Lawful people is that they are extremely organized.)
• Surrounding the Keep is a wilderness, about as far as people can explore in a few days.  In that wilderness are several different zones, some of which are linked to rumors provided to the players.  Eric used one of these rumors to introduce B1: In Search of the Unknown to one portion of the map, which apparently is pretty commonly done.
• The main action can be found in the nearby Caves of Chaos, occupied by a zillion tribes of monsters.  As we experienced the module, some of these tribes are better organized than others, and the evil Wizard who's apparently running the show is very reactive, taking precautions as needed.

It was reasonably entertaining, but hard as hell, the rewards were pitiful, and (for me) the evolution of the situation wasn't always grabby enough--I know the situation wasn't static behind the scenes, but it sometimes felt static on-stage, so to speak.

B2: Keep on the Borderlands as Teaching Text
Ron's essay is mildly critical of deriving ideas about sandbox style play from "imaginative notions about what it's like, or must have been like, to play Keep on the Borderlands."  I'm not sure how many people are doing this--it is perhaps the most widely played module in the history of the hobby, not only on original publication but especially within the revivalist community.  But even if people are simply theorizing, I don't see it as inappropriate.  B2: Keep on the Borderlands is extraordinarily precise about exactly how it is to be used.  In the early days of the hobby, designers' intent has rarely been communicated more clearly.

Throughout its early history, Dungeons & Dragons was pretty much horrible at telling people how to actually, y'know, play the game.  0e was a disorganized mess and successful play seems to have been passed along by word-of-mouth.  In 1979 AD&D came along and really wasn't much better.  There's an extended example of play on pages 97-100 of the DMG, but not very much about what campaigns are actually like in play.  (Lots of information on the world-building aspect of campaigns, though.)  There are some snippets in the 1e DMG suggesting a sandbox style of play, particularly on pages 86-87, but it's far from explicit.  The 1978 Holmes Basic rules don't really address the topic, and neither does the 1981 Moldvay Basic rules, though it does have some good procedures for designing a dungeon.

So B2: Keep on the Borderlands is pretty much the first, sustained effort at a teaching text to campaign play.  (B1: In Search of the Unknown performs a similar function for stocking a dungeon.)  B2 was shipped with, I believe, all printings of the Moldvay rules, and though Gygax's name is listed on the cover, it looks like practically everyone at TSR at the time was involved in "revising" or "editing" it. 

And what does it have to say for itself?

Using the KEEP as "home base", your players should be able to have quite a number of adventures (playing sessions) before they have exhausted all the possibilities of the Caves of Chaos map. . . . In fact, before they have finished all the adventure areas of this module, it is likely that you will have begun to add your own separate maps to the setting. The KEEP is only a small section of the world.  You must build the towns and terrain which surround it. You must shape the societies, create the kingdoms, and populate the countryside with men and monsters.

The KEEP is a microcosm, a world in miniature. Within its walls your players will find what is basically a small village with a social order, and will meet opponents of a sort. Outside lies the way to the Caves of Chaos where monsters abound. As you build the campaign setting, you can use this module as a guide. Humankind and its allies have established strongholds - whether fortresses or organized countries - where the players' characters will base themselves, interact with the society, and occasionally encounter foes of one sort or another. Surrounding these strongholds are lands which may be hostile to the bold adventurers.  Perhaps there are areas of wilderness filled with dangerous creatures, or maybe the neighboring area is a land where chaos and evil rule

After the group establishes itself and obtains equipment, they will either follow clues gained in conversation with residents of the KEEP or set out exploring on their own (or both).  Naturally, they will be trying to find the Caves of Chaos, but this will take some travelling, and in the meantime they might well run into more than they can handle. Thus there are two maps - an AREA MAP for use when the party searches for the caves, and the CAVES OF CHAOS MAP which is a dungeon level map.

TRIBAL ALLIANCES AND WARFARE: You might allow player characters to somehow become aware that there is a constant fighting going on between the goblins and hobgoblins on one side and the orcs, sometimes with gnoll allies, on the other - with the kobolds hoping to be forgotten by all, and the bugbears picking off any stragglers who happen by. With this knowledge, they might be able to set tribes to fighting one another, and then the adventurers can take advantage of the weakened state of the feuding humanoids. Be careful to handle this whole thing properly; it is a device you may use to aid players who are few in number but with a high level of playing skill. It will make it too easy if there are many players, or if players do not actually use wits instead of force when the opportunity presents
itself.

