10.1177/1046878105282279ARTICLESIMULATION & GAMING / March 2006Zagal et al. / COLLABORATIVE GAMES
Lessons learned from board games
José P. Zagal
Georgia Institute of Technology
Collaborative mechanisms are starting to become prominent in computer games, like massively multiplayer
online games (MMOGs); however, by their nature, these games are difficult to investigate. Game play is
often complex and the underlying mechanisms are frequently opaque. In contrast, board games are simple.
Their game play is fairly constrained and their core mechanisms are transparent enough to analyze. In this
article, the authors seek to understand collaborative games. Because of their simplicity, they focus on board
games. The authors present an analysis of collaborative games. In particular, they focus on Reiner Knizia’s
LORD OF THE RINGS, considered by many to be the quintessential collaborative board game. Our analysis
yields seven observations, four lessons, and three pitfalls, that game designers might consider useful for
designing collaborative games. They reflect on the particular opportunities that computers have for the
design of collaborative games as well as how some of the issues discussed apply to the case of computer
KEYWORDS: board games; collaboration; collaborative games; cooperation; computer games; deci-
sions; game design; individuals; lessons; multiplayer games; LORD OF THE RINGS;
payoffs; pitfalls; teams; utility
Although the vast majority of games played all over the world are collective in
nature, practically all electronic games are individual (Zagal, Nussbaum, & Rosas,
2000). Many reasons have been proposed for this dichotomy such as high costs of
technology (Zagal et al., 2000), the isolated location of computers in homes (Bunten,
1996), and the inherently single-user nature of the personal computer (Costikyan,
1998). The good news is that this is changing. Faster always-on Internet connections
together with cheaper technology have witnessed an increase in the amount of games
that can no longer be played alone. Multiplayer is now an important part of computer
However, the design space for computer collaborative games remains largely unex
plored (Manninen & Korva, 2005; Salen & Zimmerman, 2004; Zagal et al., 2000).
Recent years have shown an increase of cooperative game mechanisms in games that
do not always result in players collaborating to play the game. Li (2004) describes the
“honor system” implemented in AMERICA’S ARMY (2002) as a system designed to
SIMULATION & GAMING, Vol. 37 No. 1, March 2006 24-40
© 2006 Sage Publications
“entice people to play as teams and lead.” However, “everyone just kind of does their
own thing. There is not really any consistent teamwork or leadership being displayed.”
Another example is LEGEND OF ZELDA: FOUR SWORDS ADVENTURES
(2004). In a recent interview, Eiji Aonuma, producer of the game, stated that “although
it’s a game that four players have to cooperate to solve puzzles, when you play it ...,
you actually end up competing a lot more in that game than you do cooperating”
How can electronic games be designed so that collaboration is a worthwhile, inter
esting, and attractive option? In this article, we address this question. First, we focus on
the problem theoretically. Using game theory, we classify games into three different
categories: competitive, cooperative, and collaborative. We argue that collaborative
games are particularly suited for encouraging collaboration. Yet, few electronic games
are collaborative. For those that are, their complexity makes it difficult to extract
design principles. So, we focus on board games, which are easier to understand. Our
analysis yields seven observations, four lessons and three pitfalls, that game designers
might consider useful for designing collaborative games. We apply these observations
to computer games. By doing so, we demonstrate how board-game design can inform
Competitive, cooperative, and
In traditional game theory, games fall into two basic categories: competitive or
cooperative. Competitive games require players to form strategies that directly oppose
the other players in the game. The goals of the players are diametrically opposed.
Many traditional board games, such as Chess and Checkers, fall into this category
(Jones, 2000). In contrast, a cooperative game models a situation where two or more
individuals have interests that are “neither completely opposed nor completely coinci
dent” (Nash, 2002). Opportunities exist for players to be able to work together to
achieve a win-win condition. A cooperative game does not always guarantee that
cooperating players will benefit equally or even benefit at all. Cooperative games
include enforceable rules for negotiating or bargaining that allow players to identify a
desirable outcome for the parties involved. The classic cooperative game is the itera
tive version of the prisoner’s dilemma (Dawkins, 1989). In the traditional prisoner’s
dilemma, two prisoners are given deals if they defect and rat on their accomplice.
Because of the reward system, both rationally defect and both end up with a harsher
sentence than if they had collaborated with each other. In the iterative version, how
ever, collaboration becomes a rational strategy. In a cooperative game, nice guys (col
laborators) can finish first, as long as they make sure they are not being taken advan
tage of (Dawkins, 1989).
Though it was not acknowledged in game theory for a while, a third category exists.
In a collaborative game, all the participants work together as a team, sharing the pay
offs and outcomes; if the team wins or loses, everyone wins or loses. A teamis an orga
Zagal et al. / COLLABORATIVE GAMES 25
nization in which the kind of information each person has can differ, but the interests
and beliefs are the same (Marschak, 1972). Collaboration as a team differs from coop
eration among individuals in that cooperative players may have different goals and
payoffs where collaborative players have only one goal and share the rewards or penal
ties of their decisions. The challenge for players in a collaborative game is working
together to maximize the team’s utility.
Competitive and collaborative models are at opposite ends of a spectrum. Competi
tive games preclude collaboration. Collaborative games necessitate collaboration.
What about cooperative games? They lie between competitive and collaborative
games. Can collaboration be a worthwhile strategy in cooperative games? At first
glance, the answer would seem to be yes. However, because the underlying game
model is still designed to identify a sole winner, cooperative games can encourage
anti-collaborative practices in the participants such as free riding and backstabbing.
