Content uploaded by Maria Hwang
Author content
All content in this area was uploaded by Maria Hwang on Feb 20, 2018
Content may be subject to copyright.
Proceedings of DiGRA 2014: <Verb that ends in ‘ing’> the <noun> of Game <plural noun>.
© 2014 Authors & Digital Games Research Association DiGRA. Personal and educational classroom use of
this paper is allowed, commercial use requires specific permission from the author.
Communalizing the Interfaces
Of Single Player Games
Mark Santolucito
Yale University
New Haven, CT
mark.santolucito@yale.edu
Maria Hwang
Teachers College, Columbia University
525 W. 120th street,
New York, NY
mlh2169@tc.columbia.edu
ABSTRACT
New interfaces for games have reinvented the experience of play, but every interface
works on the basic assumption of responsiveness. If a user sends a command via the
interface, they expect to see a change in the game state. The recent multiplayer game
Twitch Plays Pokémon is revolutionary because it rejects this assumption of
responsiveness. By taking user commands in the aggregate, a user does not build a direct
mental correlation between input and state change. We call this type of interface a
communally controlled interface (CCI). A CCI can be used as a game design element to
act as an educational tool, to help build narrative, or to inspire players to engage with a
game in unforeseen ways.
Keywords
interface, game design, input, responsiveness, multiplayer, gameplay, emergent game
mechanic, meta-narrative, communally controlled interface (CCI)
INTRODUCTION
We are interested in the emergent gameplay and communal construction of meta-
narrative resulting from the communally controlled interface (CCI). By comparing the
emergent gameplay mechanics in an MMO adaption and the MMO-CCI adaption, we can
explicitly attribute gameplay phenomena to a CCI. In particular we compare:
Pokémon: Red Version - an RPG
Twitch Plays Pokémon - an MMORPG with a CCI (hereafter MMO-CCI)
PokeMMO - an MMORPG (hereafter MMO)
Twitch Plays Pokémon (Anon. 2014) streams video of an emulation of Pokémon: Red
Version on a gaming community site, www.twitch.tv. Pokémon: Red Version is a single-
player, turn-based, third-person, role playing game in which the player battles her own
Pokémon against non-player combatants (NPCs) and explores an environment as a means
-- 2 --
to complete a number of goals. A player must “catch” her Pokémon and “train” them by
engaging in battles. A primary in-game character explicitly states that the primary goal is
to catch all 150 Pokémon. However, the game is traditionally played towards the goal of
beating the “Elite Four”, group of NPCs that are equivalent to the final boss in many
adventure games.
Twitch Plays Pokémon is controlled via the chat feature built into the twitch.tv website.
To navigate the world, players type commands (“up”, “down”, “left”, “a”, etc.) in the chat
window, which are then translated into an input command. Because of a technical
limitation of the platform, commands are delayed by approximately 30 seconds (Anon.
2014). A further limitation of the platform prevents the execution of all commands
received. If the game is in a state of executing an action, all commands will be ignored.
Thus, a command sent at time T will only be processed if the game is in an idle state at
time T + 30 seconds. With at times more than 100,000 concurrent users, individual users
were not able to verify the success or failure of their own input.
The PokeMMO, as the name suggests, is a massively multiplayer online version of the
original Pokémon: Red. In PokeMMO multiple players are playing essentially their own
game with an independent respective agency. The shared platform of course allows the
players to interact and collaborate with each other during the gameplay, but it is also
possible that a player could go through every stage of the game without interacting with
any other player in particular. In the global chat feature of PokeMMO, players mostly ask
direct questions that aim to solve their own issues at any given time that does not
particularly benefit others than themselves. Therefore, even though PokeMMO is
designed for sharing a communal space online, fundamentally the shared space does not
go beyond the mere definition of co-inhabitance and every player continues with their
own gameplay with the one-to-one mapping of input to game action intact.
