I'm still working on the NWN1 portion of my statistical analysis of modding, so I'll have to get to that another time. However, I've recently replayed Baldur's Gate I and have moved on to Baldur's Gate II. One of the last quests I completed in BGI was Durlag's Tower. It was a dungeon I hadn't faced in several years and so seeing it again got me thinking about my own ideas for what constitutes a good - or even great - dungeon. A quick brain-storming produced the following list.
1) The dungeon can't be too big. I'm sure several people will scratch their heads over this rule. I don't list the converse, which is obvious, but this less-obvious rule is just as true when you think about it. A great dungeon should have several memorable encounters, but the bigger the dungeon the less likely each additional encounter is to be memorable. And simple fodder encounters add little awesomeness to any dungeon but potentially add irritation. After all, very few players enjoy meaningless fights unless their objective is experience farming. Plus the bigger the dungeon, the more likely we are to face the dreaded repeated encounter trigger giving us (in effect) hordes of the exact same thing coming at us. All designers should learn and/or know that when the good ideas for encounters run out, so should the dungeon.
So what size is perfect? In my opinion, three sizeable levels is just about right. This doesn't count one-off encounters outside these levels. For example, a couple guards on an exterior map before you enter level 1 doesn't make the exterior map a "level."
Take a look at this map I pulled from the web by simply Googling "Big Dungeon Map." Now honestly, would anyone like to get lost in this? If a DM I was playing with pulled this out, I would fire him. And if he were doing it for free, I would stop being his friend. Utterly Ridiculous.
2) The dungeon should have more rooms than corridors.
This goes doubly for a party-based game. Take a look at the map to the left. It's the Firewine Bridge ruins in BGI... and it's terrible. A big NO. Getting six characters around that maze of corridors is a nightmare. They each run every which way and get themselves into some dumb trouble before I know what's going on. Oh, and corridors are boring. Far better to have a few big chambers with interesting content than what we got here.
3) The dungeon should feature different types of adversaries. The more types of monsters, the better. Note, this doesn't mean having an orc, an orc soldier, and an orc chieftain. It doesn't even mean having a Fire Salamander and an Ice Salamander, although this is better because each is immune to different energies. Rather, I mean having a few totally different species that require different abilities and pose different dangers. Essentially, if you've used three or more of the different creature groups from the NWN2 editor (i.e. Undead, Humanoid, Magical Beasts, etc.), then you're probably good to go.
4) The dungeon should feature at least one memorable non-battle encounter. In modern CRPGs, this encounter most often falls into the puzzle category, but it doesn't have to. At the very least, it doesn't have to be a traditional puzzle. The NWN2 module Trinity featured a collapsed bridge that needed to be crossed. Durlag's Tower had a room that constantly exploded fireballs and could only be crossed during safe intervals.
4a) Make sure all puzzles don't suck. A corollary to point #4 is that if you do go the puzzle route, try to make the puzzle non-annoying. Mostly, this means allowing the character to do something other than reading lengthy texts or endless dialogs. Durlag's Tower had statues that asked questions about the Tower's history. This was a real drag. The Umar Hills Temple Ruins in BGII had a point where you had to detail the daily rituals of Amaunator - information that could only be gleaned from actually reading the scrolls scattered around the dungeon. *yawn*
But an even bigger no-no as far as puzzles go is the endlessly repeating and/or random rooms and locations. These are the "puzzles" where the trick is to learn the correct sequence of map exits in the correct directions to get through a supposed maze. Often, this comes combined with the endlessly respawning enemies every time you reload a map. Nothing gets me headed to the walkthrough sooner than this tired idea. The worst offender ever was The Fell Wood in IWDII. Yes, I know that wasn't a "dungeon," but it still sucked. Watcher's Keep Level 3 in ToB was another unfortunate example in an otherwise wonderful dungeon.
So what are examples of good puzzles? The Machine of Lum the Mad on Watcher's Keep level 4 is a good one. The HotU puzzle on level 1 of the Maker's Island (in which you deactivate the golems) is another good one. While the NWN1 campaign was generally dire, most of the puzzles were pretty good - or at least I've managed to forget the bad ones. Remember the creator ruins that took you back in time so you could learn the correct sequence of notes to play to open a door in the present? That was a pretty good idea.
5) The dungeon needs enough traps and secret doors to keep things interesting, but not so many that you destroy the flow. Let's face it. Finding hidden stuff is part of what exploring is all about, and the rogue in my party needs a time to shine and feel all manly (or womanly, as the case may be), but I do like these items to be in places we would all at least suspect. Traps on chests are obvious. Traps around that odd statue or the evil altar on the dais are fair game. Even traps in the middle of an open, empty, inviting room are fine. After all, the wary and/or intelligent adventurer might ask why the denizens of a dungeon would leave such a nice spacious area alone.
But the 100+HP trap of death in the hallway in-between all this is a sign of suck. Why? Because it then means that I literally have to have the rogue creep inch-by-inch throughout the entire dungeon. Move forward a bit and wait. Then move forward a bit and wait. Rinse and repeat forever, and that equals no fun.
6) The dungeon needs a memorable boss.
