Dungeon layout, map flow and old school game design


I posted my spoiler-laden renditions of the Rob Kuntz's Maure Castle levels "The Statuary" (Dungeon 112) and "The Chambers of Antiquities" (Dungeon 124) @ http://www.greyhawkonline.com/grodog/temp/maure_castle_mapping_analysis-grodog.pdf

Digging into the diagrams for the Maure Castle levels shows them to be very different levels, in terms of how their maps are organized.

The Statuary

The Statuary appears to be, and is I think, pretty complex: the encounters are grouped toward the lower- and the right halves of the map, while the use of many secret doors prevents the quick detection of the level's interesting challenges. Compensating for this (and shown on the map as a blue path) are the footprints of Tomorast and his cronies, which may help to guide the PCs toward that foe (or serve as a deterrance tactic, if the PCs opt to explore areas that Tomorast hasn't frequented yet), as well as to one of the key nexus points on the map (where many secret doors and access points to other levels converge). (FWIW, the adventure doesn't detail the exact path to 117 taken by Tomorast, that's my artistic license showing). I don't think that The Statuary is quite as complex as B1 In Search of the Unknown, but that may be a matter of opinon; it's certainly in the same league, at the least. It also resembles the right half of D1 as Melan mapped it, but with even more access points to the large chambers, which creates many looping paths to and from various encounter areas. In addition, the secret doors really are gateways to discoveries (unlike in B1, where they mostly provide an alternate path to a location that can be reached through other means).

The Chambers of Antiquities

While the Chamber of Antiquities is certainly simpler than The Statuary, I think that the map for this level appears to be much simpler than it is. At first glance, it looks like a circular route, with side branches (like S2 White Plume Mountain), but Kuntz added many branches and side-tracks to the map, to the point that the PCs can in fact avoid the central area completely and access the west, north, and south branches (if they manage not to awaken the juggernaut). They thus have complete freedom to choose the order in which to tackle the level's challenges. Also interesting is the area within the large central chamber, which is almost like a mini B2 cave complex---PCs can choose to approach any of the encounter areas (I almost mapped it like a starburst, but thought that the fishbone skeleton better reflected the progression of the challenges as the room is crossed). A good level, given the vast number of options it makes available to PCs; I also think that this map provides a relatively simple mapping expericence, which would be a nice change of pace from The Statuary (which is also the only level that connects to this one, via a secret area, which makes the discovery of The Chambers of Antiquities level a reward in and of itself).

Melan mentioned that is may be possible for some maps to be too complex:

Melan said:
Generally, branching, complex maps offer many possibilities for decision making, but overly complicated maps do not: they just cause frustration.

I debated about whether or not The Statuary crossed that line, and I think it skirts it closely, but manages to stay within the realm of "not too frustrating." I'm curious to hear what other folks think, though (I haven't run any of the new MC levels yet, though I'm looking forward to doing so in the autumn).

I probably won't have time to map The Whispering Cairn and A Gathering of Winds until after this weekend, but I'll post here when I make them available too.
Last edited:

log in or register to remove this ad


bumping for the current dungeon discussion thread "How dungeons have changed in Dungeons and Dragons."

edit - I'll also work-up a flowchart for the newest Maure Castle level over the holiday....
Last edited:


First Post
Means to get information to make meaningful choices rather then blind/random choices are important in the non-linear dungeons. The old school methods of hirelings, consulting with sages or demons, interrogations, communing with rocks and nature are all good methods.

Ultimately, in a game as opposed to a reality simulator where time and effort in preparation and play are limited resources, having a fun, entertaining, and for me appropriately challenging encounters are more important then the illusion of grand strategic choice.

You have to have cooperative players and a cooperative DM to have a rewarding game session. In essence, the players need to be good sports to hit the proper dungeon that was prepared to begin with. A whole world can be created and populated, much like Morrowind but like Morrowind most of the encounters will not be tuned for your players. Unless of course encounters are always tuned for your players which again brings up the illusion of choice. If you choose dungeon A or dungeon B and the results are approximately the same did you really have much choice.