Free riding (e.g., when they are being evaluated and rewarded as a group, individual
group members do not pull as hard as they can) is a problem that causes group perfor
mance to suffer (Bornstein, Gneezy, & Nagel, 2002). Free Riding may appear in col
laborative games as well, but cooperative games exacerbate the problem as defectors
are often rewarded for their free riding behavior. Backstabbing, often unavoidable in
cooperative games, is the act of defecting when your partner cooperates. If it is done at
a particularly good moment, backstabbing can be an advantageous competitive
maneuver in an otherwise collaborative game. For example, the key to DIPLOMACY
(1959) is establishing the right alliances and knowing when to backstab your allies
(Costikyan, 1994). In effect, the best strategy in a cooperative game is knowing when
to behave competitively.
Behaving competitively in a collaborative scenario is exactly what should not hap-
pen in a collaborative game. One of the problems of designing a collaborative game
thus becomes one of dealing with the competitiveness that players bring to the table.
Computer games represent closed systems that are highly complex as well as
opaque to in-depth analysis. For example, most massively multiplayer online games
(MMOGs) are open-ended virtual environments with extraordinarily rich possibilities
for play. These games are played successfully using a wide variety of strategies.
EVERQUEST (1999) could be called collaborative because many people play it that
way. Yet, the same people could play it individually or competitively if they wanted to.
The multifaceted nature of a complex game like EVERQUEST is problematic for our
purposes of analyzing collaboration and game mechanics that foster it successfully. In
contrast, the nature of board games implies a transparency regarding the core mechan
ics of the game and the way they are interrelated. This transparency makes them more
accessible for in-depth analysis.
26 SIMULATION & GAMING / March 2006
Our approach also attempts, in part, to address the need for a language and a unified
vocabulary for describing the design of existing games and thinking through the
design of new games (Church, 1999; Costikyan, 1994; Kreimeier, 2002; Zagal,
Mateas, Fernandez-Vara, Hochhaleter, & Lichter, 2005). Our results are in many ways
similar to design rules, which offer imperative advice and guidelines for specific
design situations (Falstein, 2004). However, we are reluctant at this stage to categorize
our observations as rules.
We surveyed a number of collaborative games, team-based games, and competitive
games with interesting cooperative game mechanisms. We played these games and
watched others play them until we felt we had a good understanding of the game mech
anisms. A game mechanism is a physical artifact, rule, or type of interaction that
implements an action in the game. Trading between players is an example of a cooper
ative game mechanism. Capturing territory with a token is an example of a competitive
game mechanism. Many recent game designs feature balanced combinations of co
operative and competitive game mechanisms. For example, THE SETTLERS OF
CATAN (1997), has the object of determining which player has built the best contribu
tions to the island of Catan while allowing players to freely trade resources among
each other to further their progress. A similar mechanism is also present in MMOGS
such as EVERQUEST and WORLD OF WARCRAFT (2004) where players can trade
and auction goods to other players.
Toward a depth approach
To study issues of collaboration in game design, we investigated board games that
encouraged collaboration and collaborative thinking and the design features of those
games that allowed them to be both effective at encouraging collaboration and enjoy
able to play. We focus on conventional board games, rather than paper and pencil role-
playing games (RPGs).
RPGs, like DUNGEONS & DRAGONS (1974), are a successful genre of collabo
rative games. In them, a game master guides a group of players on an adventure. As the
team has the same goals for completing the adventure, most RPGs fall squarely within
the category of collaborative games. Yet, RPGs differ from other board games: There
is added emphasis on playing a role. In a conventional game, a player might be repre
sented by a piece or token. So, for example, a player might consider himself to be the
king in chess. However, the commitment to that role is low. There is no need or impetus
to behave like a king to play chess. In RPGs, the commitment to a role is much higher.
Players in RPGs frequently seek to create a satisfying storyline for their character,
rather than successfully complete the adventure (Fine, 1983). From a game-theory
perspective, these players are not playing a game; instead, they are creating a narrative.
Consequently, RPGs are often understood in terms of narrative theory, rather than
game theory (Heliö, 2004). Because we are interested in game mechanics, we exclude
RPGs from our analysis.
Zagal et al. / COLLABORATIVE GAMES 27
In our survey of board games, we came across a wide variety of games. Some, like
DER TIGER IS LOS . . . and IM MÄRCHENWALD (2000) were interesting and
enjoyable, but failed to encourage interesting collaborative practices; these games had
been designed for young children and, consequently, their game mechanisms were
simplistic. Others had interesting collaborative mechanisms, but failed to be enjoy
able. A few proved to be successful on both fronts.
In this article, we focus on one: Reiner Knizia’s LORD OF THE RINGS. This in-
depth approach allows us to give a better description and feel for how a collaborative
game works. Another reason is that LORD OF THE RINGS is an extraordinary game.
It is the most popular collaborative board game ever. It received the prestigious game
award Spiel des Jahres (Game of the Year) in 2001 for “Literature in Games.” Further
more, because of its Tolkien theme and the fact that it was designed by one of the great
est and most prolific game designers alive, it was one of the most anticipated games of
all time (Levy, 2001). Many critics feared, with good reason, that an interesting collab
orative game could not be made. As accomplished game designer Bruno Faidutti
(2004) noted in his review, “it [LORD OF THE RINGS] is the first collaborative game
that really works.”
