You can read more of my writing over at the Meeple Like Us blog, or the Textual Intercourse blog over at Epitaph Online. You can some information about my research interests over at my personal homepage, or on my profile at Robert Gordon University.
You can find part one of this series here.
There still exist vibrant communities devoted to building new games in the fields of text games, including a burgeoning renaissance in the field of interactive fiction (Montfort and Short, 2012). Prospects with the world of multiuser dungeons and their variants have been grasped less enthusiastically, but there are still new titles being developed (such as my own Epitaph Online) and existing titles continue to be enthusiastically expanded and refined even twenty or more years since their inception. The health of a hobbyist scene can be partly measured in terms of the external activities occurring around it, and text games are the subject of several modern academic papers (c.f Heron, 2013; Townsend and Heron, 2013; Montfort, 2013) and are a significant subject of study in the fields of the Digital Humanities and Electronic Literature. Likewise, several electronic magazines such as Imaginary Realities5 or the Society for the Protection of Adventure Games newsletter6 revolve around commentary and analysis of the area. As such, the lessons of text game puzzle design are still of contemporary interest as well as historical relevance.
We can see in the style of the text game a wide range of ways in which puzzles have been presented. Some of these are so common as to be considered almost cliché, and their reimplementation in different contexts is something most creative developers seek to avoid. These are the standard set of ‘logic puzzles’ which may interest and excite players new to puzzles generally, but merely elicit a weary sigh of boredom for those who have encountered them before. These include mazes; stock puzzles such as the Towers of Hanoi or the three and five litre water jugs; or common, well-known riddles. As Dan Shiovitz remarks in Scott’s (2011) text game documentary:
No-one wants to see the old puzzles again, because we’ve already solved the old puzzles. Like, when you’ve solved a puzzle, you’ve drained all the juice out of it.
And Mike Berlyn adds:
I don’t know anyone who erases crossword puzzles to fill them in again. I really don’t!
Logic puzzles in these games then are ‘fun the first time’, if you’re lucky – rapidly though they become less fun the more often they are encountered. Within a piece of single-player interactive fiction, there is a limit to how much puzzle-space can exist and be realistically and profitably exploited given the large number of games and relatively small number of players. A puzzle may be unique to your game but still utterly derivative when examined in the context of all the other games that your player may have completed. Standard logic puzzles are not just endemic in video games, they’re common in the world outside too. This is perhaps, in part, why modern video games largely shy away from the use of these kind of puzzles at all. Even in unabashed adventure games like Tell Tale Games’ The Walking Dead, puzzles largely consist of ‘finding one object in an obvious place and using it in the one situation where it’s obvious to use it’. Modern interactive fiction tends now to focus on narrative rather than ludic considerations (c.f. Douglass, 2007; Sorolla, 2002; Ryan, 2009). The removal of a commercial incentive means there is no longer a requirement to artificially ‘pad’ game length with complex puzzles to ensure a sense of value for money. However, it is still possible to create an inherently satisfying game experience using logic puzzles– consider the Professor Layton series of games which present (largely bespoke) logic puzzles as obviously artificial barriers to narrative progress. The replayability of a Professor Layton title is limited however because the answer is the same each time. The expectation is that a player picks up the title, plays it to the end, and then doesn’t play it again because it’s already been solved. If a player wishes to re-experience the thrill, they must move on to another title in the series.
The dynamic is different when discussing MUDs because of the different game focus and expectation of persistence. MUDs, like MMOs, are designed for people to engage with for the ‘long haul’. Dedicated players of MUDs may invest years of their life into building up their characters and integrating into a complex social environment. The creation of alternate characters, or ‘alts’ is common – when all the challenge has been extracted from the game for one character, a player will often start again anew to capitalize on their valuable expertise and social networks to experience the game from a different perspective. They will take on new gameplay roles, make different advancement decisions, and so on. As such, it’s expected that players in these games will indeed encounter the same quests many times and ‘solve’ them many times. Rarely do these quests have a purely intrinsic reward for accomplishment – they are ludic elements used to progress game goals. This has created an unusual dynamic in multiplayer games where there is a need to protect the integrity of the game from the quests being too easy to solve.
