There's one topic that keeps me up at night. Dungeons. Rightly so, as it is half of the initial promise when I got into the hobby.

It is not the first time that this topic has come up. And it will not be the last. One question looms over it all: What makes a good dungeon? More specifically what makes a good dungeon that I want to run?

Beauty is in the eye of the beholder. What constitutes as a good dungeon equally so. For me it is important, that I can run a dungeon many times and the experience is different every time. Or at least has enough variety that I don't get bored running it.

There's also a distinction to be made, if the dungeon is part of a one-shot or a larger campaign. Each scenario has its own requirements. It is comparable to a short-story and a full-length novel. A one-shot dungeon is more likely to provide spotlight to all different characters, and thus has to cater to different aspects of the game. In a campaign I have more of a luxury to dedicate a dungeon to one specific aspect and then another for a different one. I can focus on a singular character and advance their story and give the spotlight for a whole session.

I've also gone the complete other way and abstracted dungeons completely. Dungeons as a means to an end. Not the central point of and adventure but just background.

Two things reveal themselves time and time again though: Motivation and Purpose.

The motivation lures the players into the dungeon. It is the reason why they should care. The hook. In a campaign this is most often very easy to provide. MacGuffins can be hidden in dungeons. Characters will be automatically drawn toward it. A one-shot has more problems establishing motivation.

There's a weak social contract, that keeps the group together. Otherwise there would be no game. A suspension of disbelief that the characters want to undertake that specific quest at this given time. They can't ignore the call for adventure. So the call needs to be interesting enough, but it should still leave enough room for interpretation. There should be more than one solution to the problem.

As important, is the purpose of the dungeon. This goes beyond the motivation for the players. The dungeon must make sense and isn't just a dungeon for a dungeons sake. It is there for a reason and didn't just pop up one day and was a dungeon.

It can have a past purpose. Why was it build in the first place and by whom? This purpose defines the architecture, places rooms and corridors. This is why you will find a pantry next to the kitchen and not 100 meters apart from each other. Somebody had to live in the dungeon. It also defines what rooms you actually might find and potential hazards.

Analogous, there's the present purpose. There's somebody now living in it. They may have repurposed the dungeon for their needs. These might be completely different from the original builders. Animals don't have need for a kitchen, so what happens to that room? Did the new occupants make changes to the architecture to fit their new purpose?

Past and present can be at odds. Something from the past can still linger in the dungeon and create an internal conflict, to which the players get thrusted against. The dungeon is a character. It wants to tell a story. Furthermore past and present don't need to be far apart. They can be in fact close together.

How does this now help us to create dungeons? And can these dungeons then be considered good?

Use it as layers to construct the dungeon. Let the past inform the layout and initial purpose. The present creates the obstacles and presents a new purpose. How does it clash with the original one? Are there still remnants of that past in the place or has it been completely overrun? On top of it all is the motivation. The players are now entering the dungeon. Do they know about the original purpose, was it their motivation to come here? Or do they find out about it, as they venture into it?

Tension arises from these three layers. Think of it like factions that want to advance their goal. From this tension, conflict for the players to solve is created.

Situations not plots.