Musings on metapuzzles, part 3: Resource metas in puzzle games

At the start of this I defined metapuzzles as being about collecting information or resources, but as I bring it around to actually talking about puzzle games I'll flip the order and talk about resources first.

This'll have some extremely minor early spoilers for Talos Principle 1 and Bonfire Peaks, and significant spoilers for the Hall of Innovation and Factory Floor rounds from the 2023 MIT Mystery Hunt.

Screenshot of Bonfire Peaks. The protagonist sits by a campfire marked "Burn Your Belongings (Solved)". A box has appeared by the fire.

Many games with individual puzzles feeding into a meta have each one reward a resource, sort of like puzzlehunts’ token metas. For instance, solving a puzzle in The Talos Principle rewards a tetromino that goes into a tetromino puzzle to unlock something new, or solving a puzzle in Bonfire Peaks rewards a box that can be used in the overworld to get to a new area. Just like answers in puzzlehunt puzzles, these resources create an abstraction around the individual puzzles, so you get the same practical advantages of the exact details of the puzzle not mattering.

That sort of thing can just be a glorified way to say "you need to solve X puzzles to proceed," but a resource can be more complex. For example, though I'm kind of stretching the definition of "resource" to say this, sometimes solving a puzzle provides a way to change state of the metapuzzle. In some cases, changing the meta state involves solving the individual puzzle in a different way, using a different exit or with resources left over. This breaks the abstraction layer: the design's more challenging, but the solver can think about multiple layers at once for a more unique experience.

(Unfortunately I don't have a good example I can put here because the ones I know of are very late-game spoilers)

The disadvantage of more elaborate metas like this is losing partial solvability- more complex ones tend to need all of the state changers to solve, and sometimes even become strictly linear in how you have to do things. It's something to be mindful of when designing the difficulty curve.

Can the state change meta idea work its way back into puzzlehunts? Well, I promised Mystery Hunt spoilers and haven't gotten to them yet, so you can probably guess the answer to that question.

A maze with colored squares and some letters in it. Notably, there's no path through the maze.

The 2023 Mystery Hunt took place in both a museum-themed puzzlehunt and the "factory" where the puzzles were being created. The Hall of Innovation round in the museum had a bunch of warnings about how the puzzles weren't finished yet. The puzzle above, Maze, is trivial by Mystery Hunt standards: find a path through the maze, write down the letters on the path, and that's your answer. The problem is, of course, that as shown the maze is impossible.

It turned out that solving puzzles in the factory unlocked state changers: a set of interactive "gizmo" controls on the hunt website that changed the Hall of Innovation puzzles in various ways. The Maze had two gizmos: a set of dials that rotated some of the walls to change the solve path, and a text entry that added more letters to the grid. Using those could make the puzzles solvable, or even change their answers. (And when you're on a team of dozens of people, anyone on your team changing the devices would change the Innovation puzzles for everyone on the team, to the chagrin of whoever was working on them.)

Two images of controls: a series of four pairs of dials, and a text box with "HELLO" input into it
The same maze as above, except the walls in the colored squares have changed, making a path through the maze, and the letters "HELLO" have been added to a few squares
While HLL is not an English word, the system would accept it as "correct" if it matched the current state of the puzzle.

The gizmos each affected multiple puzzles, so the first layer of the metapuzzle after understanding how all the individual puzzles worked was to come up with a state where all of the puzzles could simultaneously be solved. Doing that provided information on what all the puzzles' answers were supposed to be, so the next step was to arrange everything's state to produce all of those answers. And because it's Mystery Hunt, there was another more complicated metapuzzle after that that used all of that information along with the answers from the factory puzzles, but that's not really relevant to this post since it was a more conventional meta.

This example is kind of singular in the puzzlehunt world- token-based metas aren't very common, and when other online hunts have interactive elements they're usually on a much smaller scale. Next time I'll go over to the "information" side of my definition, which lines up more directly to the standard puzzlehunt answer-based metas. In games those don't show up as much in systemic games like the examples I've discussed here, but do show up in more adventure-style, riddle-based, or otherwise "messy" games.

You should also read:

Musings on metapuzzles, part 6: Non-spoilery summary/conclusions

The Tunic post might come later, but for now I'll try to put together some sort of takeaways that don't spoil stuff:

  • The definition I'm working with for metapuzzle is a puzzle that brings together information or resources across a broader scope than a single puzzle.
  • In a puzzlehunt, information usually takes the form of puzzle answers, which can be combined either on their own or with some additional information (a shell).
    • In a puzzle game, the information can take many different forms (codes, rules knowledge, lore/story knowledge...), so there needs to be a way to determine what's even part of the metapuzzle.
  • In a puzzlehunt, resources are usually more abstract like a count of puzzles solved, or something other than the answer that you're given as you solve each puzzle.
    • In a puzzle game, that can be something like an object in the puzzle's rule system like "you get a block in the overworld when you solve a puzzle", or a more complex way to affect the metapuzzle's state.
  • Puzzlehunt metas are usually solvable without solving all of the puzzles that feed into them, but the more you have the easier they get.
  • Looking for commonalities is a common skill in puzzlehunt puzzles, and doing so in a puzzle game can be a first step in its own metapuzzle.
  • Breaking the abstraction layer and having the meta more deeply interact with individual puzzles makes design a lot more complex, but oftne makes the experience more cool and memorable.

Thanks for reading!

Continue reading...