MONSTERS LEARNING FROM EXPERIENCE: Allow intelligent monsters (even those with only low intelligence) to learn from experience. If player characters use flaming oil against them, allow the monsters to use oil as soon as they can find some. If adventurers are always sneaking up on them, have the monsters set warning devices to alert them of intruders. If characters run from overwhelming numbers, have the monsters set up a ruse by causing a few to shout and make noise as if there were many coming, thus hopefully frightening off the intruders. This method of handling monsters is basic to becoming a good DM. Apply the principle wherever and  whenever you have reason.

EMPTIED AREAS: When monsters are cleared out of an area, the place will be deserted for 1-4 weeks. If no further intrusion is made into the area, however, the surviving former inhabitants will return or else some other monster will move in. For instance, a thoul might move into the minotaur's cave complex.

There are also specific notes about how the Caves of Chaos will change in response to player depredations.  For example:

(DM Note: Orc losses cannot be replaced, but after an initial attack by adventurers, the males at location 10. will move four of their number into area 9., arm these orcs with crossbows, and lay an ambush for intruders. If the leader is slain, all surviving orcs from this locale will seek refuge with the tribe at C. (see below), taking everything of value (and even of no value) with them, and B. will thereafter be deserted.)

All of which indicate to me that B2: Keep on the Borderlands is trying very hard to teach its users a particular style of play in which the PC's, operating from a  safe but not boring base town, plan raids on a menu of nearby dungeons (i.e., largely unconstrained choose from a menu of small-scale "Situations"), and then the environment updates itself in response to players' deeds.   

Now, when I was setting up the Black Peaks game in early 2008, I hadn't read or played Keep on the Borderlands, but I pretty much did the same thing independently.  (Well, maybe Robbins's West Marches was built off this module, I don't know.)  So I think inspirations for sandbox-style play go back to the early 1980's if not earlier, and if people want to use this module as an inspiration or teaching text, I don't think that's too far off from original intent. 

(Historical things worth chasing down for comparative purposes but outside the scope of this essay: Douglas Niles's section on campaign design in the otherwise worthless Dungeoneer's Survival Guide; this section was edited and reprinted in Paul Jaquays's 2e-era Campaign Sourcebook and Catacomb Guide.  The rules for wilderness design in the 1981 Expert Set by Cook & Marsh.  And 2e's failure to explain itself, which seems like one of the worst unforced errors in the history of gaming.)

Value Added: Melan's Diagrams
Check something out here: Here's Melan's famous post about non-linear dungeon design.  Here's his map of the Caves of Chaos (but you should read the rest of his post, because it's really solid!)


See that flat base line?  IMO that's the essence of the sandbox.  And so long as that base line is there, it's okay to have linear things branching off (like the "D. Goblins" cave).  That cave permits for very little player choice, but it's tolerable since it's embedded in a very open macro-structure.  (You can also embed a sandbox-within-a-sandbox, which is what Keep on the Borderlands was within the larger structure of Eric's Glantri game.) 

Unrelatedly, our experience with the Keep on the Borderlands suggests that sometimes, all of the choices on the menu are unappetizing.  "Ugh, not an endless menu of poverty-stricken humanoids and bickering with the very uptight denizens of the Keep.  Let's get out of here."  So it's sometimes helpful to include an escape hatch, where you can jump one layer out and find some other sandbox to muck around in.  In the Glantri game, the players eventually ended up exploring Eric's mega-dungeon, which I know very little about.

James, Will You Ever Shut Up?
Soon!  I promise!  I think I've said everything I need to say about sandboxing as a functional style of play.  Hopefully there's enough here, either in the posts themselves or in linked examples, to provide a definition--the fact that C. Edwards, Dave Berg, and I all seem to be on the same page (and arguably on the same page with Gygax et al.--is very encouraging.

Having come this far, I want to talk about some failed sandboxes, pseudo-sandboxes, and Story When.  But I've got a lot of work to do later this week, and then I'm going to be traveling, so I won't be able to respond very frequently between now and November 14, 2011.

[contracycle]

If people want to discuss the early history of D&D with respect to this type of play, preferably from actual experience, I'm cool with that.  But it's not my primary focus.  If it turns out that nobody played this way in 1974 ("sandbox," as a term, appears to come from relatively recent video games, so maybe the idea behind it is brand new), it wouldn't change the fact that in 2011 these procedures work reasonably well. 

I don't know that it was a "reaction".  I would have thought this was pretty explicit in the text of the Expert set, from 1983:

"As a Dungeon Master, your D&D wilderness adventures will be far more challenging than a simple dungeon or two. For example, you should have a general idea of what is in each area of the wilderness, for the party may go anywhere! Although a few hints may help to guide the characters toward a desired area, you must be ready to make up minor details as needed, often during play."

