Mystery House and the Trouble with Parsers

by Matt Brennan

Mystery House, an Apple II mystery game that roughly follows the plot of Agatha Christie’s groundbreaking 1939 murder mystery novel And Then There Were None, is an interesting hybrid of the many concepts of contemporary video games that were being thrown around at the time of the game’s development and release in 1980. The game is ostensibly a text adventure in nature, relying on the player to input commands via a text parser as in text adventures like Adventure or Zork, but with primitive graphics similar to those seen on Atari 2600 systems of the time. This was a great feat for game development, and doubtless Mystery House has its place among the original cornerstone video games that shaped the medium into what it is today. However, it does not hold up in any real regard when returned to in the present day.

Mystery House is at its core a balancing act. Its developers deserve their flowers for what they were able to accomplish with it, combining graphics and story-driving text into a single product, but at its core is a balancing act between these two main components. The text and image don’t operate independently of each other, but they don’t always work together either and can create some nasty overlaps that affect the player’s experience. In addition to this is the text parser, which is regularly a hindrance to the player and nearly impossible to move around with, let alone solve a string of murders.

The game’s ambition and scale is obvious from the start, and the inventive use of graphics puts these qualities on full display, but that just makes it all the more frustrating when the player is stuck wrestling with the parser for minutes on end just to accomplish the slightest interaction with the world of Mystery House. Whether it’s inspecting a crime scene, picking something up, moving between rooms (and even just walking forwards is a challenge), or most infamously attempting to turn on the water, the terrible text parser will find ways to rebel at every turn and turn an experience that should be awe-inspiring for its accomplishments into a downright miserable slog due to its shortcomings. (Worse still, according to the game’s guide on the microm8 emulator, the way I experienced Mystery House, the inept text parser was packaged as a feature of the game rather than a bug as it should have; the game pats itself on the back for its immense difficulty, neglecting the fact that so much of the difficulty with the game is artificial and a result of Mystery House’s chief misuse of the adventure form.)

From a technological standpoint and a development perspective, Mystery House is still a triumph in every sense of the word. While this does make the glaring issues that make the game a difficulty to play more glaring and somewhat ironic, but the game succeeds in its strange marriage of text-based game design and graphics in a way that would ultimately set off a whole genre of adventure games after itself. While Adventure and contemporary adventure games could describe the cave or the dungeon using its words, Mystery House was alone in its capability to take the plot of an adventure and show it directly to the player as a picture.

However, again, the parser comes back to center stage, the only way to control the player character fighting the player every chance it gets. Mystery House doesn’t have many places in which it falls short. The display is often perfectly fine, and while having text block the action on the screen can be a pain, it’s easy to fix. The graphics are excellent for their time, especially when deployed on the scale that they are in Mystery House; the fact that they were able to create such stellar graphics for a game that was for all intents and purposes text-based and make it work deserves commendation. The story itself is excellent, a loose take-off of Agatha Christie’s And Then There Were None that plays to its strengths to create an air of mystery and danger within the game that gets genuinely tense as the bodies start to pile up. The only significant issue with the game is the controls, the text parser—but that’s a backbreaking issue considering how crucial the parser is to the game experience.

Mystery House is a fascinating game on a truly ambitious scale for its time and level of sophistication, sunk by one single fatal flaw. Looking at the game around this flaw shows that all elements are clearly there for an excellent adventure, as the game has a lot of good ideas and executes them all well. Sadly, and let this be a lesson to aspiring game developers everywhere, the player’s experience is quickly derailed by a control parser that regularly feels unfinished and unable to handle the demands of the game.

Text-Based Games and the Trouble of Simplicity

Article by Matt Brennan

Last week, I played through the game Colossal Cave Adventure for class, and in my playthrough and the subsequent discussion we had as a class, one specific topic popped up with regularity: the text parser. The text parser was no small innovation for its time, but returning to it nearly 50 years later was an exercise in frustration. The parser was extremely simple, as it had to be for its time, and thus only accepted a very small—and very specific—list of inputs. This confused every player, myself included, and led to the game becoming a tug-of-war between the player and the controls, with the game itself, while never quite becoming secondary, taking a very clear back seat to the struggle to play at all.

For movement, the parser of Adventure accepts cardinal directions (though not the command to “go” or “move” in these directions), as well as up and down. Everything else gets more arcane from there, as the parser allows for “get” and “take” but not “use” or “investigate.” “Look” is allowed but only gives a description of the room around the player character, and “throw” does the same thing as “drop,” preventing any chance at lobbing objects at enemies as weapons. Eventually, it occurs to the player that their best bet at succeeding is to input only the exact thing they want to interact with and no other words, and allow the game to fill in the rest of the data. This works more often than not, which is its own issue. In a game like Dungeons and Dragons, which was brought up in discussion and was a clear inspiration for Colossal Cave Adventure and every other early text adventure of the day, it’s a reflex for the players to describe what they do in great detail so that the game’s adjudicator, its Dungeon Master, knows exactly what the players want to attempt and how to judge them on the fly. That level of detail, necessary in D&D, becomes an active hindrance in Adventure, and the player must rely on simple, often one word, inputs to move the game forward.

The important distinction between a game like Dungeons and Dragons and a game like Adventure is the level of freedom in both. D&D is nearly pure improv; the players work together and keep each other focused and up to speed on how the game is working at any given moment, but nothing is set in stone in terms of story, and anything can change on a whim or an unexpected die roll. Adventure, for how much it’s inspired by D&D, is nearly the exact opposite; the single player is completely alone, has to figure everything out themselves, and is working to advance the single plot down its defined tracks until it reaches the ending, which is always the same. If the plot of a D&D campaign is an adventure, the plot of Adventure is a train ride with a sticky handbrake. There are interesting and clever moments—for instance, the player needs to keep their inventory clear, but if they don’t set down a porcelain vase in the same room where they set down a decorative pillow, the vase will fall and shatter—but those are hard-coded into the game, not the result of a clever piece of judging by another person. The mechanics can be interesting, but there’s a certain attempt to make Adventure something it isn’t, and the text parser makes this abundantly and uncomfortably clear.

Not to say I didn’t enjoy playing Adventure. I thought it was a very solid game for its age, and the fact that the most frustrating thing about a game over twice as old as I am was that its controls were counterintuitive does say a lot of good things about the game. A lot of it holds up really well, but at every turn when I played, the text parser became a more and more pressing issue. Thankfully there were a handful of passwords to skip most of the long walking sections (where the game has the most chances to screw up and dump the player somewhere in an endless forest), but that was only a workaround, hardly a solution. A text-based MUD I have some experience with, Discworld MUD, also uses a text parser to control the game, but this parser is much more sophisticated and, more importantly, flexible than the parser used in Adventure, and as a result the game experience feels much more akin to D&D and other roleplaying games than to the old text-based adventures we played for class. While this is certainly an unfair comparison between the two, I find it fascinating how far the technology for text-based games has developed even amidst the advent and growth of graphical games. It’s a far fairer comparison between the two, I think, to bring up how beloved and propulsive for the genre both games have been.