Musings on metapuzzles, part 2: Puzzlehunt metapuzzles
If you're familiar with puzzlehunts, this one might be a little redundant. If you aren't, here's a pretty detailed post with an introduction to the sorts of puzzles you see there.
I'll be spoiling an old puzzle of mine: Wario, from the 2021 Microsoft Puzzlehunt. If you want to try solving it yourself, you can check the final answer here, or the individual minipuzzles at 9-Volt, Ashley, Dribble & Spitz, Jimmy T, Mona, Orbulon. There's a full solution here.
So I'm defining a metapuzzle as "a puzzle that requires collecting information or resources across a broader scope than a single puzzle," and in a puzzlehunt meta the information usually takes the form of answers. The Wario metapuzzle's what's often called a "pure meta", in that the only information you have to work with is some flavortext and the answers to the individual minipuzzles: PLATEAU, SILENT AUCTION, COAUTHOR, APPLAUD, GAUDY, DEBAUCHERY. The first step here is useful in a lot of puzzlehunt puzzles, but especially pure metas, finding something in common. Wario's reputation for greed and the flavortext referring to his (at the time) latest game, Warioware Gold, are hints that the AU appearing in all of them is important. Then you need to find a way to use that connection to pick out a letter from each answer and an order to put those letters in to get your answer.
The other common type of meta is the "shell meta", where the metapuzzle comes with some information you use along with your answers. For instance, if I'd put this at the bottom of the Wario puzzle, it would've been a shell meta instead:
The obvious logistical advantage of shells is giving more flexibility to the author. This version has the same mechanic and answer as the original, but if I rearranged the rows or circled a different box I could make the answer almost anything that uses a letter from each answer. But they also provids more information to the solver- if you hadn't figured out the commonality, knowing that you'd normally put one letter in each box, but none of the answers have exactly 4 or 12 letters, might be a nudge in the right direction.
Relying on answers as the chief source of information is convenient for the author overall because it creates a layer of abstraction around the individual puzzles. You don't need to know about the details about the individual puzzles when making them, as long as it provides the correct answer. Often, a hunt will design and test the metapuzzle first and create the list of needed answers, then hand off the answers to individual puzzle writers.
Can a puzzlehunt meta rely on collecting resources instead? There's a less commonly used term called the "token meta", where when you solve a puzzle you get some other thing, a message or a PDF or an image puzzle piece, and you use those things instead of (or in addition to) the puzzle answers. And, yes, technically all these meta tokens all constitute information, but since the information isn't coming from the puzzle itself there's another layer of abstraction: a meta creator can create a puzzle without knowing anything at all about the puzzles other than how many there are.
One other common consideration in hunt meta design is partial solvability. Most puzzlehunt metapuzzles are solvable without all the data or tokens from individual puzzles, if you get GOB_ET and can guess the last letter. This makes individual puzzles/data points less likely to be a bottleneck, of course, but also allows for backsolving. Knowing that the sixth answer has LAU in it might help you go back and solve that puzzle, or even guess its answer with no idea how it works.
Next time I'll look at how some of these concepts apply to video game metapuzzles.