The communally controlled interface (CCI) used in Twitch Plays Pokémon fundamentally
differs from previous interfaces in the lack of a one-to-one mapping of input to game
action. Traditional singularly controlled interfaces are popular because of the importance
of narrative or personal achievement in many games (Fuller and Jenkins 1994; Jenkins
2004). The CCI’s rejection of responsiveness to the individual in Twitch Plays Pokémon
is acceptable to the user base in part because of the modern familiarity with the concept
of creating a narrative through the collective. As an example, take viral memes, which are
created by the anonymous group behavior of online communities. No one person can be
attributed to the success of a particular meme, even the creator of the content relies upon
the community to make the meme “go viral.” Similarly, in Twitch Plays Pokémon, no
singular person is responsible for success, yet all participants share in a collective feeling
of achievement.
We now explore the implications of using a CCI in a game. We look at two major
affordances that the CCI adds to the Twitch Plays Pokémon that are absent in PokeMMO:
emergent game mechanics and meta-narrative. We observe these phenomena by
examining quantitative and qualitative data from chat logs and external forum sites. With
an understanding of how the CCI functions in Twitch Plays Pokémon, we lay a
framework for how a CCI might be applied in other games. This abstract conception of a
CCI is a novel and practical theoretical framework that may be applied to future research
and development.
-- 3 --
DATA ANALYSIS
Emergent Game Mechanics
The predominant emergent game mechanic we observed was based around the concept of
planning. The limitation of interface responsiveness triggered players of Twitch Plays
Pokémon to coordinate future plans to overcome difficult stages. To observe planning as
a game mechanic, we graph the frequency of chats containing the word “plan” (see
Figure 1).
We choose to discuss three peaks in the graph representing significant events. The
Pokémon Tower and Viridian Gym are two places that were particularly hard to navigate
in Twitch Plays Pokémon. To successfully complete these sections of the game, players
need to organize and work together around a single plan. “Bloody Sunday” is the name of
a particularly unfortunate sequence of choices made in the game. In the aftermath of the
event the community needed a plan for recovery and a plan to avoid making a similar
mistake in the future.
Figure 1: Frequency of chats containing the word “plan”
sampled over four hour periods of Twitch Plays
Pokémon game chat logs.
In the MMO, we see a similar, but distinct group planning behavior. PokeMMO has
implemented “teams”, in which small groups of player may band together both in and out
of game. Since in PokeMMO gameplay is not team based, coordination between members
is limited to more of a chatroom style interaction. However, it is not uncommon in other
MMOs for small teams to coordinate at a high level. “The players of [Dark Age of
Camelot, EverQuest, and Anarchy Online] reported that their guilds staged pre-planned
-- 4 --
activities, often called raids, 1-2 times per week” (Saey et al. 2003). Although we observe
planning behaviors in both MMO-CCIs and MMOs, each is a specific style of planning
that provides a different intellectual challenge to the players.
To understand these differences, we qualitatively analyze the planning strategies
employed by exploring the out of game communities. The PokeMMO just like any other
MMO game germinated activities and communities outside the game itself. We
investigate our preliminary inquiry of CCI in comparison to MMOs on the notion of
“community of practice” (Lave & Wenger, 1991; Wenger, 1998) as research supports the
proliferation of “out-of-game” communities based on the communally shared experiences
developed through MMOs. The ties and bonds that are formed through these
communities encourage learning, offer rapport, and extend a safe place for identity
formation and development. “At the aggregate level of the community, ...learning process
takes the form of an emergent reorganization of the patterns of member participation
coupled with a growth of shared knowledge through changing practices and artifacts”
(Steinkuehler, 2004).
Certainly, with the PokeMMO there are many communities that show evidence of the
previously mentioned characteristics of community of practice. However, these
PokeMMO communities are limited in the sense that the “collaboration” and “planning”
could not be brought back into the game, unlike the guilds’ pre-planned activities in the
aforementioned MMOs. Because PokeMMO is not a team based game, the discussions
involving planning in these communities is reduced to an individual level and must be
interpreted as such.