For a long time, MUDs suffered from the illicit and back-channel trade in what were known as ‘quest lists’. These were comprehensive, sometimes complete, accounts of how every quest in the game could be solved (Drakkos, 2004). They consisted of instructions such as ‘go here, do this, wait for this to happen then type this command’. Armed with a quest list, a player could rapidly advance their character to relatively high levels with no more intellectual investment than the ability to read instructions off of a sheet. Many games worked to ban the sharing and writing of such quest lists, with predictable ineffectiveness. MUDs exist online, and if someone can connect to a MUD they can send an email with a quest list as an attachment. Countermeasures such as tracking player communications or setting up elaborate honey-pots7 served only to create an ‘us versus them’ mind- set in some game communities. For those players who had already successfully completed the puzzles themselves, their own personal quest lists may be of use to them and their alts without implied moral outrage. However, the trivial nature of quest design in many such games means that there’s no need for someone to understand what they’re doing, or why they’re doing it. They just need to follow the instructions that they are given. Within a single player game using a standard walkthrough, this can easily be ignored in terms of ‘you’re only cheating yourself’. In a multiplayer game there is an expectation that there will be a general balancing regarding the time and effort a player invests versus the benefit gained. Quest lists in that respect represent a distortion of the balance curve and thus a potential threat to the game’s integrity and as a consequence, the community upon which it depends.
The type of quest most susceptible to this is what I have termed as ‘syntax quests’. They focus on getting the player to type a particular command into the game in the right place at the right time based on interpreting game output and contextual interaction options. Once that command, or sequence of commands, has been determined by anyone, then no further contemplation need take place. A slight variation to this quest style involves entering a command which may be slightly different for each player, in which case the instructions will need to highlight where the player specific version may be encountered within the game. Being asked by an NPC to repeat a particular phrase for example may have a different phrase for each player, but it’ll likely come from the same source.
Various techniques were used to combat the endemic quest list issue, and none of them had a satisfactory outcome. Discworld MUD, which fought a prominent battle against the issue of quest lists, trialled out a several year period where quests gave no in-game reward for their completion. They were to be completed just for the ‘fun of doing them’. Predictably, quests simply weren’t completed nearly as often and the game underwent a five year period where new quests weren’t developed because nobody was completing them. Within volunteer environments like text-game development, the key benefit developers experience is the feedback they get from players enjoying what they have produced. When that feedback goes, so too does the incentive to develop.
However, battling the power of the quest list has resulted in various innovations within text game puzzle design. Within Epitaph we have had a focus for some time on ‘dynamic’ quests where the full instructions for completing the puzzle are available within the game itself. However, the quest is designed in such a way that the instructions only provide the context for a dynamic logic puzzle with a randomised internal state. One such example is a quest called ‘The Dastardly Deed’ which requires players to quiz NPCs on their backgrounds, relationships and whereabouts in order to solve a murder. Another such quest creates a bespoke substitution cipher and encrypts a piece of standard text with it – the player’s job is to find the cipher keyword which is selected at random from all the room descriptions that are currently loaded in memory. Other quests are implementations of standard grid-based games such as bejewelled, minesweeper and light’s out. The way in which these are to be solved is well known, but the actually completing a grid is a task that requires considerable mental agility.
The puzzle element that is attached to these is to present them both within a narrative context and with a particular win condition. The Bejewelled puzzle for example is used to handle hacking into computer terminals – that provides the narrative context. Bejewelled is a game when it is played for the accumulation of points, but it becomes a puzzle when the task is to accumulate X number of points in Y moves, or in Z time. These titles are an important inspiration for creating dynamic puzzles within contemporary text games – they leverage existing familiarities and skill-sets whilst still providing a challenge for those playing in the new context of a text game.
These type of quests both invalidate the use of set solutions and also help resolve the issue of the ‘juice’ being sucked out of a puzzle. It is true that people rarely rub out the answers and complete the same crossword twice, but they often complete multiple different crosswords and it is in this metaphor that dynamic quests are best understood. However, such quests are difficult to design and much more complex to implement than syntax quests. There is a limited design space too where such puzzles can thrive, and that space also tends to preclude two categories of players – those with cognitive impairments in fluid or crystallised intelligence and those who simply do not enjoy the recreational stressing of their ‘left brain’ faculties.
Another category of quest is what has been termed ‘action quest’ where the goal is clear and the methods by which it should be accomplished are equally so. It is the player’s ability to perform the task under duress and against a time limit that offers the challenge. One such example of this quest within Epitaph is called ‘Rat Maze’, in which four ‘super zombies’ are loose in a small complex. The player must navigate this complex avoiding the zombies whilst hitting a number of switches. Zombies are attracted primarily to noise, but secondarily to the player and they will thus converge on the player’s location if not otherwise distracted. The puzzle then is to work out when it’s best to make noise; what path is best to take through the complex; when to wait and draw zombies towards you so as to clear other parts of the complex; and when to flick the final switch which is at the end of the complex. Flicking the final switch causes the zombies to berserk and speed up rapidly. Here, the syntax is clearly provided but the task is one of constant evaluation of a situation and manipulation of enemy spatiality as well as managing the resources of sound. You have a limited number of sound generation tools; they only work for a limited period of time; and they are distributed throughout the complex and immovable. In many ways, it is this kind of puzzle that is most common in video games generally – the accomplishing of a well-defined task under duress and within time constraints. In this respect, dealing with a boss encounter in an FPS or navigating Mario through a complex topography of platforms are ‘action puzzles’.
