Zelda Dungeon Analysis, Part 1

The game I am currently making is a procedurally generated Zelda-like. This means that I have to try to recreate the dungeons of Zelda. This means that I have to understand what makes the Zelda dungeons work that special way that they do. To do this, I turn to my friend, systems analysis.
So, what does a sample Zelda "dungeon" (by dungeon, I don't necessarily mean dungeon, I just mean any location in the game that isn't overworld) look like?
Well, here is a lovely example:Now, you may say to yourself, "Bwah? I know Zelda dungeons, and that, my good sir, is no Zelda dungeon!" Well, that is a fair assessment.
What if I were to show you this instead:

Compiled map courtesy of http://ian-albert.com/games/legend_of_zelda_a_link_to_the_past_maps/, art assets owned by Nintendo

That's more like it, right?

Now, you've probably guessed the punchline , those are the same thing.
Ok, so with that out of the way, what does the top graph mean?
Here is a handy dandy legend:
Blue Diamond - Entrance or Exit
Blue Circle - Boring node
Red Circle - Enemy node
Green Circle - Puzzle node
Brown Circle - Enemy and puzzle node
Yellow Circle - Treasure node
Orange Circle - Treasure and Enemy node
connector with arrow - one way path

By "Boring node" I mean a room that has no enemies, puzzles, treasure, or traps. I.e. a room that is boring.

So, what are some statistics about this map?

Exits/Entrances: 2

Total Nodes Count: 20

Boring Nodes Count: 5

Interesting Nodes Count: 13

Total Non-exit Count: 18

Percent Interesting: 72%

Connections: 21

Percentage unnecessary nodes: 0%

Enemy nodes: 7

Percentage Enemy nodes: 38.9%

Treasure nodes: 4

Percentage Treasure nodes: 22.2%

Puzzle nodes: 5

Percentage Puzzle nodes: 28%

So, interesting things of note:
1. This map has two loops, one big, the other small. The small loop is easily erased as it doesn't substantively change the map. The big is easily erased by taking the one way arrow from the top orange node, connecting it to the exit node, and erasing the other connection to the exit node. Essentially the one-way arrow to a node near the exit node is already putting the final exit of the level near the orange node.
2. For all it's maze-like tendencies, there isn't a single room here that isn't absolutely necessary
(unless you count the two false starts at the beginning, more on those later).
3. There are also quite a few instances of two separate nodes occupying the same screen space. This will be tricky to achieve but should pay off well. To make a boring node less boring, segregate it completely from an interesting node but allow the interesting node to be visible. This achieves 3 things, the boring node gives the player a breather, the boring node is made more interesting, and, most importantly, the player is given a glimpse of something they want (action, treasure, etc.) creating a yearning to reach that node. By showing them the unattainable (at least currently) they desire it more.

4. Important treasure is always placed with lots of enemies (the orange nodes). This is good as it forces players to fight.

5. Puzzles ease the player into the level, which then gives over into a prolonged battle.
6. The two bottom boring nodes are actually rather fun as they represent a 3-way fork right at the beginning of the level and show the player the dead end they chose, but also the one true path that they should take. Perhaps they shouldn't be treated as separate nodes, but rather as a more complicated node.

So, what can we take away from this?
1. In the early stages of the game (this is from the first dungeon of Link to the Past) a small number of nodes is desirable (20 in this case).
2. It is ok to have "boring" nodes (although these shouldn't be boring). Players need the ability to catch their breath.
3. While enemies are the most common way to make something interesting, they account for under 50% of all nodes AND under 50% of all "interesting" nodes. Enemies are important, but are certainly not the only way.
4. There are very few branches. All that matters is giving the player the illusion of choice, not actually giving them choice. By making the rooms meander, you make the player feel like they aren't just moving in a straight line (See FF13 for an example of why making players just move in a straight line is a bad idea.).

Thus concludes part 1 of our Link to the Past Dungeon Analysis. Next time, we shall delve into the dark world!