A customized version of Stonehell dungeon that is consistent with the history of my campaign world and reinforces careful play and resource management.I will need to sharpen that as I go, but it will serve to keep me pointed in the right direction for now. I can't talk about the history part yet because my players have spent the last several sessions almost stumbling on some big reveals. Some of my players read this blog, so I don't want to spoil anything. I can discuss the play style of my group though. My current Stonehell group consists of people who either work in, or are students in, the science and technology area. There are both professionals and skilled tradesmen in our group. I am a lawyer with an undergraduate degree in mechanical engineering. As you can imagine, my group has a very methodical play style. They carefully explore, usually with a plan, always with a system. This doesn't mean that they don't get into crazy situations, just that they usually have an exit plan ahead of time.
Since I know my group likes to plan, I want to make sure that my customization efforts reward and challenge their careful play style. This means I need to make it possible for them to get the information they need to make good risk versus reward choices in advance. That last statement seems obvious, shouldn't we always design our dungeons so that players can make informed risk versus reward choices? Of course, but if you have a group that puts extra effort into planning ahead and gathering intelligence, you want to make sure there are plenty of opportunities to do so. You also need to make sure that there are enough twists, and not cheap ones, that they can't ensure victory through their careful play.
Next, I need to determine the scope of my customization. Stonehell is a very large dungeon, even if you only count the five levels Mike has released so far. One quirk of the way Stonehell is formatted is that all five levels are the same size. This is not necessary for my purposes, in fact it even works against my goals to a certain extent. I also only plan for the group to be focusing on the dungeon till level 7 or 8, so I only need one more level than what has been released.
In general, I think that the levels closer to the surface need to be bigger than the lower levels. There are several reasons for this. First, low level characters die more often and extra rooms are needed to provide XP for the replacements. Second, because resources are tight, covering large distances is an actual challenge for lower level groups. They have limited access to healing and are usually spent by the time they stop to rest, when to stop or turn back is a huge decision at lower levels. As characters gain levels they also gain the ability to avoid or escape encounters on their way back out, they are also moving through easier encounters on their way to the exit. Third, you have to spread the treasure (XP) around a bit more at lower levels because it is very difficult for low level groups to survive the types of encounters that should be between them and large rewards. You should certainly have some of these high risk opportunities available that clever parties can exploit, but you also need enough "slow XP" to get a bunch of characters off the level. Ask me about dragon fishing some time.
I need to make Stonehell smaller, especially the lower levels. This is not a criticism of the original dungeon, it was written for a much larger audience than I am running it for. It was also written to provide modular opportunities, that I am taking advantage of by removing some sections. Reducing the size of the lower floors will allow me to provide more concentrated, and varied, challenges and rewards to my more experienced players. It isn't just about character experience, in the early levels you are trying to give the players certain core dungeoneering skills. They learn mapping, resource management and how to move through the dungeon tactically. They are also calibrating their baseline for the fictional world you have created. They are constantly testing to see what the consequences are for their actions.
Remember, the very fact that there is a megadungeon and magic means that the world they are exploring doesn't "make sense" to a certain extent. They are going to need a bunch of data points to figure out how it departs from the real world. Repetition is the key to learning those skills and establishing that baseline. They need chances to test their assumptions before having to apply those assumptions in high risk situations. Death in the early levels should be like an airline accident, unless they make a clearly boneheaded move, it should be the result of multiple errors over time that suddenly become very dangerous when combined. If there is a total party kill on levels 1 or 2, the players should have a list of choices they made to get them into that situation in the first place.
On lower levels this all changes. The players should be faced with challenges that they apply their hard won skills to overcome. It is far less important for them to map a large area to determine where secret doors may be than it is for them to notice several very subtle clues in a smaller map. They should be facing more challenging encounters for greater rewards. Instead of pacing themselves to deal with a string of smaller encounters, they should be faced with situations where they are trying to survive long enough to figure out a strong enemy's weakness. I just don't need much real estate to pull this off.
Another thing to consider is the way I run Stonehell. I will have one or two ongoing groups exploring the dungeon, which is persistent across all games I run. I will also have a much larger number of groups who only enter the dungeon one time at a convention or game day. This means that very few characters are ever going to make it below level 2. Players at conventions do not generally come equipped with megadungeon exploring skills. I don't know that these skills were ever common, but they certainly are not now, RPGs are far more diverse, even under the D&D umbrella. This means that I really can't hand out 5th level characters and start a convention group out on level 3 or 4. These levels are going to be far too intolerant of minor mistakes for groups to be probing them by trial and error, they are intended to challenge groups that have spent many, many sessions doing that in the levels above. If only my ongoing groups are going to see the lower levels, I just don't need as much level to be there to deal with the fact that my dungeon is persistent.
Keep in mind that "smaller" does not mean small from an objective standpoint. Half of a Stonehell level is still a lot of dungeon.
Right now my plan for the dungeon is something like this: Levels 1 and 2 will remain largely unchanged, they are large XP farms and mapping areas, with the mapping getting a little more challenging on level 2. These levels have two or three sub-levels each. Level three will be a clear shift in direction, it will be more compact and challenging. I also plan to have the distinction between level 3 and level 4 be more blurred than in previous levels. This means that players will have to think more about relative vertical distance to gauge challenges. Level 4 will be made up of many smaller areas and will require a lot of vertical movement between those areas. It will probably also require things like going up to go down in certain places.
Level 5 continues the trend of smaller areas and vertical movement, the dividing line between level 4 and level 5 will be even more blurred than the line between level 3 and level 4. You will not be able to reach all areas of level 5 from another area in level 5, some will require going back up to level 4, others a more dangerous excursion down to level 6. Careful mapping will be required to determine that these possibilities exist. Level 5 will also introduce dangerous one-way movement. Level 6 will be a return to a more compact level, but be very challenging. One way movement and magical movement will be much more common and often leave the party in very dangerous situations. I expect that this 6 level version will have a bit more real estate than four current levels.