LORD OF THE RINGS:
Detailing a collaborative game
People say, you can’t play with each other—you have to play against each other, other-
wise there’s nothing to do. Of course, that’s not true. I actually believe that playing with
each other and really facing a common opponent in the game makes a much richer play-
ing experience. My challenge was to create an atmosphere in the game that pushed people
together and made them naturally want to stay together....Theplayers realize after the
first few turns that they get hit so quickly with so many bad things that if they want to just
go off by themselves they have no hope. (Reiner Knizia, in Glenn, 2002)
In LORD OF THE RINGS, you are a hobbit. Together with your fellow hobbits (2
to 5 players total), you journey to carry the one ring to Mount Doom and destroy it.
Along the way, you will face many obstacles and get corrupted by Sauron and the
power of the ring. It is your decisions that will shape the fortunes of all.
Regardless of whether they succeed, the entire fellowship receives a score deter
mined by how much progress is made through the game; scores are recorded on a “Hall
of Fame” sheet that comes with the game. Because all the players receive the same
score, LORD OF THE RINGS is a collaborative game.
The game has two basic components, a main board that keeps track of the story and
displays the corruption track that marks the progress of the individual hobbits toward
Sauron (represented as an ominous black tower with a single, red lidless eye) and a
scenario board that tells a part of the Lord of the Rings story. In this case (see Figure 1),
the scenario is the journey through the mines of Moria. If Sauron reaches a hobbit on
the corruption track, that hobbit (and the player) is removed from the game. If that
28 SIMULATION & GAMING / March 2006
hobbit is also carrying the one ring, Sauron regains his ring and Middle Earth is
plunged into a second darkness; the game is lost.
Each player has a hand of cards that either perform an action or have a symbol
matching one along an activity line. Each turn, a player must first draw a tile out of a
bag. These tiles have either good outcomes (advancesa marker on an activity line earn-
ing the hobbits some progress or resources) or bad outcomes (requires the hobbits to
sacrifice resources, move closer to Sauron, or triggers one of the scenario events con-
nected to that particular board). Tiles and events can cause the hobbits to either
advance toward Sauron or, worse, they cause Sauron to advance toward the hobbits.
After drawing tiles, the current player can play up to two cards matching the symbols
next to an activity line. Progressing on an activity line enables the hobbits to move
closer to their final objective and/or gain resources that will help them later. Alterna
tively, they may draw cards to replenish resources or move themselves one space away
from Sauron to avoid danger. When the hobbits reach the end of the main activity line
(which also tells them their current score) or if the last event is triggered, they advance
to the next scenario.
Why does LORD OF THE RINGS work?
LORD OF THE RINGS is an effective collaborative game because players are
tempted to behave competitively but winning the game requires them to behave col
laboratively (Knizia, 2004). Winning the game is possible with good luck and careful
resource management but is more likely through good collaboration: Specifically,
Zagal et al. / COLLABORATIVE GAMES 29
FIGURE 1: LORD OF THE RINGS Game: Moria Scenario Board
active communication amongst the players and timely sacrifices for the good of the
group. We have identified four lessons that can be observed:
Lesson 1: To highlight problems of competitiveness, a collaborative game should introduce a
tension between perceived individual utility and team utility.
In LORD OF THE RINGS, there are many opportunities for selfish behavior at the
expense of the team. Usually, selfishness is expressed in a player making a decision
that has high utility for their hobbit but is not the best decision for the team. For exam
ple, a hobbit might choose to draw two cards to replenish resources or “heal” himself,
moving further away from Sauron. A better decision might be to play the last card in
the hand and take the consequences of an evil die roll. Competent play requires that
players choose strategies that balance or forgo self-preservation to help the team.
For example, Frodo draws an “active hobbit moves 2 or Sauron moves 1” tile that
requires that either he is corrupted two steps or Sauron moves one step closer, corrupt
ing all hobbits one step. Frodo gets to make the decision. If he chooses the former, he
will achieve the least total corruption for the team (2 steps) but the most corruption for
himself (2 steps). If he chooses the latter, he will achieve the most total corruption for
the team (5 steps, one for each hobbit) but the least corruption for himself (1 step). A
selfish hobbit will choose the latter, a collaborative hobbit the former. Notably, Sauron
achieves his best result (5 steps) when a hobbit is corrupted to behave in a selfish,
The design of the game accentuates the problem that the participant mistakes a col-
laborative situation for a competitive one. Because success in a collaborative situation
requires a concentration on team utility over perceived individual utility, the individu-
alistic approach is problematic. In LORD OF THE RINGS, players who behave
individualistically are likely to run into difficulties, no matter how well they play
according to their perceived individual utility (Knizia, 2004). When players discover
this, they learn that a collaborative situation requires a fundamentally different ap
proach than a competitive one.
This lesson is similar to the definition of a social dilemma. Social dilemmas are sit
uations in which individual rationality leads to collective irrationality (Kollock, 1998).
In other words, they are situations in which individual rational behavior leads to a situ
ation where everyone is worse off. What is interesting about LORD OF THE RINGS
in the context of this lesson is how the tension between the individual rationality and
the effect on the group is highlighted.
Lesson 2: To further highlight problems of competitiveness, individual players should be
allowed to make decisions and take actions without the consent of the team.
In LORD OF THE RINGS, a selfish hobbit has the ability to act selfishly without
the consent of the fellowship. In the previous tile example, Frodo makes the decision.
If the game were designed so that the team, not the individual, makes the decision, the
other four hobbits could vote for the “Frodo moves 2” (collaborative) option. But,
30 SIMULATION & GAMING / March 2006
LORD OF THE RINGS is not ruled by votes or consensus. If Frodo wants to choose
the “Sauron moves 1” (competitive) option, he can. Because there are no ways for the
team to stop or override an individual, competitive choices can be made.