However, within text games we have an additional tension – many are drawn to these games for the extremely high potential accessibility (Heron, 2013). Those with visual impairments can use a screen-reader to verbalise the game text, and then they have the same ability to perceive the world as anyone else. Those with physical impairments can find that the slower, more contemplative pace of gameplay allows for restrictions in terms of speed and accuracy to be less impactful on their gameplay experience. Accessibility in video games is generally woeful (Heron, 2012) and a respectful and mindful developer can create a truly inclusive text game design if they are willing to put in the effort (Heron, 2015). This effort can be invalidated by adopting the above variations of the classic game puzzle. Dynamic quests that incorporate grid-based ASCII representations are inaccessible to screen-readers. Dynamic quests that do not may be inaccessible to those with cognitive impairments due to the difficulty of the task and the difficulty of cognitively parsing complex textual state representation. Action quests, with their focus on speedy execution of commands under stress may be cognitively and physically inaccessible. Thus, the unique characteristics of the input and output environment create restrictions on what developers may freely adopt – those who do not care about accessibility can innovate more freely in these areas, but in turn may alienate a group of players who may be one of the most significant in terms of retaining the viability of text gaming as a genre.
These and related issues resulted in the adventure game evolving down two routes – one was towards straight interactive fiction where puzzles were de-emphasised and gradually removed. The second route was to incorporate graphical displays into text parser games – first as mere ornamentation and visual cueing, but later for handling all elements of world representation. Consider Sierra Online’s 1980 game Mystery House, which used wire-frame graphics rendered in real-time to represent the key elements of room inventory as is shown below.
As rudimentary as this was, it marked a turning point in the evolution of the text game. As the graphical capabilities of home computers increased, so too did the graphical sophistication of games. It took some time before the processing abilities of computers truly allowed for instantaneous representation of these kind of game graphics, and so players were often subjected to a noticeable lag in which the game would first draw the vector shapes of the environment, and then (sometimes) slowly colour them with fills. Melbourne House’s 1982 title, the Hobbit, was one notable example of a game that incorporated this regime. One scene from the game is shown below:
Later titles, such as Adventure International’s QuestProbe series would use licenced characters to provide visual annotations of the story. The Incredible Hulk (1984) shows an example of this:
These graphical regimes represented a step away from the pure interactive fiction being developed by companies such as Infocom. That particular company shied away from attempting to add these elements for a long time. They hoped instead that focusing on their titles as literature rather than games would simultaneously allow for an expansion of their artistic aspirations and cement the success of the company when their competitors moved into a different area of the market. They focused instead on building collaborations with successful authors such as Douglas Adams and Robert Pinksy (Montfort, 2013, p267) and in bringing out all of the potential they saw in interactive text. In the end, both groups found themselves consigned to a commercial graveyard as the period 1985-1992 saw most of the big players in the text and graphical text adventure field closed down (Heron, 2013).
Other companies had chosen to distance themselves further than pure parser games with token graphical ornamentation. Sierra’s King’s Quest (1983) used full graphical representation with an on- screen avatar, combined with a text parser, to create the game interface. Music too was added, giving it an additional channel of multimedia representation lacking in other titles. The various incarnations of the King’s Quest series continued to enhance the graphics and sound, as well as refine the parsers and the presentation of the puzzles up until King’s Quest 5 (1990). For this title a ‘point and click’ interface was adopted – bringing the adventure game full circle back to closed parsers. The puzzles represented in these games began to evolve in line with the restriction of scope implied by a closed parser – a limited set of interaction opportunities that can be applied to a limited set of items or environments represented by graphical iconography. This convention became core to many games of the period, with the point and click adventure games of Lucasarts perhaps being the most significant examples of excellent and largely logical puzzle design (Fernández-Vara, 2009). However, for every Lucasarts game there were ten from rival companies where the puzzles presented were often baffling in their lack of intuitive or even retrospectively comprehensible solutions. The ‘hunt the syntax’ problem was replaced with an equally frustrating solution strategy of simply rubbing every item in your inventory against every object in your environment in the hope that you’d find the magical combination that the developer had been considering at the time of writing the puzzle.