An example of an out of game community for the CCI is represented in Figure 2. In order
to navigate particularly difficult areas, users create strategies on external sites. For
example, a particular path might be successfully navigated in two moves by either “up-
left” or “up-right”, where failure can be caused by a “left” on the first move. Some users
will recognize that the optimal path is “up-right” as it reduces risk of failure (specific
ordering of moves is impossible in a CCI) and attempt to organize the community to
follow the plan.
Because of the necessity of coordination and planning between players to create
probabilistically successful strategies, a CCI can be employed to introduce planning as a
game mechanic. We will explore the implications of this observation in our discussion of
the theoretical framework.
-- 5 --
Figure 2: A sample document made by user
PhreaksChinstrap from Reddit. This is an example of a
planning strategy to avoid a worst case scenario.
Meta-Narrative
A further consequence of using a CCI in Twitch Plays Pokémon can be seen in the
community behavior of players. The community has created an entire meta-narrative on
top of the narrative constructed by the original game designers. For example, the
community treated a particular game item as a holy relic, where no such connotations
exist within the original game design. This narrative construction is driven by the lack of
agency individual players have over gameplay, allowing for a suspension of disbelief that
the game in fact is autonomously generating a narrative. This meta-narrative
demonstrates the especially rich potential for CCI games to engage players in ways
unforeseen by designers.
In Figure 3, BirdJesus and Lord Helix are major plot components to the meta-narrative,
whereas Misty and the Moonstone are secondary plot elements. Major plot elements,
once introduced, continue to consistently appear in game chats, at times having moments
of particular prominence. Secondary plot elements, may persist in low levels through the
game as with the Moonstone, or may have isolated moments of high importance as with
Misty.
This phenomenon of meta-narrative is rarely observed within traditional MMO games.
By retaining their own agency in an MMO, players also have the responsibility of putting
that agency to use. In order for a meta-narrative to be established, all players must not
only agree on the meta-narrative, but take an active role in executing the plot.
-- 6 --
Figure 3: Frequency of chats mentioning some meta-
narrative elements sampled over four hour periods of
Twitch Plays Pokémon game chat logs.
A more common meta-narrative in MMOs is a feud between guilds or teams. This meta-
narrative can persist only as long as members of the guild make an effort to continue the
feud. In an MMO-CCI, the lack of agency of individuals also absolves them from the
responsibility of executing the narrative. The narrative instead adapts to the agency and
choices of the collective.
THEORETICAL FRAMEWORK
On a more theoretical level, we look at the requirements for a CCI to function properly.
To do this we first examine Pokémon: Red Version and the elements of its game design
that made it well-suited for a CCI. Abstracting these elements allows us to build a more
generalizable framework for a CCI.
A critical feature of Pokémon: Red Version is the experience and leveling system. In a
trainer battle, the player must defeat a number of pokemon in succession. Every time the
player defeats a single opponent pokemon, the player’s pokemon will gain experience
and be more likely to defeat that same opponent pokemon in the future. Even if the player
loses the overall battle, defeating a single pokemon will increase the likelihood of future
success. This allowed Twitch Plays Pokémon to make progress, however incremental,
even after repetitive losses. This concept will call recoverability.
Recoverability is best understood if the game is conceptualized as a decision tree. In the
first attempt at the beating a stage, the decision tree may contain few pathways to success
-- 7 --
states. Upon a failure, the decision tree must be updated to contain relatively more
pathways to success. Essentially, we are dynamically adjusting the game difficulty. In
this way, the systematic randomness of the CCI can be overcome and game can be
guaranteed to progress.
Other important mechanics include turn-based combat and limited input options. Turn-
based combat removes time as a confounding factor to gameplay. Especially with the
thirty second delay in the implementation, the idea of a time dependent game seems
unachievable. In Pokémon: Red Version, you may also only press one of any seven
buttons at a time. This makes it easier for players, to build plans, predict, and execute
appropriate moves.