Also, an important design feature is that although it is the individual who makes the
decision, it is the team who needs to work together for that decision to be properly
informed. Because the other members of the fellowship can also evaluate the decision
(from their perspectives), they have an incentive to persuade that hobbit to make the
right decision for the entire fellowship. This encourages the kind of co-construction of
meaning essential to good collaboration (Barron, 2003). That communication then
turns into a way for the fellowship to convince the individual hobbits to behave in a
Lesson 3: Players must be able to trace payoffs back to their decisions.
Players need to be able to reflect on the consequences of their actions. In particular,
they need to experience expectation failure. They expect their decision to be a good
one but later discover it to be problematic.
In LORD OF THE RINGS, the action one player takes often directly affects the
next player. For example, each hobbit needs to collect a certain amount of “life tokens”
for every scenario board. Often, a hobbit will take an action based on their individual
perspective that inadvertently makes it hard for the next hobbit to collect any life
tokens. Without consulting with that next hobbit, that hobbit could not have realized
this. However, their mistake quickly becomes obvious. The next hobbit does not have a
good action to take. In the game, the next player is likely to complain that the last action
has just put him into trouble. So, the first player discovers that his individual and/or
competitive decision making is problematic.
Axelrod (1984) notes that making interaction more durable or frequent, increasing
information about an individual’s actions, and increasing identifiability are all ways of
facilitating strategic solutions to social dilemmas. In the context of games, these obser
vations help players understand the consequences of their actions as well as those of
their fellow players. This applies not only to decisions that affect themselves, but also
others. In fact, Komorita, Sweeney, and Kravitz (1980) note that cooperation rates
increase significantly as the benefits to others from one’s cooperation increase.
Lesson 4: To encourage team members to make selfless decisions, a collaborative game
should bestow different abilities or responsibilities upon the players.
First, the hobbits have individual abilities. Frodo is particularly good at using his
resources, Sam is less vulnerable to corruption, Merry does not require as many life
tokens, and Pippin is particularly good at playing many cards. For good play, it is
important to recognize these strengths and use them accordingly. For instance, a useful
strategy is for Sam to take the move that requires the active hobbit to roll the evil dice,
as it has less effect on him. Eventually, somebody would have to takethe dice roll and it
makes sense for Sam to be the one to sacrifice. Unless Sam is particularly stubborn or
Zagal et al. / COLLABORATIVE GAMES 31
competitive, it is relatively easy to convince him to make the sacrifice. Though it is bad
for his own utility, it is relatively easy to see that it is in his ultimate interest to sacrifice
for the team.
Second, the ring of power can move from one hobbit to another hobbit after each
scenario board. If the ring-bearing hobbit is captured by Sauron, the game is lost. As
such, that hobbit is (often temporarily) more important than his comrades. So, occa
sionally (particularly late in the game), the correct decision regarding an “active hobbit
moves 2 or Sauron moves 1” tile might be to have Sauron advance toward the hobbits,
instead of the ring-bearing hobbit moving two steps closer to Sauron.
The events of one of the game sessions observed illustrate how selfless decisions
result in an engaging, enjoyable and ultimately successful game experience. During
this game, the players were on the second to last board, Shelob’s Lair, and it was cer
tain that the ring would pass to Pippin for the last board, Mordor. Unfortunately, Pippin
was only one step away from Sauron and would have a hard time staying uncorrupted
in Mordor. The fellowship discussed various options of how to help Pippin survive
Mordor. Could we afford to hang around in Shelob’s Lair and allow Pippin to heal him
self on his turn? Should we purchase the Gandalf card that allows one player to heal 2
steps, depleting our resources? After 5 minutes of intense discussion over various
approaches, the solution was arrived at: Pippin made the ultimate sacrifice and volun-
tarily got corrupted by Sauron before Shelob’s Lair concluded and, thus, the ring could
not pass to him. Instead, it passed to Frodo, who was least corrupted. Frodo and Sam
stormed through Mordor and plunked the ring into Mount Doom. The game was won;
every player rejoiced. Pippin sacrificed his life to save Middle Earth. If such a moment
of self-sacrifice is interesting as a story, it is even more engaging when you are the one
to make the decision (in the game).
Challenges in designing collaborative games
In this section, we shift our focus from mechanisms that work to challenges that
designers need to overcome. We extract three pitfalls that illustrate some of the particu
lar difficulties involved in the design of a collaborative game. These pitfalls are high
lighted from the collaborative and team games surveyed.
Pitfall 1: To avoid the game degenerating into one player making the decisions for the team,
collaborative games have to provide a sufficient rationale for collaboration.
It is easy for a collaborative game to degenerate into a solitaire game. A solitaire
game is one that can be abstracted to one player performing all the actions, or giving
orders, to achieve the win condition set out by the game. This was a major problem in
many of the collaborative and team games surveyed. One person, the most competent
player, could direct the entire team. For example, SCOTLAND YARD (1982) can
become boring as one person can tell everyone what to do (Aleknevicus, 2002).
Avoiding this solitaire feel can be accomplished by several design techniques.
32 SIMULATION & GAMING / March 2006
One technique is to give the players different roles and abilities so that optimal
game-play depends on good coordination and decision making on the part of the play
ers. LORD OF THE RINGS gives different abilities to each of the hobbits so that each
hobbit has a useful role to play at various points in the game. Furthermore, resources
are hidden so that each player only sees their own. From an optimal information-
awareness perspective, this design choice might be considered a problem. But, from a
game-play perspective, it is a good choice as it forces communication. A better (or
dominant) player is unable simply to command the other players, as it is difficult to
make good, informed decisions without the help of the others.