As this evolution was occurring, so too was the evolution of the MUD. Many new styles of MUD, codebase and game engine were developed over the ten or so years in which MUDs were in the ascendency (Barton, 2008, p37; Aarseth, 1997, p150-151). Lucasfilm’s early experimentation with Habitat (Morningstar and Farmer, 2008) were interesting but failed to truly marry multiplayer elements with engaging gameplay. Neverwinter Nights (NWN) in 1991 was perhaps the first significant graphical MUD, and it used elements of the SSI ‘Gold Box’ architecture (Barton, 2008, p160-162) in a multiplayer context to create a long-running game for AOL users – those who played would spend $6 an hour to adventure with and against other players, of which there could be many. Each NWN server could accommodate 50 players in 1991, growing to 500 players in 1995. At the end of its run, it was estimated to have over 100,000 subscribers (Brown, 1997). It was mothballed, under protest, as part of a restructuring of AOL’s online gaming provision (Brown, 1997). Other games at the time of NWN’s demise were starting up, including notable titles such as Meridian 59, Ultima Online and Everquest. Each of these at the time was referred to as ‘Graphical MUDs’, indicating the strong perceived lineage that preceded their release (Bartle, 2010; Keegan, 1997).
Everquest in particular was so close to design to the popular Diku MUD games that it was rumoured to be infringing on the Diku MUD game licence (Taylor, 2002) because of common adoption of terminology, textual phrasing and gameplay conventions. Issues of licencing are complex in these kind of environments (Heron & Belford, in press) and the Everquest team were eventually exonerated. The adoption of these conventions is unsurprising considering that several prominent Diku players were part of the team that created the game (Koster, 2009). Meridian 59 and Ultima Online likewise incorporated MUD players and developers on its staff. With them came much of the design expectations and vocabulary of development, and this in turn showed in the way in which these games offered quests to their players. The nature of the interface had implications though for how this could be done. Lacking the flexibility of a parser, the quests had to become simplistic enough that the limited set of interaction options possessed by a player could be harnessed. Lacking the nuance of textual representation, quests had to be promoted to be obvious for all to see. Lacking the narrative context of their single-player variants, quests had to be heavily ludic in their design. The result was the formulaic, often desultory provision of standard MMO quests (Doran & Parberry, 2010) – kill quests, craft quests, collect quests, escort quests and courier quests. Their limitation in design ambition meant that they could no longer rely on the intrinsic sense of accomplishment to entice players to find and conquer the challenge. Instead, they were made ‘first tier’ advancement options and sold to players on the basis of their efficiency of advancement. See for example the way that ‘questing’ within World of Warcraft is the most efficient way of gaining levels.
We can see much of the history of the text game puzzle in the design of both the modern graphical adventure and the MMO quest. However, it would be hard to argue against the proposition that much of the depth and nuance of puzzle solving has been lost in this evolutionary process.
Text games offer an interesting view of an atypical environment within which puzzles are emphasized. They are also, in many ways, one of the primary inspirations of the ways in which video game puzzles have been incorporated into modern games. Look back far enough through the evolution of any video game title and you’ll likely find a text game somewhere in its distant past. The lessons learned dealing with text game puzzle design greatly influenced the design of games that followed.
In this short blog series, I have attempted to discuss both this historical context and the reason why text games remain relevant in the modern gaming landscape. This in itself has been a puzzle given the constraints of a reasonable word count and the vast depth of the topic. Issues as complex as accessibility intermingle with topics as broad as video game history. Even within the relatively niche area of text gaming itself we find there are two largely distinct forms – interactive fiction and multiuser dungeons. These in turn have their own highly individualised way in which they have addressed the issue of puzzle design given the relative expectations of their respective audiences.
There are few gaming genres that have been as influential as the text game, although younger gaming audiences now may not even be aware that once there was such a thing, or that their development could survive and thrive as a commercial endeavour. It’s easy to forget that companies such as Infocom, Level 9 and Magnetic Scrolls once dominated software sales charts. However, in many ways the death of the commercial text game brought with it an opportunity for the genre to truly innovate in terms of game and puzzle design. Modern titles, developed on an indie basis by hobbyists with all the complexities that implies need no longer pander to the need to provide ‘value for money’. Few of these games require payment to play. While that can make sustained development difficult to justify, it does mean that puzzles can be considered in terms of their narrative or ludic value rather than as a mechanism for slowing progress through a story.
It is for this reason that many modern interactive fiction titles tends to shy away from puzzles generally, looking instead to offer an alternate view of progressing through a contextually shifting narrative – the role of a puzzle has been de-emphasised as it no longer meets the requirements of the form. However, with more ludic titles and environments the quest or puzzle is more important than ever – whether that is in graphical MMOs or in text-based MUDs. Understanding where we came from in this respect is important in understanding
Bibliography to follow!