Dungeon Logic

Let’s consider two degrees of freedom in RPGs. One is the Environmental Boundary of a game; this is how far the PCs are allowed to roam. The other is the System Boundary of a game, which is how much their actions are etched in the system.

I. Environmental Boundaries

Forget for the moment how big the space is, and instead think of how rigid the boundary is. A hard boundary is one that cannot be crossed in game, ever. There may be good in-game reasons, e.g. you can’t go to the Moon because your society hasn’t invented the technology yet. There may be metagame reasons, such as a D&D game that focuses only on the dungeon, not what’s outside it — in which case, the hard boundary comes from the game’s social contract. There may be no good reason other than the GM’s viking hat, but since that’s a special case I’ll cover it later.

Then there are soft boundaries. These aren’t directly restrictive, but there are consequences of crossing them. If the soft boundary is the GM saying “plot is OVER HERE” and the players ignore her there may be social consequences as she then scrabbles to GM on the fly for a couple of hours as the party ignore her lovingly prepped encounters.

That’s a negative example. A positive example is any kind of isolation or survival scenario where if the players attempt to leave, there are in-game penalties. The purpose of the soft boundary is to telegraph the penalties so the players can make a decision.

Is there a no boundary case? A sandbox might be considered unbounded if the play area is bigger than the ambitions of the party to explore; but by definition a sandbox has a boundary, somewhere. The value of a boundary is not only to physically restrict but also to provide a restriction on perspective, and a lot of the attraction of large-world exploration gaming comes from the expectation of something beyond the boundary. Really despite the name the sandbox is kind of irrelevant to boundary considerations.

II. System Boundaries

Now let’s consider if the system is similarly bounded like the environment. All components of a system must be related to either PCs trying to do things, or a measure of the game environment trying to do things to the PCs.

What’s the difference between a hard and a soft system boundary? An element with a hard boundary has a precisely defined function. The scope of its operation is non-negotiable. Armour, saving throws, D&D style spells are old-school examples, and Apocalypse World’s various moves are new school examples.

Soft-bounded system components have a lot more negotiation value. This is good because they can take up the slack when an unfamiliar situation arises and the GM has to find something for her players to roll dice against. D&D (and Runequest) attributes are softly bounded because they can be used for a catch-all test in a pinch, in addition to whatever they do in the RAW*.

III. Remarks

Why does this all matter?

a. History

First, I think it’s interesting to think where trends in RPG design have swung between hard and soft boundaries. Most games have non-negotiable system and environmental boundaries. This includes war-games, but RPGs are a peculiar exception.

The roots of our hobby lie in the Braunstein game. As noted in this excellent summary:

2. Players weren’t limited to actions in a rule book. It allowed players great latitude to take creative actions that would be interpreted by a game master.

In other words, Major David Wesely introduced softly-bounded concepts into a hard-bounded game, and p&p RPGs were born.

The capacity for negotiation is what sets RPGs apart from all other games; it is the USP of the hobby. CRPGs can’t do soft system, or soft environments. Players appreciate being able to apply their personal experience to a game and use that to influence its outcome. It’s pretty much what we do in real life, as we learn that rules-as-written are only one part of our day-to-day jobs, for example — a significant portion of probably everyone’s job is tacit experience and skill.

The early games still had a lot of hard boundary design, with saving throws, to-hit rolls, damage dice, etc. But into the nineties the preference for softly bounded system emerged. The World of Darkness with its negotiated Attribute + Skill = Dice Pool is very soft in terms of design, a flexibility that appealed to me but came a cropper when the power gamers got hold of the system, and divided the Vampire-playing community into one half who wanted to largely ignore the rules, and another that wanted to play fanged superheroes. It causes tremendous problems for player expectation.

I reckon the apex of soft system design came with my favourite Everway. That game is so soft, it’s positively fuzzy. But that’s OK, it’s designed for heavy player-GM negotiation and interpretation, particularly with the fortune deck.

Post-Everway we have games like FATE with very soft and negotiable boundaries around Aspects (more about FATE later). But also we have a new school with very rigid system boundaries. Burning Wheel is very rigid, but modular. Apocalypse World (and progeny) appears flexible because it’s so simple, but I consider the system boundaries to be very hard. Yes, there are some negotiation aspects but generally the game is about using moves creatively, but within the framework of the move as written.

What about environmental boundaries?

D&D’s wargame roots have strong environmental boundaries, something that’s carried into the dungeon. Gradually though the preference for soft boundaries in system crept into scenario design, because it makes sense to players that they should be free to explore. D&D transitioned from dungeon to wilderness, sandbox play happened, etc.

Then Vampire came along, and scene-by-scene railroading came along. Like the sandbox it’s a special case, but nevertheless it’s a form of GM-imposed, rigid environmental boundary.

And ever since the World of Darkness we’ve been forced to wear the hair shirt of railroading. Phrases like player agency creep into the indie design terminology — but at the same time, indie games are strongly scene focused. The story-now ethos forbids the GM from imposing a plot on the players, but she’s free to impose a temporary (and often arbitrary) environmental boundary in the scene; and if a character in a scene chooses to leave, they are consciously walking off-camera and out of the bounded area. Whether the boundary is soft or hard is open for debate; I would argue that it’s a hard boundary, because the cost of crossing the boundary is to the player (disengagement with the game) rather than the player-character.

