A Walk in the Dark A look in to the mind of an RPG designer


Maze of Possibilities

Hedge maze from "Harry Potter and the Goblet of Fire"
(c) 2005, Warner Bros.

There is one part of A Night in Seyvoth Manor that I've been having problems designing for a while: a hedge maze.

You see, this maze is intended to be huge with changing walls and paths, so that aspect alone makes it virtually impossible to represent on a tactical map. Not to mention that navigating such a thing on a physical map is somewhat of a time consuming process and this module's designed to be lean and run in a single day, so I've been looking for a means to make it fairly quick and painless.

After thinking about it and throwing the question to the Twitter collective, I decided to make it a sort of card game. There will be some progress cards (+1 point), neutral cards (no points or failure), threat cards (+1 failure), a few treasure cards and one "bonus" card. The DM would go around the table, asking each player to draw a card; the player can do so or try a skill check (hard difficulty) for a peek at several cards and pick the one of them, returning the rest to the deck. The objective: get "X" points before "Y" failures.

That all sounds well and good, but there's a problem: how do you determine "X" and "Y", and how do you determine what the deck consists of? How "rich" should the deck be on either side to ensure the party isn't drawing cards until the end of time or drawing every card in the deck? Although there are a certain amount of assumptions that can be made - for example, there *must* be at least "X-1" progress cards (the bonus cards counts as 2 points) and "Y" threat cards - but beyond that there's a whole lot of variables that makes it difficult to gauge the difficulty of the challenge.

My solution was to write a program in C# (.NET 4.0) which, given certain criteria for deck building (number of cards of each type) will perform random draws and determine the success/fail rate. This allows me to run several thousand iterations of the challenge and tweak the numbers as need be. It's the closes I'll get to actual playtesting, and one could argue it's even better because it's doing thousands of iterations.

So, for example, given the following guidelines...

  • The "main" deck contains 6 progress cards, 2 neutral cards and 4 threat cards.
  • The "secondary" deck contains 2 progress, 4 neutral, 3 threat, 2 treasure and 1 bonus card. Every time a card is drawn from the main deck, a card is taken from the secondary deck to replace it and the deck is reshuffled.

...I currently get about a 34-38% chance of success (across 10,000 iterations). The average number of card draws is a little over 9, which means two full passes around the table in a group of 5. To compare, chance of success in a complexity 2 skill challenge is lower than that.

In addition to that, players will average 1.5 threats per challenge, will get at least one treasure about 30% of the time and will get the bonus card about 13% of the time. That sound acceptable. but the convenient thing about this little application is that I can tweak the numbers and see how they impact the percentages. Heck, I even simulated a skill check to see how much of an impact it has (one skill check increases the chance of success by almost 10%).

The only concern right now is the lack of player skill being a determining factor. I am going to allow skill checks so that PCs can draw two cards from the main deck and decide which of the two they want to keep, but that's all I can think of right now. The challenge ends up being more about luck than anything, which I guess isn't such a bad thing but I'm trying to think of a better way to actively engage the players.

Do you have any ideas on how to approach something like this?


Disclaimer: I've been told that there is a labyrinth in the Fane of the Heresiarch from SVD Press. I'm currently an active player in a play-by-post version of that at Grind 4E, so I have purposely avoided reading that module and do not intend to spoil myself until the adventure is over. Those are the sacrifices we must make...

Note:  One A Night in Seyvoth Manor is released, I will probably also release the source code for the above application. Just in case you're curious or could use it for something similar.

Comments (5) Trackbacks (0)
  1. D,
    Why not a two or three different decks to simulate the skill, then the roll determines the # of cards the player can draw? Maybe 1 card per 10 on the skill check? Nat 20 = an extra card? Would add to the complexity of the program, but also give the maze ALOT of flexibility.

  2. I agree with DeuceDM – there should be “shades” of success. Higher check = get to look at more cards. Also, give the decision to roll a skill check some teeth – what is the consequence for failure? Give some treasure back, automatic failure, automatic encounter? Decisions are always engaging.
    You could also give a choice of some sort upon success – draw x number off the deck OR draw 1 off this other “minor treasures” deck, which has small consumables that boost skill checks or whatnot. OR the “minor treasures” deck could just be a range of skill bonuses, BUT the labyrinth has to be beaten in a certain number of turns. So then the decision becomes, “Do I take a bonus for a future turn, increasing the likelihood that I can draw more than one card, or do I draw a progress card because we have to beat this thing quickly?”

    Both of these things keep it from turning into “My turn? I roll to see if I can draw two…no…I draw one.” That would get boring quick.

  3. I don’t know if you have access to the old 1e adventure “Pharaoh”or not, but it has a maze/gauntlet portion that might be worth looking at to see if it gives you any ideas.

  4. I liked the “Chaos Maze” mechanics presented in the Fourthcore module “Fane of the Heresiarch” a lot. Instead of card draws, the 25 cards are placed face down in a grid, and essentially you play a game of memory. Match pairs of “progress” cards to earn a success. Failure to do so incurs certain penalties. There is also a “reward” pair, and five “monster” cards. The cards are removed when matched, and reshuffled after each player picks two.

    I used this to simulate an “impossible geometry” maze created by a Lovecraftian horror in my game. It allows for player skill (ability at memory) and creates better choices, I think, than card draws or a random generator.

  5. Fuck fuck – didn’t finish reading to the disclaimer. Delete my post if you don’t want to know!

Trackbacks are disabled.