Another technique is to make the problem sufficiently difficult so the players need
to work together to solve it. In collaborative games, players work together by sharing
knowledge and resources, exploring the information space as completely as possible
to identify the best strategy to use. If there are insufficient information or resource-
management requirements, the collaboration becomes forced and it usually falls on
one player to make the majority of the decisions. In LORD OF THE RINGS, there is
enough variability of play and in the resources held by the players to require individual
management. Communication among the players about the available resources for a
particular task becomes more efficient than a single player trying to marshal all the
resources at one time.
Another collaborative game that manages this well is EAGLE EYE AGENCY
(1999). In that game, the players are detectives gathering clues to solve a caper. They
have limited resources, so there are only so many places they can visit to pick up clues.
The detectives have to work together to gather the right clues and then analyze them to
solve the mystery. As there are many clues and several red herrings, it is useful to dis-
cuss theories with others. The collaboration feels natural and useful.
Pitfall 2: For a game to be engaging, players need to care about the outcome and that out
come should have a satisfying result.
This pitfall applies to all games; however, we feel it is particularly important for
collaborative games. If players do not care about the outcome, then they are not moti
vated enough to help each other or improve on their performance. If players find the
outcome to be unsatisfying (either boring or random), they are unlikely to learn any
thing, understand the consequences of their actions, or want to play it again. Games
require a good narrative and flow to be entertaining to the players. A good game can be
like a good story. A good collaborative game can be even more entertaining because it
involves the collective contributions of all the players. It also helps if the outcomes
have some variability to them to promote surprise and tension. Yet, the outcome
should still be largely accountable to the decisions made by the players.
Watch a Lord of the Rings player as he flips over a tile during the last part of the game.
Some inhale sharply or wince or close their eyes . . . [This game seems] to involve people
in a way more than the quiet mental gyrations of Chess or Go. (Branham, 2001)
Zagal et al. / COLLABORATIVE GAMES 33
The story of the hobbits as they journey through Middle Earth and face a myriad of
dangers combined with the limited randomness provided by the cards, tile, and die cre
ates a very engaging experience for the players. Because the outcome is often uncer
tain until the very end, the game manages to maintain (and often build) interest and
tension. A well-played adventure that fails at the edge of Mount Doom is often as
exciting as one that succeeds.
Pitfall 3: For a collaborative game to be enjoyable multiple times, the experience needs to be
different each time and the presented challenge needs to evolve.
Although this pitfall also applies to all games, it can be particularly prevalent in col
laborative games because these games face unique problems in replayability. People
learn skills through practice. To put in more practice time, they need to be able to
repeatedly play the game. However, if a collaborative activity has an easily learned
deterministic solution, then the participants will find it pointless to repeatedly play the
game. The repeatability of a game can be enhanced by random elements in setup and
randomization of the resources and obstacles through the course of the game. Too
much randomization and the players will have no reliable information to formulate
and discuss strategies. LORD OF THE RINGS is a game of controlled chance, as the
main mechanism is drawing from a bag filled with evil and good pieces (Branham,
2001). Unlike completely deterministic games, like Chess, LORD OF THE RINGS
cannot be played exactly the same way twice. It maintains good replayability by hav-
ing constrained randomization of the tiles and cards, which leads to different decisions
and situations each time the game is played.
Competitive games are most engaging when opponents are closely matched. Two
novice Chess players will have competitive games. Likewise, two expert players will
have competitive games. However, an expert playing a novice is not a competitive situ-
ation. If we expressed playing ability numerically, a challenging situation in a compet
itive game occurs when the difference in the players’ abilities is close to zero. In con
trast, a challenging situation in a collaborative game is better modeled by the sum of
the players’abilities being close to the difficulty of the game. If two intermediate play
ers find a collaborative game challenging, a novice and/or expert pair might find it
challenging, particularly when the expert’s additional skill balances the novice’s lack
ing skill. That same game will be too challenging for two novices and too easy for two
A collaborative board game only has a set of static goals and rules to provide obsta
cles and counterstrategies. As a result, after multiple playing sessions, the players can
become more familiar and better at the game until it is below their combined abilities.
The game becomes unchallenging as the team is able to easily beat the game. So,
unlike competitive games, like Chess, collaborative games need to adapt to the play
ers’ abilities to maintain replayability. In LORD OF THE RINGS, the players can
adjust how many steps Sauron is away from the fellowship at the beginning of the
game. This provides an easy, relatively flexible manner to increase the difficulty of the
34 SIMULATION & GAMING / March 2006
game. Once a fellowship has mastered beating the game at 15 corruption steps to
Sauron, they can try it at 12 steps.
Implications for computer games
Because of their transparency and simplicity, we have focused on board games. In
this section, we aim to apply those lessons learned to computer games. Computers
open the game design space as fairly sophisticated computation can be done quickly
and accurately (Lundgren, 2002). They can display information in different (and often
more meaningful) ways. They can even be used to analyze how players are doing and
provide just-in-time help or make dynamic adjustments to the difficulty of the game.
In addition, computers offer communication flexibility. Whereas board games are
limited to open face-to-face communication, computer games can be more flexible
because, amongst other things, players no longer need to be co-located. Communica
tion is of particular importance to a collaborative game, as players have to coordinate
their actions and strategy. Changing the medium for communication can vastly change
how participants work together (Clark & Brennan, 1991). Restricting some types of
communication while supporting others can be quite powerful in changing the nature
of collaboration in games (Dillenbourg & Traum, 1999; Scott, Mandryk, & Inkpen,
2002). A computer program can support conflict resolution and group decision mak-
ing in ways that may be superior to face-to-face discussions (Nuñez, Aguero, &
Olivares, 1998). Thus, the computer also significantly opens the design space for
multiplayer games (Manninen, 2002), in particular those that are collaborative.
