Puzzle Dependency Charts

Shame on me for failing to post anything for over a year. I’ve been working on some puzzles and ideas of my own, but have nothing to reveal yet. As part of that work, I’ve been expanding my knowledge of Puzzle Dependency Charts so will instead share some information about them, including two full charts for LucasArts flawed masterpiece, The Dig.

Puzzle Dependency Charts (or Graphs, or Diagrams) are an excellent tool for designing and refining interconnected puzzles. Online discussions usually focus on their use in the classic LucasArts adventure games since Ron Gilbert of Monkey Island fame developed them to help design his games. But they are not limited to computer games and would be helpful for designing physical Escape Rooms or any multi-stage puzzle.

Chart 1

There are several different ways of approaching a Puzzle Dependency Chart (PDC). The most straightforward way is described by Ron Gilbert himself in this 2014 blog post. There’s also this 2013 talk from Noah Falstein. To the right (click to view full size) I’ve mapped the entirety of The Dig in a similar style to the one they describe. I’ve added some colour coding which I’ll explain in a bit.

The steps in each puzzle are represented by blocks, e.g. the “Release pig” block near the start of The Dig. I’ll call these blocks “nodes”. Nodes can be thought of as either locks that need to be opened or as the keys that open them. Except where necessary to a puzzle, the nodes don’t include plot points or character development.

There are directional arrows between the nodes to show how the locks and keys depend on each other. I’ll call these arrows “edges”.

To solve the puzzle “Release pig” in The Dig, you have to talk to Miles. If you think of the pig as the lock, then Miles is the key. So, I’ve created a blue node named “Miles” to indicate a conversation. The edge from the “Miles” node points to the “Release pig” node.

When you solve the “Release pig” puzzle, The Dig rewards you with 5 new inventory items to pick up, i.e. you get 5 new keys. I’ve created purple nodes for each item, and edges point from the “Release pig” node to each of the item nodes. The Dig continues with you using each of those inventory items in various combinations to place and detonate bombs on the surface of a meteor.

You can define “locks” and “keys” however you want and the more ways you avoid using physical locks and keys, the more interesting the puzzles become. A key could be an item, a piece of information or an entire location that gives you access to more of the game. Locks could be a sealed door, an enemy to defeat, or a broken inventory item that you need to fix.

In this approach to the PDC, edges don’t usually have text on them. My Dig chart has a few named edges where the same inventory item (e.g., the shovel) is used again later in the game. Drawing more edges all the way from the first shovel node to every later node that uses the shovel as a key would be redundant and confusing.

I’m reminded of a puzzle I got stuck on in Indiana Jones and the Fate of Atlantis where I didn’t pick up my ladder after using it to enter the city. So I couldn’t access several other areas of the map. In that case “pick up ladder (again)” would be a key node on the PDC.

Chart 2

Another way of drawing a PDC was described by Josh Weinberg in a series of articles and talks from around 2016. Once again, I’ve mapped the entirety of The Dig in his style to the right. Click to view it in full size.

In this style, the nodes represent the locks and the edges represent the keys. This means that all the edges have text on them and my colour coding has gone. It’s sometimes hard to decide what’s a lock and what’s a key and I haven’t been very careful doing it, so there are some obvious mistakes. But I hope you get the idea!

I created these charts using my favourite free graphing program, yEd Graph Editor. Regular readers of this blog will know I used it to help solve The Maze.

You can download my two PDCs here and here. Please credit me if you use them elsewhere.

Analysing The Dig

There’s a few things you can learn from these charts about how The Dig was made and why it’s not LucasArts’ best puzzle game. Huge chunks of it are very linear. The Nexus has a lot of doors, but you can only open one at a time. There aren’t many occasions where you’re able to work on multiple puzzles at the same time.

This limitation on how much the puzzles branch out is probably caused by the complex story. A lot happens in The Dig, and it has to happen in a particular order, so you have to keep on hitting puzzle bottlenecks where the next bit of story can be told. Day of the Tentacle and Monkey Island have much simpler stories so the puzzles can be more intricate.

The puzzles in the Dig are also all very similar. You’re going around fixing broken machinery and unlocking real doors with real keys, so there isn’t much variation in the types of keys and locks you’re given. Instead, the Dig is focused on exploration and finding (and understanding) the keys and less on solving overlapping puzzles.

What The Dig does well, though, is that your actions throughout the game all point towards the end. You can see this by the many long edges that extend almost the entire height of the chart. Finding metal plates and fixing light bridges are ongoing, gamelong tasks that have a big payoff near the end. Then you finally get to use the razor sharp jawbone you’ve been carrying around since almost the start as part of a major plot development.

Designing a game

So, how do you go about using PDCs to design a game? The first rule of PDC club is to start at the end and work backwards. A game will often be broken into chapters or other bottlenecks, so you can start at the end of each section instead. These sections can be wholly independent (like some of the Acts in Monkey Island), or some of the items and locks may carry over from one section to the next (as in story-enforced bottlenecks in The Dig).

Let’s say, at the end of Chapter 1, you defeat the Gorgon. How are you going to do that? Perhaps you need to get a special sword, a mirrored shield and a magic bag to keep her head in. You work backwards from each of those items deciding how they are obtained, until you reach the start of the game where the player receives their first key: a poster promising a vast reward to whoever can slay the Gorgon.

When first designing a game, you probably wouldn’t have a huge level of detail and instead just map out some key puzzles. Then, as the game is refined, each node can be broken into multiple nodes, or separate nodes can be combined, to increase or decrease the length and complexity of the game. The story can be developed and intertwined with the puzzles to create a great experience for the player.

The key thing, according to Gilbert, is to make the PDC bushy. A bushy PDC suggests non-linearity and more choice for the player. This makes the game more interesting.

In the Gorgon example, you might collect the sword, then use the sword to defeat a witch guarding the magic bag, which contains a key to unlock the dungeon of the mirror shield. This is entirely linear and probably not very interesting to play. Worse, if you get stuck at some point, there’s nothing else to do.

Far better in a puzzle game would be to make the three quests (almost) independent of each other. The player can choose whichever path they want, giving them a rewarding sense of control. And if they get stuck, they can try one of the other paths.

Gilbert describes each chapter as a diamond shape, starting from a single point and expanding until the player has several threads running at once, before contracting down to a single end point. This bottleneck can be used to tell some more story. Or, as in Day of the Tentacle, independent, parallel paths can each have their own subplot.

Final thoughts

I hope this post has been interesting. I’ve learned a lot over the past year reading up on PDCs and creating my own while playing some different games.

One observation – I think the first way of creating the PDC is much better for designing games. Having both locks and keys as nodes puts everything in plan view. It’s also easier to break up a node if you need to make one section more difficult or more bushy, or to collapse it down for simplicity. But the second type of PDC might be better for analysing a finished game since it removes a lot of clutter.

I picked up a copy of the recently released Detective Di last month. Look out for a new blog post soon analysing the puzzle structure. Short version: it’s a surprisingly linear game and while the story is engaging, and some individual puzzles are interesting, it’s not as interesting or replayable as the classic, bushier adventure games of the 1990s.