If dungeon A was 6 levels lower then your party and dungeon B was 6 levels higher then your party then either choice is not going to be satisfying for your game time.

This, imo, is also applicable to the dungeons themselves. If you are going to level and gain magical items that put you ahead of the challenge of areas you have bypassed, backtracking is going to be relatively trivial. Nonlinearity is useful if clumped in clusters of encounters that can be tackled with appropriately leveled and equipped characters.

Again this is all a matter of taste. Some groups may like very well the free form worlds where intrigue, figuring out puzzles, and gathering clues about the world to make strategic decisions based on scouting and communing etc are important. It's not my taste as I feel that as a game it's more rewarding to me to be adventuring.

Now that said, my fondest gaming memories online seem to be the freeform nature of Ultima Online...

Melan - Incredible analysis!
I would like to say though that non-linear maps are sometimes more of a detriment than a asset. Unfortunately you stated that encounters were not he focus of the essay and that they were not part of our equation, however when designing a dungeon crawl, the purpose of the encounters may well dictate a more or less linear layout.

For instance, we have all seem the 'lair' layouts from the 70s & 80s that twist and turn and bob and weave, but think for a moment if your own house did that? What if you had to walk 30' down a corridor to your living room and then up a secret passage to get to your attic? Of in the case of a combined lair think if, for no reason you had to go through a circuitous route to go to the grocery store for no other reason than poor street design or city planning? Sure it happens, but even in the Medieval time period city plans were based on geometric patterns, usually squares or pentagons. It wasn't until these Medieval cities grew through the Dark Ages, the Renaissance and the Victorian eras did they begin to become entangled messes.

So, in some instances, overly complicated or even less than simple layout is a poor thing. You mentioned Village of Hommlet, my question for you is, if the 'Keep" encounter had been a huge layout, would it have made sense? The answer is no, the encounter was small by design and strictly a gateway for a larger adventure, the Keep was a building and therefore laid out as such to include the basement area. One can agree that the resulting Temple was much more sprawling and needed the 'exploration' you described. And though they were not originally released together, they were always meant to be part of the same plot-line.

I agree that poor design in a supposed 'free-form' dungeon is just inexcusable, but in an era where story-telling is as much a part of the game as the action, sometimes that small straight hallway is just what the doctor ordered. :)

Happy gaming and keep these great thought provoking articles coming.


Jack Colby, I would be interested in that. I have not made up my mind yet how 4e's "encounter zone" style dungeon creation philosophy fits in there. I am not even sure it is a relevant issue to my points, though.

Thunderfoot: you are of course right. In some cases, there are definite advantages to avoiding too sprawling maps; I think "lair" type dungeons can work well as a sequential or mostly sequential series of encounters. Maybe the Moathouse should be interpreted that way... although I'm still saying it would have benefited from a bit more layout complexity. It is a question of emphasis and degrees.

The issue of verisimilitude is either relevant to you or not. To me, it is not very relevant beyond the superficial. While "mediaeval" architecture might often have been simplistic, it is not necessarily a good model to emulate for all games; instead, we dwell on the stranger things... or at least buildings which are, as someone put it when discussing architecture for a 3d computer game "a pleasing jumble of basic elements". A building with hidden nooks, crannies, a secret staircase to the tower and a walled off section is more mysterious and more intriguing than a simple rectangular affair, and this is what counts. I do agree, though, that too much "noise" may not be so interesting. See this map for a good example of a dungeon which is very complicated, but wastes very little room on superfluous and empty space.

Last but not least, you can see the theory in action here (scroll down a bit). The Khosura undercity was consciously designed with these ideas on my mind. I think it was mostly successful in play, although not perfectly - of my players, some would have liked less mapping and wandering around and more concrete encounters. What I did realise is that the sparse key did not work so well for me and my group, and I improvised a lot in play to compensate.