However, unlike the case of the board games studied, the use of the computer can
also have negative effects on communication between players. Many basic cues of
identity, personality, and social roles are absent in the online world (Donath, 1998),
making it harder for players to understand each other and agree on plans of action. In
addition, there is an increased risk of deceptive practices by players. This problem is
particularly evident in modern MMOGs where it is commonly referred to as “grief
play” (Lin & Sun, 2005). Typical uncollaborative practices include “taking over”
badly injured monsters from other players, stealing treasure from recently killed mon
sters from the players who did the kill, and using bots. This last point is particularly
problematic because for many players it raises the issue of whether or not they are even
playing with other human beings at all! To compensate for the factthat players may not
be physically co-located, they are many times forced to resort to out-of-game sources
such as bulletin boards to inform themselves of the reputations (positive or negative) of
their fellow players (Lin & Sun, 2005). It is thus critical that computer games take
careful consideration of how they will communicate game information to the player as
well as how players will communicate with each other. After all, computer games have
to account for much of the feedback and communication that happens naturally in a
The communication issues discussed are especially relevant in the context of
Lessons 1 and 3 (introduce tension between individual utility and team utility, play
Zagal et al. / COLLABORATIVE GAMES 35
ers must be able to trace payoffs back to their decisions). For instance, in
WOLFENSTEIN: ENEMY TERRITORY (2003), aside from the final result (Allies
win or Axis win), there are few affordances for players to have a proper sense of the
team utility of certain actions. If a player wants to play the game as a gung-ho loner sol
dier, he is not only perfectly capable but, if skillful enough, can win the game single-
handedly. There are not any explicit bonuses for “sticking together.” Usually a player
will try to “get an extra kill” unaware of the fact that his actions resulted in the death
of two of his teammates. Not only is he ignorant of the implications of his actions, but
his fellow teammates do not have a means to easily communicate these to him. The
problem, as Sellers (2001) notes when referring to the four-player arcade classic
GAUNTLET (1985) is “If only that darned narrator would have said something to
scold players like ‘Wizard isn’t pulling his weight’.“
Computer games also make Lesson 3 more important. The amount of time it takes
to play a board game can be orders of magnitude less than what a collaborative com
puter game could allow. In computer games, players must have some way to under
stand the impact of their decisions over a time span of 10s, 100s, or even 1,000s of
hours (MMOGs being a good example of the latter). Computer games can explore
these issues particularly well if we consider the number-crunching involved in the log-
ging and evaluation of a player’s performance. Computer games could provide contin-
uous in-depth individualized feedback in a timely fashion.
There are games that have taken other approaches to solving this problem.
BATTLEFIELD 1942 (2002), a modern-day militaristic first person shooter, allows a
few players to assume the role of commanders and squad-leaders. Commanders can
coordinate battles by issuing orders to squad leaders, place waypoints for soldiers to
follow on the map, and also scan for enemy activity. Players in leadership roles have
options available to help them monitor the success of their fellow players and, through
the use of voice-over-net communications, talk directly to the players in their units.
This game allows players to trace the payoffs of their decisions by having other players
play roles whose responsibility includes monitoring and providing feedback. As
Gillen (2005) comments, “My favourite squad leader utilised the voice-over-net sys
tems regularly, and his caring, encouraging tones (‘Get down, Medic! I need you alive,
man’) forged one of my favourite memories of BF2 so far.”
We speculate that these games could also benefit from assigning bonuses to actions
such as providing good covering fire, saving a teammate from enemy fire, or even
showing good leadership. Perhaps other players could also report teammate actions
that resulted in negative team utility. Another innovation could be to provide for
postgame analysis where players could explore how their actions affected the outcome
of the game.
Lesson 2, individual players should be allowed to make decisions and take actions
without the consent of the team, is not usually a problem in computer games. This is
probably due to the real-time nature of many computer games as well as the prevalent
use of individualized input devices. In other words, each player has his own controller
and the other players have no means of impeding its use. Thus, we could say that com
36 SIMULATION & GAMING / March 2006
puter games tend to be excellent examples of individual decision making in collabora
tive games. However, we have observed games in which a related problem appears:
Due to the extreme freedom enjoyed, players are forced to “collaborate” under con
In some ways, a corollary to Pitfall 1, provide sufficient rationale for collaboration,
there are many computer games in which players are required to collaborate in sit
uations that are arbitrary and contrived. For example, in THE ADVENTURES OF
COOKIE & CREAM (2000), two players each control a character that must race as
fast as possible to a goal. Their progress is constantly hindered by closed barriers that
must be opened by having the other player step on a switch or button. This cooperative
mechanism is unfortunately rather common in computer games, with some games
going so far as to require four players to simultaneously step on separate buttons in
order to proceed. These situations also tend to ignore Lesson 1 because, due to the fact
that none of the players can proceed without all of them performing an action, there is
no tension between individual and team utility. Additionally, these sorts of situations
tend to fall prey of Pitfall 2, players need to care about the outcome and that outcome
should have a satisfying result.
Lesson 4, bestow different abilities or responsibilities upon the players, is also
prevalent in many computer games with cooperative mechanisms. For example, on-
line games such as BATTLEFIELD 1942 and WOLFENSTEIN: ENEMY
TERRITORY are online first-person shooter games in which the players choose dif-
ferent classes (or kits). These vary with each game, but generally allow the players
access to different abilities such as healing teammates and planting explosives. A suc-
cessful team in these games is usually one in which all the classes are represented, with
their respective players taking the most advantage of the particular abilities they offer.
