March’s patreon project, Don’t Mind my Apocalypse Head, was a short parser game written around a fairly disturbing dream I had. If you haven’t played it, it’s fairly short and I suggest you check it out before reading on.

Apocalypse Head is superficially structured a lot like a conversation game, but this is a misdirection; the conversation that goes on throughout the game is totally linear, and the player’s only choice is on what parts to be present for. I wanted to build a game where interaction was mostly about the protagonist’s mental state; choice revolves around managing exposure to distressing things

That meant building the game around a very different set of interactions than normal. Most parser games treat EXAMINE according to a certain contract: It’s a neutral action that’s not supposed to affect the game world; at most, it advances the clock. Sometimes this is subverted, but Apocalypse Head is built entirely around repurposing EXAMINE. Another expectation that really comes from games in general is that action happens only when the player is present. There’s a long history of games based around schedules and even real-time clocks, but usually they’re constructed around the idea of trying not to miss important interactions; Apocalypse Head is an inversion of that, it’s built around trying to avoid distressing interactions.

These inversions are kind of deliberate, kind of not; they emerge directly out of trying to build a game around anxiety, and thus around the avoidance that comes with trying not to overwhelm oneself.

In puzzle parser games, there’s a long history of games with systemic mechanics that support a specific mode of interaction, but this is equally valuable for puzzleless parser stories. Building a story around a specific set of interactions has a number of advantages.

It acts as a sort of leveler. It’s difficult now, when writing parser games, to balance the expectations of players across the spectrum of familiarity with parser. Your game might be someone’s first piece of parser IF, but then again it might not. A game with interactions designed specifically around itself, and its own story, can’t rely on preexisting familiarity with parser conventions. This forces one to think hard about what’s confusing and how players are being led into the game.

Second, it opens up avenues for story choices that are discoverable, but not necessarily placed directly in front of the player. This has always been touted as an advantage of parser; the ability to let players make their own leaps of logic. In practice, the opacity of the typical parser interface, and the linearity of most parser games, means that this is rather underutilized.

Apocalypse Head has seven endings, which are never presented as a choice to the player, but rather are the outcomes of several different systems that are racing together and interacting throughout the game. The core set of interactions (examining, waiting, eating, moving to and fro from the dinner party, trying not to get bored) is the player’s tool for manipulating those systems.

One risk with games built around multiple endings is that it’s not necessarily totally clear to players how much variation there is, and even whether or not there are multiple endings. In Apocalypse Head, I do enumerate the endings (Alphabetically, Drakenguard-style), which is a bit of a blunt solution to this problem. But then you have the twin problem of players not really understanding how to reach alternate endings.

Systems give players a thread to pull on and follow towards different outcomes. They also enable the perverse pleasure of watching things degenerate; a lot of the endings in Apocalypse Head are “bad,” they stem from actively pursuing what the game is notionally about avoiding.

This is a thing that is sometimes not appreciated in discussions of choice in interactive stories: sometimes you can let the player pick between a good and a bad decision, and have those two options stand as relative equals, because you’ve convinced the player to value outcomes based on how interesting they are narratively and not based on how good they are for the player character.

Story-as-system pushes in that direction, because systems are naturally taken as toys; they invite the player to turn it this way and that to see how the pieces fall into place. And this is where tightening the focus into a few specific interactions can help, because they make it clearer, more direct, how that toy is meant to be manipulated.

In Apocalypse Head, a trick I found useful was collapsing actions into one another. There’s a very small “puzzle” that is solved with an item found inside a container. The puzzle is solved as soon as the item is in view of the player, which means that, for many playthroughs, opening the container doubles as using the item. Using the environment to suggest an obvious, straightforward action, and using that as a proxy for a more complex and nuanced one can be used to sidestep the problem of verb-guessing and other parser-related confusion.

This is a fun trick: Systematically cut away any interaction that has nuance or complexity to it. If you design around it, you can actually build a game that only needs VERB NOUN commands, and maybe for some stories that’s actually more desirable than the long drive towards parser sophistication that has been the trend for the last 30 years. In Apocalypse Head, the player is quickly taught that they can’t interact with the conversation. The game doesn’t quite verge into outright limited parser territory, but a good way of keeping the scope on a parser project under control is to swerve into concrete interactions that are obvious to express (EAT FOOD) instead of complex ones that include ambiguity (TALK TO JULIA ABOUT HER SHITTY ATTITUDE).

In other words, you can use the medium-sized-dry-goods class of interaction as a proxy for behaviors or values separate from just the mechanical manipulation of the world model. In Apocalypse Head, eating food is used as a proxy for behaving like a normal human being at dinner. If you’re struggling with how to express a nuanced interaction in a parser game, consider hanging it off a simpler, more direct one instead.

The last piece of the puzzle, in Apocalypse Head, is using the player character’s interiority to drive interaction. Apocalypse Head features a disobedient protagonist; because of the way the game is written, the player is incentivized to avoid interaction. To counteract this, I made repeatedly taking the same action divert towards looking at something reflective. Having the game’s systems incorporate some pushback against the player helps keep it from feeling too much like a toy.

This also helps demonstrate the mechanics of the system to the player. Having something happen on its own is one of the most direct way of suggesting to players how they can make it happen themselves.

This is all very similar, again, to building a puzzle-oriented parser game around a systemic mechanic. But the goals are a little different; there’s no leap of logic required of the player in Apocalypse Head. It’s designed so that all of the avenues of interaction suggest themselves, and players can pull on those threads as they choose.

A good systemic mechanic is, as Emily Short puts it, juicy; it has a lot of design space, and a lot of room for variation and interaction.

Systemic story interactions need to have directness: they need to be easy for the player to express to the parser, you are hanging nuanced story beats off simple behaviors. And they need to be loud, they have to announce themselves to the player and be insistent about that.

Ideally, on a narrative level, they need to be ambivalent; both sides of the interaction, doing the “right” and “wrong” thing, need to be interesting to the player. Often, games try to make choices interesting by giving them an ethical or strategic layer, but I believe that you can also do that just by having interesting failure be an option. All of the interactions in Aisle that lead to unhappy or dissatisfying endings are just as desirable, to the player, as the ones that lead to more conventionally satisfying conclusions, narratively.

Game narratives are often about building systems of valuation that spur the player to want one outcome over another, but what if the game’s value system is implicitly ambivalent? What if you give equal narrative weighting to “bad endings” and attract the player towards them?

In horror, it doesn’t have to be that there are no wrong answers; the wrong answers can have a morbid fascination to them. Sunless Sea does this masterfully, giving the player plenty of options where being morbid or horrific is its own reward. It’s like picking at a scab, and Apocalypse Head is a very minimal exploration of that; you could go much, much deeper into it.

This post is brought to you thanks to my Patreon supporters. Special thanks to Emily Short, Kevin Snow, Liza Daly, and Doug Orleans.