Sam DiBella is a freelance editor and designer living in Brooklyn, NY. He grew up in Rhode Island and studied philosophy of language. He spends his spare time designing chat bots and convincing anyone who will listen (see: the beginning of this sentence) that they really should be worried about Incompleteness.
The presence of meaningful choices or, barring that, the illusion of meaningful choices has long been a considered a solid standard for a successful narrative game. Normally, developers do this by creating multiple possibilities for the story to follow, allowing player actions to alter the course of the story. Instead of creating an enormous amount of possible states, Her Story, a story-based game by Sam Barlow, experiments with allowing players instant access to its entire story, provided players use the right search terms. Her Story’s structure, and how it differs from other narrative games, is the key to understanding how Her Story functions as a successful narrative game. To do so, I’ll have to explain the computational concept of the finite state machine, why it is a good model for narrative games, and how Her Story’s state machine differs from those of interactive fiction. Based on Barlow’s personal work in interactive fiction and the genre’s place as the earliest style of narrative video game, I will stick to comparisons between Her Story and interactive fiction (IF).
During the design process, game designers use conceptual tools to model game states, allowing them to track all of a game’s moving parts and search for areas of improvement. One of these tools is the finite state machine. Finite state machines are a conceptual product of computation theory. Computation theory studies the solvability of problems and the efficiency of algorithms by using models of varying levels of expressiveness. Finite state machines all have complexity far below the Turing machine that all modern computers are based on. Simplicity is not a weakness, in this case. In game design, finite state machines are a valuable way to visualize a game’s internal code for analysis.
Fig. 2 (left) Warren Robinett’s diagram for dragon behavior in Adventure on the Atari 2600
Fig. 3 (right) Warren Robinett’s map of rooms and their connections in Adventure.
A simple example of a state machine is a door. Loosely speaking, a functioning door only has two states: open and closed. Every human interaction with the door will result in the door being opened, closed, or left alone. We could write a single day in this door’s history as a set of actions: open, walk through, walk through, close, walk away, open. We can draw a diagram to represent all the possible states and transitions between states, and it would look like Fig. 4, where the circles indicate states, and the named arrows define allowed transitions.
Fig. 4 Simple Door State Machine
There are a couple of assumptions for Fig. 4 to be an accurate model of a door. First, the door must have some initial state. Second, no one attempts any “unauthorized” state transitions. Third, there are no other states for the door to be in. It’s either open or closed, and any middle state is excluded. Fourth, commands are discrete actions that are mutually exclusive to one another. You can’t open and close a door at the same time. Finite state machines have only a limited grammar of accepted commands, and as new states and transitions are added, it becomes exponentially more complex. Depending on the complexity of the object or situation, the finite state machine will be correspondingly complex. An accurate model of the real world would be impossible, but the simpler worlds of games can be represented with a state machine.
Fig. 5 More Complex Door State Machine
Mechanisms like the finite state machine can be used to track various actor’s behaviors and AI, as in Fig. 2 and 3 from Warren Robinett’s diagrams for Adventure show, or they can be used to represent the narrative and spatial structure of a game, as Twine’s development environment does. In all these cases, the action performed by a player is different (touch door, move joystick, click browser), but the player’s input results in a change in the underlying finite state machine. With interactive fiction, state machines are most often used to represent the geography of the game, the rooms and the paths between them, as well as gates locked by puzzles.
Fig. 6 Twine grid for porpentine’s With Those We Love Alive
Looking at a diagram of a game’s underlying states, like a dialog tree or a Twine grid, makes the possible interactions clear, but players often don’t have access to that information. During the course of play, a player is trying to create a mental model of the game’s finite state machine. Depending on complexity, design, and player skill, the representation that they create will be more or less accurate to the game’s actual system. For a narrative game, even if there are multiple endings, the manner in which you arrive at each of them is predetermined. Sometimes this means answering questions in a certain way. Sometimes this means having some score or attribute in a particular range of values. In any case, there are a certain number of endings, good or bad (accept states), and the player is encouraged to reach one that they think best represents how they want their fictional avatar to be.
