Making Puzzles For Games I: Vectors of Interaction

In case the content of this blog did not demonstrate this, I am deeply interested in all facets of game design, a domain that can be messily divided into video game design and tabletop game design (there are a bunch of other areas of knowledge too, like for gambling games, sports, gamification, and surely a bunch of other niches I'm unaware of). There's a lot of shared thought and space between video games and tabletop games, especially since video games from near inception to today have been using ideas from its older brother, a translation of ideas that tabletop games have started to reciprocate recently. There is, however, one idea that does not make the jump across that particular threshold so easily: the puzzle.

Puzzles, which I'll vaguely and probably incorrectly define as "a challenge in a game which is overcome with knowledge, logic, or lateral thinking instead of strength, skill, execution, or numerical advantage", are pretty tightly linked with both tabletop games and video games and have been since both mediums' inceptions. Generally speaking, people like a good head-scratcher, and a puzzle is a good way to provide a change of pace from the regular core action of a game.

I've seen a lot of puzzles in video games, and a lot of puzzles in tabletop games, and I can tell you that while puzzles in video games range from awful to transcendent, puzzles in tabletop games, in my experience, are almost unilaterally awful.

Again, some video game puzzles are definitely also bad

They're awful in a specific way, too: the puzzle is presented, players spend too long bashing their head against solutions, usually in a solution that too closely resembles brute force, until finally the energy in the room is so thoroughly sapped that someone looks at the Game Master and goes "fuck it, what is it" and the whole thing is just resolved by GM fiat or by ten goblins falling out of a ceiling tile in a desperate ploy just to create some semblance of forward momentum again.

How can it be that even the most action-packed video games can successfully include some relatively obtuse puzzles, and sliding a single puzzle into even a slow, investigative tabletop RPG session collapses immediately? I don't think it's all bad puzzle design on the part of tabletop RPG designers (although they're certainly not fully blameless), I think that there are fundamental differences in the two types of game which make one have a much easier time with puzzles than the other, the largest of which is going to be what I call "vectors of interaction".

When I refer to a "vector of interaction", I am talking about a single way in which a player can in some way affect the world of the game. You can think of these as "verbs", although I think they're a little bit broader than that. In Super Mario Bros, the vectors of interaction include moving left and right, jumping, crouching, and shooting a fireball. These are the fundamental vectors.

Since the player exists outside of the game and the puzzle exists within it, it is fairly reasonable to say that the only way a puzzle can be solved is by applying a certain set of vectors of interaction in a certain order. To open the locked door, you must move (a vector) to the enemy that has the key, attack (a vector) them until they die, pick up (a vector) the key, move (same vector) back to the door, then unlock (last vector) the door.

One of the strongest tools that puzzles in games have in creating the puzzle is by carefully curating the player's vectors of interaction. Games frequently set up puzzles by teaching the player what their vectors are, and then conditioning the player to recognize subtle hints which indicate to them what vectors might need to be used where. This can be very obvious (when you see enemies, you probably need to fight or kill them), or more subtle (whenever you need to find a ladder in The Last of Us, the ledge that you will use that ladder to access will usually be painted yellow). The Witness, famously, built an entire game, with tons of puzzle variety, around a single vector of interaction: guiding a line along a grid to a goal.

Games can also build puzzles by subverting expectations regarding vectors of interaction, perhaps introducing new ones, or making the player find new ways to use old vectors. For an example of the former, The Legend of Zelda: Phantom Hourglass has several puzzles that are solved by using vectors of interaction that the player might not even initially think to use, namely the DS's microphone and ability to close. The ending of Portal 2 does the latter exceptionally, by leveraging a gag from hours ago that the portal-friendly walls of the Aperture testing facility are painted with moon rock, in order to get the player to shoot a portal at the actual Moon.

Here in vectors of interaction we have a massive gulf between video games and tabletop RPGs, one which at first blush sounds like it might be in favor of tabletop RPGs: video games have an order of magnitude fewer potential vectors of interaction than a tabletop RPG. Every gaming platform has a limited number of control inputs, which can only be reused and combined so many ways before a game's controls become cumbersome. As a result, players' problem solving toolkits are dramatically limited in video games. You cannot hash out a verbal truce with Bowser at the end of Super Mario Bros. You cannot get through a locked door in Resident Evil by breaking it down.

Tabletop RPGs, as a (typically) completely spoken-word storytelling medium where the game exists in players' boundless imagination, has no such limitations. Players' vectors of interaction are functionally infinite: any action they can conceive of, at least any action that doesn't conflict with the fundamental rules of the game world's reality, is valid. An enemy can be shot, stabbed, talked down, tied up, avoided, thrown down a big hole, brought to exhaustion, brought to boredom, outlived until it eventually dies of old age, anything at all.

Admittedly, point-and-click adventure games actually have a similar problem, with their wide gamut of verbs available to the player and an unclear delineation between foreground and background

The problem is that infinite vectors of interaction equate to infinite potential actions at any given moment when solving a puzzle, which equate to infinite red herrings. When I as a player can talk to every single inhabitant of a town infinitely while investigating a murder, as opposed to a video game where they will inevitably run out of dialogue, I have the room to explore red herrings and dead ends forever. If they truly don't know how to solve a puzzle, and try to simply brute force it, they have infinite options to iterate through until they find the correct one.

Thus, a potential avenue to developing a puzzle that players will solve in a finite amount of time, ideally before everyone at the table spits bile at the mere mention of the word "puzzle", is the careful curation of vectors of interaction. By limiting the possibility space of a puzzle's solution, players can reason about a more limited set of options and, hopefully, come to a solution in a limited amount of time. I'll be discussing the ways you can do this in the next post in this series, but as a teaser, there are at least two strategies you can follow to do this, you can either add an additional layer of interaction to the game, limiting players' available vectors to a more manageable subset, or you can use signalling in order to let players trim down vectors themselves.

The other major solution that comes to mind, one which I'll also be covering, is what I like to call Gordian Knot puzzle design. This is to say, create a puzzle without a solution, at least not one in mind, and trust in the players' ingenuity, combined with the infinite vectors of interaction available to them, to produce a solution without you having to think of it beforehand.

These techniques are designed to either constrain or work with tabletop RPG's extreme breadth of vectors of interaction, and hopefully utilizing these techniques will help produce RPG puzzles that are actually enjoyable, for once.