However, the benefits of both turn-based combat and limited input are eclipsed by the
importance of recoverability. With a large number of input options, in a game such as
Civilizations V (Firaxis Game 2010), we simply have a tree with many branches where
many of those lead to failure. By maintaining recoverability, this tree can be updated to
encourage progress. In fact, a continuous controller game is simply an extreme case of a
game with many input options. So we see, turn-based combat and limited input are
simply tools to make recoverability easier to implement.
FUTURE WORK
Throughout the duration of Twitch Plays Pokémon certain implementation choices were
reevaluated and re-implemented. Of particular interest is the dichotomy within Twitch
Plays Pokémon that existed between the two game modes, anarchy and democracy. In
later iterations of Twitch Plays Pokémon and with fewer players, as is the current state of
Twitch Plays Pokémon, every command can be processed.
The type of planning required for these different modes is complex and requires
reasoning about uncertainty and distributed control. It seems to align closely with control-
flow analysis and scheduling theory for parallel computation (Rayward-Smith et al.
1995). To investigate this further requires a more thorough review of scheduling theory
than is appropriate to provide here. However, this could lead to development of
educational games for high level computer science topics. These games could also be
used to teach valuable soft skills such as teamwork and communication.
In an existing commercial game, a CCI mode could be introduced to add replay value.
However, care must be taken to ensure that the game has the recoverability property. If it
does not, it must be implemented in some way. A limiting factor in this is the apparent
need for a large user base. Future work may include exploring methods to effectively
implement a CCI with a relatively small user base.
It would also be possible and interesting to develop a game entirely around a CCI. While
it is clear a CCI can spur emergent game mechanics and communally created meta-
narratives, more research must be done to understand the best ways to take advantage of
these phenomena when the CCI is a foundational part of the game design.
BIBLIOGRAPHY
Anonymous Developer. (2014). Twitch Plays Pokémon. [Web], Twitch, US: played 20
Feb., 2014.
Firaxis Games. (2010). Civilization V. [Microsoft Windows, OS X], Firaxis Games,
North America.
-- 8 --
Fuller, M. and Jenkins, H. “Nintendo and New World Narrative.” Jones, Steve;
Cybersociety: Computer-Mediated Communication and Community (1994): 57-72.
Game Freak. (1998). Pokémon: Red Version. [Game Boy, Red], Nintendo, North
America: played 20 Feb., 2014.
Jenkins, H. “Game Design as Narrative Architecture.” Computer vol. 44, (2004): s3.
Rayward-Smith, V. J., Burton, F. W., and Janacek, G. J. 1995. “Scheduling Parallel
Program Assuming Preallocation.” In Scheduling Theory and Its Applications, P.
Chretienne, E. G. Coffman Jr., J. K. Lenstra, and Z. Liu, Eds., Wiley, 146-165.
Lave, J., and Wenger, E. Situated Learning: Legitimate Peripheral Participation.
London: Cambridge University Press, 1991.
PhreaksChinstrap, “The NASCAR Strat: How we’re going to beat the maze.”
r/twitchplayespokemon. Last modified February 23, 2014. http://redd.it/1y83pr
“PokeMMO: General Discussion.” Forums.pokemmo.eu. Last modified May 15, 2014.
https://forums.pokemmo.eu/index.php?/forum/7-general-discussion/
Seay, A. F., William, J. J., Kevin, S. L., and Kraut, R. E. “Project Massive 1.0: Organizational
Commitment, Sociability and Extraversion in Massively Multiplayer Online Games.” In
Proceedings of DIGRA Conference. 2003.
Steinkuehler, C. “Massively Multiplayer Online Games as an Educational Technology:
An Outline for Research.” Educational Technology vol. 48, no. 1 (2008): 10-21.
Steinkuehler, C. “Learning in Massively Multiplayer Online Games.” In Proceedings of
the 6th international conference on Learning sciences, 521-528. International
Society of the Learning Sciences, 2004.
Wenger, E. “Communities of Practice. Learning as a Social System.” Systems Thinker,
(1998). Last modified February 26, 2000, http://www.co-i-l.com/coil/knowledge-
garden/cop/lss.shtml