The pleasure of a well-designed interactive fiction game comes from learning how to divine a designer’s intention through the puzzles and text presented. It’s a delicate balancing act that the best IF manages to deliver through effort and craft. If the design fails, the player has to discern a solution from out of the blue. The player has to use their understanding of the game’s internal logic to reach the solution without access to the full state machine. If the game doesn’t provide adequate hints or isn’t internally consistent, the solution seems detached from the rest of the game’s logic.
A walkthrough of parser-style games provides clues about their underlying code. Interactive fiction uses a parser to break down player input into strings of characters, checking those strings against the parser’s vocabulary. Parsers have only a limited ability to process syntactical units and to recognize words. As a result, interactive fiction games have a walkthrough of simple commands that constitute their solution. You can learn the solution to games like Anchorhead without even reading the game’s text. This is confusing to reconcile with the story-driven nature of interactive fiction games. From the viewpoint of the state machine, all of the artistry of the author is moot, provided the user enters the correct command. Even if the player deviates from the perfect transcript, any “successful” play through will contain the string of necessary turns, plus some superfluous inputs. Barlow explored this idea in one of his previous titles, Aisle, an interactive fiction game with over a hundred endings where the player is only allowed to play a single turn.
Fig. 7 A Puzzle State Machine
Her Story does not suffer from that problem. Her Story uses the logic of the Google search, which should be an intuitive concept for the contemporary PC gamer. At any given time , the player will always know their options, large though they may be. Her Story presents a Windows 95 era OS, with a sparse desktop decorated by a British county police department wallpaper. When it begins, a program is running, an archive search engine, and the command line has MURDER typed in. When you hit enter, all the clips with dialogue containing the word “murder” appear on the UI. That is all the pretense given to you. You enter terms into the transcript search engine, and short video clips from seven different police interrogations of the same woman are returned to you. To slightly hinder your search, only the first five clips in the chronology are given to you. As you watch the clips, the woman’s statements begin to contradict one another, subtly and directly, and alternative truths swim into view. You pick the threads you deem significant and follow up via the terminal the game provides.
Fig. 8 Her Story’s UI
Instead of gating areas off and having the player discover the game’s internal grid structure, Her Story gives the player a skeleton key with too many locks to choose from. In this sense, it has more in common with a toy than a game. Will Wright once described his sandbox games, The Sims and Spore, as toys for their ability to encourage open-ended play. Compare friends casually playing catch with a baseball to those same friends playing catch and keeping track of how many times each of them drops the ball. There’s no change in the materials or environment, but with the inclusion of a scoring mechanism, there is a change in the player’s behavior. The play now has purpose. The baseball is still a toy, but to use it in a game a player has to ignore all of the ball’s possibilities and focus on a single game. As Wright puts it “[A] game is a subset of the experiences you can have with [a] toy.” As a toy that contains a story, Her Story provides more room for player interpretation and play than a rigid narrative would. Toys encourage unguided play. Just as the archive of an author, their notes, drafts and journals, can serve to complicate a straightforward reading of a magnum opus, the toy has the benefit of being inexhaustible. There are heuristics to its behavior, and its possible states are just as finite as a game board’s, but the player can play with it according to whatever internal laws they deem necessary. Rather than giving a cohesive message to digest, Her Story is, as Barlow put it in his article Break My Game, “A game that works by — that benefits from — having the player break its narrative.” It’s just as valid to play Her Story entering words alphabetically from the dictionary, or by using random strings of characters. Those are valid options available to you, given that this strategy will eventually result in clips being returned.