Have you thought about analyzing the Mouths of Madness/The Store Rooms maps for Castle Zagyg: The Upper Works, by chance?


I do not have the product yet, but I am working on it. The shipping TLG is regularly using is a killer ($38.95!), and other outlets don't have the Upper Works in stock. Sigh.

the Jester

Bump for an awesome thread.

Melan, am I correct in remembering that you have a website with more of your analyses on it? If so, could you link it please?

This is a fascinating thread, one that I missed on its initial appearance.

As I was reading the last 100+ posts, a thought occurred to me which I was surprised no one had mentioned yet -- automapping in computer games. Automapping has been around for ages and ages; it's very prominent nowadays in most games involving "dungeon" exploration (whether the dungeon is literally that, or is a sci-fi location, or is a quasi-modern-day location as in a spy game or tactical shooter).

Most automaps do not show you secret or hidden areas (until you find them), but they can make the player suspect the location of a secret area in exactly the same way that an old-school D&D player would react to an incompletely filled in sheet of graph paper. This tickles my fancy in that the new technology is being used to replicate the old. And of course it's no accident that things turned out this way, since the histories of computer games and role-playing games are tightly entangled.

The game that I was thinking of when I had the automapping thought is Elder Scolls: Oblivion, and I want to point out one of its dungeons which I think is pretty well designed. Here is Vilverin, a large "beginner" dungeon that is literally the first thing the player sees upon emerging from the tutorial level.

Note: spoilers ahead.

[sblock]If you read the explanation of the area -- scroll down to "Zone 1: Vilverin", you will note that the dungeon starts off rather mundanely. Like most Oblivion dungeons, it is less a sequence of corridors leading to rooms than it is one huge interconnected area with room-like smaller parts. For that reason, if you go charging into Vilverin straight out of the tutorial, you are likely to get your butt handed to you by the mutiple bandits who are fully capable of swarming your position. (And interestingly, this somewhat presages the 4e D&D trope of interconnected encounter areas which can lead to masses of monsters engaging the PCs.)

Another thing that's not apparent from the overhead view map on the wiki page is the use of verticality in the Zone 1 map. From the entrance (labeled "Out" on the left-hand side of map), you spiral around a short stairway and then down a much longer one to the big open area. The section to the top of the map is lower still, so you can potentially sneak up on the bandits lurking below and literally get the drop on them; or you can descend another set of stairs to attack them at their own level.

At any rate, if you do not pay attention to your surroundings and pick up on the clues present in Zone 1 (i.e., the Dirty Scroll from the bandit ringleader), you will make your way down to Zone 2, fight a single bandit, and depart Vilverin without ever plumbing its true depths.

Zone 2: Vilverin Canosel is where the clever map design starts to pay off. There's a secret door that leads to the rest of the complex (location N on the map). What's especially clever is that just looking at the automap won't necessarily tip you off to the secret: the pre-secret area is nice and symmetrical. Instead the player must observe, or stumble upon, the trigger that opens the door. The player could use use auditory clues that something (a shuffling zombie) is beyond the "wall", or could potentially use life-detection magic to make the same discovery.

Past the secret area, there's a choice of two ways forward: past a locked gate (which the PC can unlock if he's good at the lockpicking minigame) or through another secret passage. In either case, there is an entirely separate section of monsters'n'loot that you can only get to by going through a flooded area -- a reward for persistent or risk-taking PCs (and it is a risk this early in the game because you will lack the ability to breath underwater and, depending upon your magical ability, might lack the ability to see underwater!).

Zones 3 and 4 are straightforward in a map sense, but have a variety of things the player must do to proceed. Unfortunately, there is no reason not to press every single button and follow every single passage you find; I'd like these zones better if there were some sort of clever environmental observation you could make that would reveal the easiest/most lucrative path.[/sblock]

Some of the later dungeons in Oblivion are much more complex, with multiple uses of verticality, looping, secret areas, and so forth. All of them, though, are essentially linear. I think the linearity works in a game like Oblivion, because you are either in a dungeon for a specific quest or side-quest, or you're there just randomly looting, so the dungeon is not meant to be a tremendously lengthy endeavor.