This is almost self-evident, but it still needs to be stressed. All the coolness that is your dungeon needs to build up to the coolness that is the dungeon's master, and there are a couple of ways to do this. First, the boss can have a cool back-story that grips you as a player. This is probably the way to go for humanoid adversaries. Second, it can be a monster that is awesome just for being awesome. Either way, it needs to be a suitable challenge for the party. Firkraag was a cool boss that was a bad-ass monster WITH a cool back-story. Durlag's Tower fell down a bit with the Demonknight because you spent the whole time talking about dopplegangers and Durlag, but he was still pretty cool. The all-time coolest end-boss was Demogorgon in Watcher's Keep. On the other hand, the Ice Labyrinth in TotSC... had a dude with a cape. Not cool.
7) The dungeon needs a compelling reason for being. Note the word "compelling." Durlag's Tower took this to an extreme. In fact, it was essentially an entire expansion in itself. You could learn an entire new story independent of Sarevok's plans in BG1 just at this tower, but you don't need near that level of detail. A beholder cult sprung up beneath the city works fine. So does a dragon's den. Each of those creates a narrative for your dungeon that provides a framework for everything else (layout, creature selection, traps, loot, etc.). You do, however, need something more than a hole in the ground with a bunch of creatures in it. The ankheg warren in BG1 is an example. Is it realistic to think a few ankhegs would be burrowing around some farms? Sure, it certainly works logically, but it isn't the makings of a great dungeon.
8) The dungeon should have a logical ecology. It's a point I've made a dozen times, so please excuse the boredom if you've already heard this, but every dungeon should have a proper ecology. What I mean is that it should be logical that all the creatures therein could live together and survive. A carrion crawler in every room doesn't work for a number of reasons (the boredom of repetitive fights, for one), but it also doesn't make sense logically. Where do all these monsters find enough food to eat? It would have to be the cleanest dungeon in all of Faerun.
Along the same line, it is possible to jam 300 orcs into some caves if you've made some provisions for how they will eat. Are there supplies picked up from highway robbery that's feeding them all? OK, but do they have enough room to sleep?
Finally, do you really think that basilisk and troll are going to live in adjacent rooms in harmony? Don't most creatures carve out a territory? Let's try to avoid monster overload.
So when every room in your dungeon looks like the picture below, time to rethink your design.
8a) Some empty spaces are awesome and awesomely scary. This is the corollary to point #8 because if every creature has carved out its place in the dungeon, then there must be some empty spaces between them. This leaves room for other kinds of cool encounters (see points 4, 4a, and 5 above) or neat details that will really bring a dungeon to life. And sometimes the anticipation of the next encounter can be more scary than the actual encounter.
9) Just say "No" to Frankendungeon. With only the rarest of exceptions, a dungeon should look like it flows together. And the exception is when your characters have somehow traveled into the mind of a schizophrenic. The vast majority of dungeons, however, should utilize different tilesets (or styles) sparingly. This goes back to the dungeon's narrative. If it has a good reason for existing, then it should be consistent within itself. Throwing a bunch of things together just because they are individually cool creates a dungeon that overall sucks. Most often, this rule is broken between levels, but an example of this within the same level is Irenicus' Dungeon from BG2. Much of the dungeon was consistent, but then we had these odd rooms seemingly out of nowhere: a mineral cave, a copse of trees, an entrance to the Elemental Plane of Air. There was a lot of strength in Irenicus' dungeon, but the Frankendungeon aspect kept it from ascending to elite status. Is this dungeon a laboratory for experimentation or a prison or a shrine to a lost love... No wonder Irenicus failed to dominate the mind of an ascending god. He couldn't even control his own...
10) Like all real estate, it's about location, location, location. Certain types of locations are naturally more atmospheric than others. A tried and true winner is the abandoned castle/fort/temple. Almost all my favorite dungeons have the bones of an old medieval stone building. I concede that this entry on the list is much more a matter of my personal preference than the others (it obviously fits with my other interests), but I also believe I'm in the majority here. As proof, think back on what you consider the "classic" dungeons, and I'd wager that most of them fit this profile even as far back as the P&P days. Temple of Elemental Evil? Check. Keep on the Borderlands? Check. Village of Hommlett? Check. Isle of Dread (the abandoned temple in the middle)? Check. And so on. I don't think this is coincidence.
On the other hand, when was the last time a bunch of caves became a dungeon for the ages? How about mausoleums and tombs? Tomb of Horrors is a big classic for this group, but by-and-large they're pretty forgettable too. BGII had a couple dungeons that were completely outside the box like the Planar Sphere and the Astral Prison, but these never quite sang to me the way other dungeons did, and my totally unemperical gut feeling is that I'm not alone here.
I mean, this is a cool map to look at, but it isn't the makings of a great dungeon.
11) The dungeon must be fun. This is the one rule for which you can break all previous rules. All great dungeons must leave the player wanting to restart a new game just to go through that dungeon again. Thinking back on them years later, the player is likely to remember them with the same fondness reserved for first loves...
OK, maybe that's stretching it, but you get the point.
So what are examples of great dungeons? I'm glad you asked because that's the topic of my next couple posts.