That's from the Introduction.  Later text goes on to discuss the establishing of a (capitalised) Home Base for the players, mentions that the DM should prepare lairs for when PC's randomly bump into them, and populate wandering monster tables according to habitation zones.

It seems to me that this style of play, which we now call sandbox, was expressly a sort of "graduation" out of the dungeon.  Dungeons per se were training wheels for characters of level 1-3, and after that you got to go out into the big bad world.

The Companion set then introduced the tem "campaign", and discussed how to handle characters of significantly different levels.  It also contains "dominion" rules for the players to set up as territorial powers.  So by this point pretty much all the elements of "sandbox" play are explicitly present, and if we want an original term from before "sandbox" was adopted, that term would be "campaign".

So, in terms of the argument about where "sandbox" play comes from, I do not see it at all as a reaction to Story-based play of any kind.  The OSR has a pretty strong case when it argues, more or less, that this was the way D&D was "meant" to be played.  If anything, the opposite is true - story-oriented play was a reaction to the sandbox.

All this stuff did actually appear textually and people did play that way.  I certainly did, at any rate.

[Abkajud]

Hey James!
I'm almost certain that I got the term from video games and from RPGers who play video games (and thus, probably got the term from video games). Good point!

Admittedly, I wasn't clear on the time frame I was referring to, but I suppose talking in terms of time here is less important than talking in terms of, uh, traditions of play-styles, maybe?
That is to say, there has been (and still is) a great deal of play out there that consists of running the players through an adventure.
In terms of video games, sandbox play doesn't necessarily escape the railroading effect; it simply spreads it around - - in Grand Theft Auto 3, you can run around and fight and steal and get into all sorts of mischief, but once that gets old you have half a dozen or so different NPC camps you can meet up with, who will give you missions to complete.

Important: none of the missions are time-sensitive *except when you're in the middle of one*.
That is, if the Italian mafia has a mission that involves boosting cars from a particular convoy on a particular evening, you can run around and do "free play" as much as you like, and that car-jacking mission will still be there, waiting for you to take it.
Once you are on the mission, boosting said cars, you'll probably have some sort of time constraint that remains in effect until the mission is over, and then it's back to free play until you feel like taking on another mission. It's like you're still being railroaded, but not all the time, and you can choose which train route to take, to some extent. There are limits to the medium, of course.
In terms of video games, sandbox play is (for players, anyway) a reaction to rigid, linear plots found in the likes of, say, Final Fantasy 3. And yet, fascinatingly, diehard Final Fantasy geeks will gush over what they find to be amazing, tear-jerking stories they get to experience by playing those games.

This, of course, creates all kinds of terminological confusions over in our little end of the nerd-pond, as people hear "story" and think "linear, semi-rigid plot".

Going back and reading the rest of your response, I think we're on the same page. As for the problem of universally unpalatable options, I think it's possible that part of the term "sandbox" can mean/means that it is nonetheless a GM-determined structure, literally a big, open box around the PCs with stuff inside for them to do.

I read that and think about the linear plots I've been run through by GMs, and for me, the point of a sandbox is to create an adventure that's a lot harder to "break". I'm sure a lot of us have been there - - we prep something, we sit down with some players, and then at some point during play, we all can sense the ... limits of pre-planned adventure.

It's squatting right there in the middle of our social contract. And I think it's sitting there, looking at us, because of what Contra just said:

"It seems to me that this style of play, which we now call sandbox, was expressly a sort of "graduation" out of the dungeon.  Dungeons per se were training wheels for characters of level 1-3, and after that you got to go out into the big bad world."

When people are comfortable enough with the medium to do more than be led by the hand through the dungeon, but logistics, timing, getting everyone together, etc., means we have a tough time organizing a legit campaign that we're going to care about from month to month, and veteran players end(ed) up doing a lot of one-offs or single adventures that were easier to maintain interest in.
At least, that's been my experience. Heck, I'm lucky if I can get my local group to commit to picking up the same game two weeks in a row, let alone the same characters and plotline.

And now a quick derail: "If anything, the opposite is true - story-oriented play was a reaction to the sandbox." In turn, Story Now/Narrativism is/was a reaction to the "GM storytime" style, and that, imo, is why Story Now and OSR seem to be looking longingly at each other from across the dance floor. More precisely, Gamism and Story Now can pass each other in the hallway, pause, nod, and keep walking with a slight smile on their faces. But that's another subject.

-- Zac