There are some much older computer games -- Eye of the Beholder, for example, which is explicitly a D&D game -- with some highly clever layouts involving teleporters, one-way doors, and all the other Gygaxian tricks. So even though those things may have fallen out of favor in print adventures, they linger on in computer games.


I have used the reveal technique in PnP gaming by having a flat screen TV as the tabletop and creating a map in photoshop with a layer of black that gets peeled back in stages.

It works very well if you have the time to set the map up properly.


Melan, am I correct in remembering that you have a website with more of your analyses on it? If so, could you link it please?
I do have a recently opened website, but at the moment, it is only a collection of support material for my Hungarian old school game system. I plan to upload English content in a few months, including a recap and perhaps expansion on pet topics such as dungeon mapping or The Tyranny of Fun.

You may be thinking of my maps from the Khosura undercity dungeons I posted on Dragonsfoot. They were drawn with this thread in mind, and include a semi-successful attempt to graph 'em.

Kwalish Kid

While I like the idea of this essay, I have a problem with the author's rather short dismissal of The Lost Tomb of Martek.
In the end, a dungeon without any real branches would look like a straight line (A.), or a straight line that looks slightly hairy (B.). The Slaver modules or Lost Tomb of Martek would fall into this category.
In the Lost tomb of Martek, there are a number of sections that must be completed. Thus it could be argued that these are not true branches, but rather a line with an optional order.

However, if one actually takes the time to investigate these sections, one finds in many cases the chance for a significant amount of sidetracks, branching, and circular routes.

We begin with the Desert Wilderness map. This map has a definite end point, but clearly has a number of significant side branches. Important locations are those with cloudskates, those with the oracles, and the cursed city. None of these are necessary for success, but they do offer particular successes and challenges that can aid or detract from the adventure.

In the next section, the Garden, there are a number of absolutely optional locations. The PCs can wander for a limited amount of time (and the DM has a number of options to move the action along to the next section). This section has thus has a number of branches.

The Crystal Prism is clearly its own, specific challenge, as is the Black Abyss and the Al-Alisk Desert.

The Mobius Tower, however, has a number of possible branches. The PCs have a number of areas they need not explore and a number of treasures revealed only through careful search. (Certain scenarios can lead to a difficult situation for the DM, and she or he has to figure out the behaviour of a number of NPCs that would otherwise not have any role at all, other than furniture.)

Even the last section has some payoff for branching investigation, though it is obviously limited.


First Post
Yeah, I agree. I read it through once a while ago, and meant to bookmark it, but forgot. Now that it's grown a bit, I think I need to re-read it from the beginning. It's been marked.

This is in many ways one of the initial assumptions of the thread (and elaborated upon in my second long post) - that the dungeon itself is an interesting and entertaining challenge, not merely a backdrop or "skin" to a series of encounters. Instead of considering a moderately challenging map as an inherently unfun thing keeping you away from the "good stuff" (encounters) as many people here seem to say, I consider it a part of the fun.

People's mileage may vary, but I'll also argue that:

(a) Nothing thrills the players more than when they realize that a choice they made had a real and meaningful impact on the events of the game; and

(b) As a GM I want to be surprised by my players. If I want to convey a linear series of events (i.e. a plot), I'll simply write a story. I play RPGs because I enjoy watching the players create a story. I'm never happier as a GM than when I'm completely taken aback by something the players have done.

It's interesting that you would have chosen The Sunless Citadel as your original example in this thread, because I've recently been reflecting on why I've had so much success using that module as a DM. There are a couple of answers:

(1) From a non-physical stand-point, the module is refreshingly non-linear. Both the kobolds and the goblins are presented as challenges which can be overcome in a variety of ways. (I think there's some mention in the module about the goblins not being amenable to diplomacy, but if you ignore that you can get even more interesting dynamics out of the scenario.)