Fig. 9 Her Story Screenshot
Rather than requiring the player to follow the standard branching tree structure, Barlow allows the player to reach any part of the story at any time: “By allowing you to interact by Googling keywords the player sees the story through subsets, slices of the whole (unlike a sculpture, where it is possible to sense the entirety of the sculpture, or at least its bounds). And what is lovely about this is in a sense the player’s input can become thematic, or symbolic.” (Barlow. A Magical Incantation: Playing Games with Words – HER STORY) Because success, as normally defined in a linear narrative (finding new content), is so easy to reach, even by purely mechanical means like picking words at random, the player is free to read meaning into how they managed to get there.
|Anchorhead – Day 1 walkthrough:||Her Story – My Search History:|
|SE. Push can against wall. Climb can. Up. West. Push play. East. Look up Verlac (2p). West. Unlock Door. West. West.West. NW. West. Read book. Show keys to Michael (2p). East.SE. East. South. South. East. South. SW. NW. Unlock door. North.||attic (12), dollhouse (5), doll attic (1), her story (3), ginger (1), beard (1), mirror (11), silver (3), eric (14), diane (5), the (164), florence (5), last week (3), hannah (18), my name is hannah (2), eve (7), my name is (4), tattoo (4), palindrome (2), dream (2), ouija (2)|
Fig. 10 Comparison of player input in Anchorhead, an IF game, and Her Story
Looking at my search history in Fig. 10, I can instantly pick out the particular moments that lead me to choose one term after another. After hearing about a doll house from the “attic” videos, I tried to find more about the toy, with only one or two new results. At that point, I had an epiphany that the title of the game should probably be used. No matter when I typed in “Her story” to the terminal, it would still give the same results, but its location within the order of commands instantly gives that turn an additional contextual meaning, derived from my actions.
Fig. 11 Example Her Story Turn
Her Story is, in many ways, a game about reflections. The screen you view the story through regularly flickers, and shows your fictional visage in the darkened glass. The woman’s monologues have a heavy motif of fairy tales, of mirrors, of copies and doubles. Sam Barlow has shown one possible way to make a game’s story intellectually, although not literally, compelling. Games like Bioshock or The Stanley Parable experiment with ideas of player agency by suggesting its possibility and then taking it away. As you search the files of Her Story, it is up to you to bring your thought process to the game. The game won’t force you to pay attention, or to find the “right” clues, and in the end, the clips you see and the order you see them in is not a result of a specific plan Barlow had for your experience. He made the possibility space of the game and the tool for its exploration, but because that tool is language itself, he did not have to predetermine how you approach the game’s single problem. Contrary to the idea of a branching story, Her Story has one timeline, with no deviations. Yet Her Story still has the player making choices that matter, and those choices take place on the level of how the player accesses the story. Without looking at the finite state machine that undergirds Her Story, its approach to narrative couldn’t be formally distinguished from a game with a branching path, or even a game with a linear timeline that the player goes through chronologically. Her Story’s method is a way of shifting the burden of interpretive vision onto the player, but one that works.
Barlow, Sam. Aisle. Z-machine, 1999.
Barlow, Sam. Her Story. Steam, 2015. 10 Aug 2015.
Barlow, Sam. “Break My Game – HER STORY.” HER STORY. N.p., 10 July 2014. Web. 10 Aug. 2015.
Barlow, Sam. “A Magical Incantation: Playing Games with Words – HER STORY.” HER STORY. N.p., 31 Oct. 2014. Web. 10 Aug. 2015.
Crecente, Brian. “Maria Montessori: The 138-Year-Old Inspiration Behind Spore.” Kotaku. N.p., 29 Mar. 2009. Web. 02 Nov. 2015.
Gentry, Michael. Anchorhead. 5th ed. 1998. Web. 11 Aug 2015.
Robinett, Warren. “Adventure as a Video Game: Adventure for the Atari 2600.” The Game Design Reader: A Rules of Play Anthology. Comp. Katie Salen. Tekinbaş and Eric Zimmerman. Cambridge, MA: MIT, 2006. N. pag. Print.
Sipser, Michael. Introduction to the Theory of Computation. Boston: PWS Pub., 1997. Print.