Pitfall 3, the experience of a collaborative game needs to be different each time and
the presented challenge needs to evolve, should be easier to avoid in the case of com-
puter games. However, despite the computational capabilities of computers, few com
mercial developers have implemented dynamic difficulty adjustment systems to ad
dress this (Hunicke & Chapman, 2004). Although these systems have mostly been
used in single-player games, there have been some approaches for dealing with the
additional complexities of multiplayer games such as large MMOGs (Carpenter,
To conclude, we maintain that games have a unique potential to engage people in
collaborative activities. On the other hand, collaborative games are rare and extraordi
narily difficult to design. This article has hopefully illustrated some of the particular
difficulties inherent to the design of these games as well as showing that simply having
cooperative elements is generally insufficient for collaborative play. We have noted
how many computer games do apply some of the lessons we have identified, though
most tend to fail when it comes applying them all. We believe that computer games not
only have the potential for addressing many of the issues discussed but also many
affordances to solve them. We are hopeful to have provided insight that game design
ers might be able to use to create more and better collaborative games.
Zagal et al. / COLLABORATIVE GAMES 37
Aleknevicus, G. (2002). Player interaction in games. Counter Magazine, (16). Retrieved from http://
AMERICA’S ARMY: OPERATIONS. (2002). U.S. Army.
Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books.
Barron, B. J. S. (2003). When smart groups fail. Journal of the Learning Sciences, 12 (3), 307-359.
BATTLEFIELD 1942. (2002). Electronic Arts.
Bornstein, G., Gneezy, U., & Nagel, R. (2002). The effect of intergroup competition on group coordination:
An experimental study. Games and Economic Behavior, 41 (1), 1-25.
Branham, F. (2001, September). What is the goal? The Games Journal. Retrieved from http://www
Bunten, D. (1996). On-line multi-player games. Retrieved 4 August 2005 from http://www.anticlockwise
Carpenter, A. (2003). Applying risk analysis to play-balance RPGs. Gamasutra.com. Retrieved 10 August
2005 from http://www.gamasutra.com/features/20030611/carpenter_pfv.htm
Church, D. (1999, August). Formal abstract design tools. Game Developer. Retrieved from http://www
Clark, H. H., & Brennan, S. E. (1991). Grounding in communication. In L. B. Resnick, J. M. Levine, & S. D.
Teasly (Eds.), Perspectives on socially shared cognition (pp. 127-149). Washington, DC: American Psy
Costikyan, G. (1994). I have no words & I must design. Interactive Fantasy, 2. Retrieved from http://
Costikyan, G. (1998). Why online games suck (and how to design ones that don’t). Retrieved 4 August 2005
Dawkins, R. (1989). The selfish gene. New York: Oxford University Press.
Dillenbourg, P. J., & Traum, D. (1999). Does a shared screen make a shared solution? In C. M. Hoadley &
J. Roschelle (Eds.), Proceedings of the computer support for collaborative learning (CSCL) 1999 con-
ference (pp. 127-135). Palo Alto, CA: Stanford University.
DIPLOMACY. (1959). Calhammer, A. Avalon Hill.
Donath, J. (1998). Identity and deception in the virtual community. In P. Kollock & M. Smith (Eds.), Com-
munities in cyberspace (pp. 29-59). London: Routledge.
DUNGEONS & DRAGONS. (1974). Gygax, G. & Arneson, D. TSR, Inc.
EAGLE EYE AGENCY. (1982). Deacove, J. Family Pastimes.
EVERQUEST. (1999). S. Clover, B. McQuaid, & B. Trost. 989 Studios.
Faidutti, B. (2004) Le Seigneur des Anneaux. Retrieved 20 July 2004 from http://faidutti.free.fr/jeux/
Falstein, N. (2004). The 400 project. Retrieved 29 October 2004 from http://www.theinspiracy.com/
Fine, G. A. (1983). Shared fantasy. Chicago: The University of Chicago Press.
GAUNTLET. (1985). E. Logg. Atari.
Gillen, K. (2005). Battlefield 2 review. Retrieved 29 June 2005 from http://eurogamer.net/article.php?article
Glenn, S. (2002). Interview with Reiner Knizia. Retrieved 30 June 2005 from http://web.archive.org/web/
Heliö, S. (2004). Role-playing: A narrative experience and a mindset. In M. Montola & J. Stenros (Eds.),
Beyond role and play. (pp. 65-74). Helsinki, Finland: Solmukohta.
Hunicke, R. & Chapman, V. (2004).AIfordynamicdifficultyadjustmentingames.Challenges in game arti
ficial intelligence (AAAI Workshop). Pittsburgh: AAAI Press.
IM MÄRCHENWALD. (2000). Nikish, M. Adlung Spiele.
Jones, K. (2000, August). Non-predatory games. The Games Journal. Retrieved from http://www
38 SIMULATION & GAMING / March 2006
Knizia, R. (2004). The design and testing of the board game Lord of the Rings. In K. Salen & E. Zimmerman
(Eds.), Rules of play: Game design fundamentals (pp. 22-27). Cambridge, MA: The MIT Press.
Kollock, P. (1998).Socialdilemmas:Theanatomyof cooperation.Annual Review of Sociology, 24, 183-214.
Komorita, S. S., Sweeney, J., & Kravitz, D. A. (1980). Cooperative choice in the N-person dilemma situa
tion. Journal of Personality and Social Psychology, 38, 504-516.
