I made a game called Dungeon wish. It’s a 1-minute roguelike. I described it in detail in my notebook. Do not do that when you design for fun! Your intuition is enough. Just start with a basic idea, test it and continue designing from there.

Dungeon generation

I wrote it on the first day. It was unnecessarily complex:

1. One level – 3×3 grid with blocks (4×4, 16 tiles).
2. Algorithm for connecting 3 blocks
3. Algorithm for connecting doors with adjacent blocks. Doors can have 2 orientations: vertical, horizontal.

Instead of this system I should have tested block destruction mechanic as early as possible.

Convoluted energy mechanic

Nobody understood my energy mechanic. I failed to communicate this mechanic clearly through gameplay. But it’s the core mechanic of DungeonWish. When you understand it the game becomes interesting. Only then you can appreciate wish system I implemented.
Basically, energy is the only resource in the game which is restored with fountains. When you loose all the energy, you die.

Magic mechanic

GIF with dungeon magic

Initially green platform was supposed to open doors and trap player. It proved to be too hard to implement with randomly generated levels. So I made green blocks perform some dungeon rebuilding magic. I figured this out through playtesting!

Roguelike with only two items

Hammer and rope. You can use rope to climb down a pit. I planned to use hammer for throwing but never got to that.

Conclusions

Notch is designing for fun. It means he has a playable prototype at initial stages of the game. I only had playable buggy version in the middle of the second day. And after I got it playing I understood that some of my design decisions were improper for the game feel.

Overall LD28(LD48?) was great! It helped me advance my design thinking.

Tags: dungeon crawler, dungeon generation, postmortem