The saving grace of this practice is that the scene is sort of negotiated up front between players; the hard boundary is built into the game. Hollowpoint is scene (or mission) focused, and because that’s the game the players have already agreed to play by those boundaries. Still, there’s pressure to “resolve the scene” and the implication that nothing happens outside the scene. The Story Game vs. RPG debate swiftly follows, but that’s something for another time.

b. Coherency

What is an incoherent game design? From Ron Edwards’ article it’s one of two things:

  1. The design “fails to permit one or any mode of play”, by which I assume Edwards means it’s impossible to focus on playing in one of the G, N or S modes;
  2. The design is a mixed bag among the modes, where some bits play as S and others play as N, for example.

If incoherent design means poorly signposting the mode of play, then we can infer that mixing system elements with soft and hard boundaries could also be incoherent. It certainly makes the game harder to learn and teach — take the attributes and saving throws on a D&D character sheet, for example. If both are given similar priority on the sheet how is a newcomer to know that you can negotiate activity based on one set, but not on the other?

The worst example I can think of is the good ol’ WoD, where everything is given the same 1 to 5 dots rating, but some traits are clearly broad in scope (e.g. skills) whereas others are very narrow (e.g. disciplines) and procedurally very different. Clearly the issue is signposting which traits are softly bounded, and which are hard-bounded. For example:

  • Runequest/BRP draws boxes around the different soft (skills) and hard (hit points, magic) elements, so gets away with having a lot going on on the character sheet.
  • Monsterhearts is mostly hard-boundary stuff, but soft-boundary items tend to be in separate headings and in text, rather than codified into moves.
  • Barbarians of Lemuria has a hard/soft divide around the combat/non-combat parts of the system (the latter being the very broad-scope careers).
  • FATE’s aspects are soft, but FATE points are hard; it separates the two by having aspects written down but points represented by tokens.

Coherency is probably easier to achieve when you have only soft or only hard-bounded items. Everway is the former and Hollowpoint is the latter. Both of these games are also fairly minimalist, so it’s unsurprising that they should be easier to teach.

c. Utility

By inference with environmental boundaries, system boundaries that are soft should include some penalty when the element crosses over into unfamiliar action. Usually when a player tries to use a skill that doesn’t directly apply, the GM will make up a penalty or consider the scope for success with an imperfect skill.

FATE deserves a special mention since its default mode is compelling/invoking Aspects and passing around fate points; use of Aspects is designed for players to try to push the soft boundaries, and the fate point expenditure is a cost of the crossing.

There may be something to be said for a hybrid arrangement in system elements; defining for a particular skill what behaviour comes at zero cost, and what stretch behaviour will require the player to pay in whatever game currency you’re using. A modification of WFRP’s skills could be used here. WFRP’s stat block largely covers “common actions” but the character is then “tuned” by individual specialised skills. The implementation in WFRP is not consistent — some skills just provide a bonus (e.g. a “hide” skill), others give permission to attempt the special action (e.g. a “specialist weapon” skill). You could change the function of the skills to allowing certain actions to be performed out of the normal scope for free. This means you’d need some kind of economy, perhaps a finite resource like Gumshoes’ investigative points.

IV. Dungeon Logic?

Consider Lamentations of the Flame Princess–very stripped down D&D with one purpose in mind: explore dungeons and get treasure, advance your career.

It has rigid environmental boundaries associated with dungeons and other dangerous places — if you don’t fight monsters and recover treasure from remote areas, you don’t advance. It also has rigid system boundaries associated for encumbrance, attributes, advancement, saving throws, even skills (if you’re not a specialist, your chances are 1 in 6).

This is “Dungeon Logic”. All character attributes have one function, and collectively they are directed at one task, which is dungeon exploration. Those traits don’t really work outside the context of a dungeon, but hey, that’s no problem if you never go outside the dungeon.

Now, back to all those games that aren’t RPGs, with their non-negotiable boundaries. Everything in such games is contextual; it doesn’t need to fit into a holistic worldview, although it ideally needs to be consistent with the player’s expectations to be satisfying. This is the new-school credo–if RPGs are games, shouldn’t they function like any other game with defined rules?

The conclusion is that new-school hippy gaming applies dungeon logic as well. Rules are hard and contextual, environment is bounded by scene.

Closing Comments: some examples

Soft Environment Hard Environment
Soft System WoD (freeform)


WoD (railroaded)

Live action games with heavy GM intervention

Over the Edge

Hard System WoD (munchkin)


Don’t Rest Your Head


Monsterhearts (scene bounded)

I’m mostly interested on soft system / hard environment games… which could be a recipe for disaster. A small environment with hard boundaries will often end up with one group trying to dominate others, but if the system is soft it’s very difficult to be “fair”. This is particularly obvious in live action games where success depends on getting the attention of the GM and negotiating for benefits.

But… is the soft/soft case any better? The main difference between soft/hard and soft/soft is in the latter case, a player is not forced to compete for resources, if that’s how the game goes. But if they choose that strategy and decide to leave the rest of the party, can they reasonably expect the GM to divide attention between them and the other players? That would be like James Hurley’s road trip in the second season of Twin Peaks.

  • Rules as Written, not in the nuddy.