Kreimeier, B. (2002). The case for game design patterns. Retrieved 29 October 2004 from http://www
Levy, L. (2001, September). Reinier Knizia: The special k’s of German game design. The Games Journal.
Retrieved from http://www.thegamesjournal.com/articles/SpecialK.shtml
Li, Z. (2004). The potential of America’s Army the video game as civilian-military public sphere. Unpub
lished MA Thesis, Massachusetts Institute of Technology.
Lin, H., & Sun, C. T. (2005). The “white-eyed” player culture: Grief play and construction of deviance in
MMORPGs. In S. Castell & J. Jennifer (Eds.), Selected papers of the Digital Interactive Games
Research Association’s second international conference (DiGRA 205) (pp. 91-100). Vancouver, Can
ada: Digital Interactive Games Research Association.
Lundgren, S. (2002). Joining bits and pieces: How to make entirely new board games using embedded com
puter technology. Unpublished Master’s Thesis, IT University of Goteburg, Goteburg, Sweden.
Manninen, T. (2002, June). Towards communicative, collaborative and constructive multi-player games.
Paper presented at the Computer Games and Digital Cultures Conference, Tampere, Finland.
Manninen, T., & Korva, T. (2005).Designing puzzles for collaborative gaming experience—CASE: eScape.
In S. Castell & J. Jennifer (Eds.), Selected papers of the Digital Interactive Games Research Associa
tion’s second international conference (DiGRA 205) (pp. 233-247). Vancouver, Canada: Digital Interac-
tive Games Research Association.
Marschak, J., & Radner, R. (1972). Economic theory of teams. New Haven, CT: Yale University Press.
Moledina, J. (2004). Doing mushrooms, Miyamoto-style (Special Edition). Retrieved 30 June 2005 from
Nash, J. (2002). Two-person cooperative games. In H. W. Kuhn & S. Nasar (Eds.), The essential John Nash
(pp. 99-114). Princeton, NJ: Princeton University Press.
Nuñez, G., Aguero, U., & Olivares, C. (1998, September). Group decision-making for collaborative educa-
tional games. Paper presented at the 4th International Workshop on Groupware (CRIWG 98), Buzios
Salen, K., & Zimmerman, E. (2004). Rules of play: Game design fundamentals. Cambridge, MA: The MIT
SCOTLAND YARD. (1982). W. Schlegel, D. Garrels, F. Ifland, M. Burggraf, W. Scheerer, & W. Hoermann.
Scott, S. D., Mandryk,R. L., & Inkpen, K. M. (2002, January). Understanding children’s interactions in syn
chronous shared environments. Paper presented at the Computer Support for Collaborative Learning
(CSCL) 2002, Boulder, Colorado.
Sellers, J. (2001). Arcade fever: The fan’s guide to the golden age of videogames. London:Running Press.
THE ADVENTURES OF COOKIE & CREAM. (2000). Agetec.
THE LEGEND OF ZELDA: FOUR SWORDS ADVENTURES. (2004). Nintendo.
THE SETTLERS OF CATAN. (1997). Teuber, K. Mayfair Games.
WOLFENSTEIN: ENEMY TERRITORY. (2003). Wedgwood, P. Activision Publishing Inc.
WORLD OF WARCRAFT. (2004). Blizzard Entertainment.
Zagal, J., Nussbaum, M., & Rosas, R. (2000). A model to support the design of multiplayer games. Presence:
Teleoperators and Virtual Environments, 9 (5), 448-462.
Zagal, J., Mateas, M., Fernandez-Vara, C., Hochhaleter, B., & Lichti, N. (2005). Towards and ontological
language for game analysis. In S. Castell & J. Jennifer (Eds.), Selected papers of the Digital Interactive
Games Research Association’s second international conference (DiGRA 205) (pp. 3-14). Vancouver,
Canada: Digital Interactive Games Research Association.
Zagal et al. / COLLABORATIVE GAMES 39
José P. Zagal has an MSc in engineering sciences and a BS in industrial engineering from Pontificia
Universidad Catolica de Chile. His doctoral research deals with computer-supported collaborative learn
ing. He is a member of the Electronic Learning Communities Lab and the Experimental Game Lab at Geor
gia Institute of Technology. In his free time he loves to design and play games.
Jochen Rick has an MS in electrical engineering from Georgia Institute of Technology. His doctoral
research, in Georgia Tech’s College of Computing, focuses on the role of personal home pages in academia.
His broader research interests center on how new media can further learning.
Idris Hsi has recently escaped from Georgia Tech clutching a doctoral degree from the College of Comput
ing. Currently he is working for the Microsoft Office Design Group as a usability engineer. He has used com
puter games, board games, and game theory books to prolong his undergraduate studies by at least 2 years
and his tenure as a graduate student for at least a decade. In his spare time, he is investigating the role of nar
rative in game design, especially how it structures game mechanisms and objectives toward particular
ADDRESSES: JPZ: College of Computing, Georgia Institute of Technology, Atlanta, GA 30332-0280,
USA; telephone: +1 404-894-1558 (w); e-mail: firstname.lastname@example.org; URL: www
.cc.gatech.edu/~jp. JR: College of Computing, Georgia Institute of Technology, Atlanta,
GA 30332-0280, USA; telephone: +1 404-385-1105 (w); e-mail: jochen.rick@
cc.gatech.edu; URL: home.cc.gatech .edu/je77. IH: Office Design Group, Microsoft Cor
poration, Redmond, WA 98052, USA; e-mail: email@example.com.
40 SIMULATION & GAMING / March 2006