(2) From a physical stand-point, I knocked out a couple of walls and added a secret door in order to create more looping paths between the kobolds and the goblins.

PCs have made treaties with the kobolds; wiped out enough goblins that the kobolds were able to take control of the entire upper level; laid siege to the goblins only to discover that they've been duped into wasting their time while the goblins looped around and wiped out the kobolds; wiped out the entire complex without care; negotiated with the goblin leader to bring back the head of the kobold queen; and so forth. There was even the one memorable session where, faced with the problem of kobold raiders coming out of the complex, the PCs simply collapsed the entrance tunnel and called it a day.

As Justin Alexander says: Don't prep plots, prep situations.

Even on a smaller scale, non-linearity can be fun. In Monte Cook's Night of Dissolution there is a small dungeon complex accessible from a sewer tunnel. Ignoring the single staircase down to the second level, there's nothing particularly spectacular about this small sewer complex. There are only 3 or 4 encounters.

But Monte Cook did something clever: He added a second entrance to the complex through a secret door. With this simple bit of looping, PC choices now make a huge difference in how the adventure plays out: Do they enter through the heavily guarded front entrance (where they're likely to be ambushed) or do they go through the secret door (where, if they can avoid the sentries, they can instead ambush the ambushers)?

This is also an example of how non-linearity doesn't necessarily mean wasted prep time: The PCs are still going to experience 90%+ of the content regardless of which entrance they use. Their choices, however, determine how they experience that content.

In fact, in my experience, embracing situation-based design usually involves less prep-work than plot-based design.

3. Lastly, do you think dungeons (meaning maps of any interior space; buildings, towers, sewers, etc.) should put more priority on non-linear style or on logical construction by the in-game designers?

I think it's a false dichotomy.

For example, I'm currently sitting in my house. There's an interior loop of kitchen-dining room-parlor-entry hall-kitchen. There's also a front door and a back door (which also creates an effective loop). Sub-branches (upstairs, sun room, parlor, media room, basement) extend from the entry, parlor, dining room, and kitchen.

IMO, reality is far more likely to get sacrificed in effort to achieve linearity than it is in an attempt to achieve non-linearity.

How often, for example, do we see the manor house of the Evil Patriarch mapped up without any windows? Why aren't there any windows? Because then the PCs could theoretically smash through the windows at the back of the complex instead of going through the front door.

Design in the real world is almost always non-linear.

Which doesn't mean that everything needs to connect to everything else.


I think it's a false dichotomy.

For example, I'm currently sitting in my house. There's an interior loop of kitchen-dining room-parlor-entry hall-kitchen. There's also a front door and a back door (which also creates an effective loop). Sub-branches (upstairs, sun room, parlor, media room, basement) extend from the entry, parlor, dining room, and kitchen.

IMO, reality is far more likely to get sacrificed in effort to achieve linearity than it is in an attempt to achieve non-linearity.

How often, for example, do we see the manor house of the Evil Patriarch mapped up without any windows? Why aren't there any windows? Because then the PCs could theoretically smash through the windows at the back of the complex instead of going through the front door.

Design in the real world is almost always non-linear.

Which doesn't mean that everything needs to connect to everything else.
That makes sense. It's been some time, but I think my question was more about design intent. Should designers aim for greater non-linearity, like that so often seen in Gygaxian dungeon design (e.g. B1) or should reality simulation / verisimilitude based upon the needs and desires of the in-game characters be paramount? You're right that these things are not really at odds. But I do believe they are different goals and will result in different designs. I'm still digging Gygax's last design, his Upper Works, which does a heck of a job at doing both well. Incredible non-linearity, while bringing in verisimilitude and in-game needs. The insanity of the caves and corridors below mixed with the sensible defensive design of the castle and castleworks above.

An Advertisement