A Big Mistake: Random Dungeons, and a note on dungeons in general

Nirrum knows how to scry and teleport

I, On the other hand, know Ctrl+F, Ctrl+N and Shift+Home/end. 

For whatever reason, your party has begun heading underground or into a walled area that they're unlikely to try to tunnel through. You want it to be interesting so you hop onto Donjon, Find the appropriate generator after an hour of mucking around with some others, and you set it up to more or less do the biggest, most labyrinthine, most secretive, obstructive, and monster-filled cavern your players have ever laid eyes on. You have made a mistake, and it's going to suck. 

Great Dungeon Dai-Chungus and its immortal weakness

In game design, one pure truth rings out like a sickening bell. Players hate going slow. Or Rather, players hate when their actions don't reveal new information. It's the reason the water levels in so many games are reviled. I see the door, let me get to it without having to look at the same barely shifting screen for a minute. In map design, this means something different, If each room looks the same, then there's no reason to think about each room. When you get stuck in the fugue of "did that do anything? No?" it really sucks. Thirty Azers and fire salamanders later, you've got nothing to show for it except a bunch of brass and spears.  The big map thus shows its flaw. It's not actually big. Not conceptually anyway. Your players have already figured out this dungeon, they already know more or less every challenge that faces them, so suddenly, every locked door is the same locked door. Didn't I unlock this already? Oh good, another trap.The shape and import of the dungeon inside of the player's mind is homogenizing, and maddening as it does.

On Dungeon Size

There isn't actually an answer to how big a dungeon should be. I've been in dungeons for literal months, and felt fine. Similarly, I've run dungeons that ran less than a session that my players were 130% done with when they emerged.The lore was there, the traps and theme all had reasons, but even then, It sucked. So what gives? As mentioned in my previous post on Map Size, a good designer can do a lot with a little. A bad designer does little with a lot. The amount of prep time, care and consideration that you put into your dungeons matters more than how big they are or what's in them. A friend of mine, who hit their stride as a designer right away says "Two sessions maximum" and seems to lean between 6-10 rooms at a maximum. The largest dungeon I've ever made that was fun was the temple of the world spirit that had about 13 and some change for hallways and secrets. The most Notorious dungeon ever, the tomb of horrors is pushing 30 and it's modern counterpart, the tomb of annihilation, has more than that in the first three floors. There isn't a real maximum number provided your players know what they're getting into, but I'm going to go with my friend on this one, keep it short

On Dungeon Content

I've talked before about keeping verisimilitude, that things are places for reasons. A dungeon should really be no different. Why have puzzles when you can have traps? Why bother having a "solution" to any of the puzzles if you want to keep something out? Acererak, everyone's favourite saturday morning cartoon villain lich kept the power of the people who died in his traps, trying to figure out his puzzles, so it suited his purposes to have them. The worst dungeon I've ever run had a similar excuse. It was the honor of each hobgoblin to die while building a room to make their artifact-house that much more confusing (and unfun). A dungeon's content, in my minor experience with them should be a series of rapid-succession, high-damage encounters that can be done in four sessions at maximum,  or,  to be a bit more broad, it shouldn't last much longer than an hour in-game time (of actually doing. And remember, Never use riddles.

The Dungeon has a long an interesting history, but conceptually it runs into what is in retrospect an obvious weakness. Gamers hate going slow. A puzzle is an obfuscation of the path forward, and confusion is such a thing as well. These things have their place in designing a dungeon, but only when they're not expected. Traps on the other hand, traps make dungeons worth it. They're fast and they happen because of carelessness. Crank up the lethality and status effects on those.

Narratively speaking, there are two types of dungeon that I have found work. Ones that the party chooses to go to, such as the old tomb of horrors, or ones that they have to go to, such as the tomb of annihilation. In both cases, the party has a goal. In the first case, there's a good chance that doing research about their location will reveal details about it's layout (either a floorplan or a description written by someone who was there when the location was being built or in use). In the second case, there's a good chance that it will be quick, as their goal should be straightforward and exclusive.

I never did understand why the tomb of annihilation wasn't approached by a romanesque army, literally carving a pathway through chult in a straight line, Cutting down all trees in their path and retreating into formation if they're attacked by any of the jungle's denizens. I just feel that a party of 100 would have an easier time. Upon entering the tomb, a ready supply of clerics and paladins would be a massive asset that you can rotate regularly. One guy's job could easily just be Gentle Repose.

This context changes a bit for Videogames, where the solving of the puzzle comes with tangible and immediate reward, but even then, the Ocarina of time water temple, one of the best designed puzzles in the game is infamous for its slow progression. Puzzles are great but if they're too hard,  that obvious pathway forward becomes a very pretty wall. A wall of frustration and incompetence. I relieve my players in letting them know that Picks, chisels, and shovels are things that exist.

A final admonishment

The end of this is to tell you that random dungeons, especially big ones are not ideal in any situation I've encountered them in. Aside from laziness, they serve nothing to the players narratively and are frequently paired with the reduction of choice. It's a perfect storm of boring, frequently linear (two unknown paths does not constitute a choice), time-consuming and narratively pointless.

Vorbet neshta'u'lo
-Nirrum the mad