The Ninth Annual Game Design Think Tank
Project Horseshoe 2014
horseshoe Group Report: Making Friends While Killing People
Participants: A.K.A. "The Pay It Forward Phenomenon"
Isaac Barry, EveryMove Patricia Pizer, Way Too Far Games
Laralyn McWilliams, The Workshop Entertainment Vijay Thakkar, Jolly Entertainment
Dan Hurd, Playful Corp Jonathan Hamel, Amazon Game Studios
Yuri Bialoskursky, EA Tiburon Joel Gonzales, Wargaming
Laralyn McWilliams, The Workshop Entertainment Terry Redfield, Wicked Fun Co
  Facilitator:Kathy Astromoff
  download the PDF

Brief statement of the problems on which the group worked

Even as online multiplayer games create opportunities for social connection, the interpersonal dynamics, resulting from game system structures and behaviors, can be incompatible with the development of healthy friendships. Problems manifest in areas such as lack of empathy, deficiencies in communicative expressiveness, and dehumanizing other players, and result in significant challenges to making friends from strangers in online games.

A brief statement of the group’s solutions to those problems

This workgroup focused on general recommendations for designs that might foster player friendships. We have identified several elements with which to address the problems inherent in creating friends from strangers in games:

  • A simple model of elements necessary for environments conducive to new friendships
  • Systems that enable players to see each other with greater emotional connection and as potential friends
  • Providing memory aids to aid the memory of good social interactions, and the contexts of past encounters
  • Methods for players to interact in ways that open doors and encourage further interaction

Expanded problem statement

How might game designs support players befriending each other? What’s in our toolbox that can be leveraged to enable strangers to become friends? Online games have been evolving to incorporate player connectedness into their core experience, with more and new interaction layers between players. Considering the growth of mobile device markets and ubiquity of internet connectivity, we see that multiplayer game design is awash with new opportunities. As game designers, we might be driven by various motivations—aesthetics, politics, audience, etc.—but most would seem to agree that great social glue can be measured by increased rates of player retention.

There’s no need for a precise, focused definition of “friendship.” Even limited to just one game, we may consider two players sufficiently “real” friends if they feel connected, socially and personally, to each other. So long as more new friendships emerge, we have success.

Even with this broad description, the problem isn’t simple. Multiplayer games frequently cast players as NPCs to be overcome or utilized, rather than as human beings to befriend; players make excellent content. But, more and more, new games are putting players into cooperative roles with social mechanics. How can we ensure that they do not continuing to fail to connect people in significant, valuable, and persistent ways?

As the capability for players to express themselves has increased, so too has abuse, trolling and other toxic behaviors in communities. The combination of anonymity and lack of persistent social contracts in game worlds allows some of the most negative human behaviors to present themselves; trolling, toxicity, etc.. Hostile environments muzzle prosocial dialog but, ironically, attempts to prevent abuse can create dynamics that squelch, not only negative social behaviors, but meaningful communication and healthy social connections as well.

Take, for example, menu-chat approaches that limit communication to “friendly”, non-threatening utterances. The communication environment may tends toward the emotionally positive and upbeat, but one-dimensional expressiveness limits the kind and depth of meanings shared, and offers no effective opportunity for friendships to spark.

The next generation of multiplayer game design needs better approaches than our current standards. Solutions should allow, and strengthen, positive social experiences—support, coaching, and healthy playful rivalries—but also prevent toxic and unhealthy behavior. A marriage of game mechanics and connection can create and cultivate player to player relationships that grow from casual in game acquaintances, to close and long term friendships, clearly, but the design of games’ social environments, is complex, even treacherous. How can we start with a grounded approach to improvement?

We found that players can fail to connect with each other in many ways and for many reasons. We noted some typical behavior:

  • A lack of empathy for other players or perceived lack of empathy from other players
  • Opting out of methods of communication and expression; ignoring other players
  • Dehumanizing others players, misunderstanding motives, seeing them as unreasonable or malicious
  • Exhibiting toxic behavior such as trolling and griefing
  • Inability to recognize and remember other players from earlier encounters

The next generation of multiplayer game design needs better approaches than current standards. Solutions should allow and strengthen positive social experiences—support, coaching, and healthy playful rivalries—but also prevent toxic and unhealthy behavior. A marriage of game mechanics and connection can create and cultivate player to player relationships that grow from strangers to casual acquaintances, and to close, long-term friendships, clearly. But the design of games’ social environments, is complex, even treacherous. How can we start with a grounded approach to improvement?

Over the course of our work, we recognized a natural gravitation toward one expression of social mechanics or another in accordance with the particulars and standards of each game, mechanics, genre, platform etc.. We returned often to a short list of popular, well-understood games as proxies for these many types. Other games were discussed, but these were used most often to direct speculations, concepts, and thought experiments.

  • Hearthstone—1v1 session-based play, limited interaction, limited communication
  • League of Legends—strong player-player dependencies and robust interaction in session-based play
  • World of Warcraft—high-interaction, long-form play, acting in narrative environments

Expanded solution description

We began by creating a model of friendship development; the things needed for friendships to form and grow. Ideas were taken from psychological studies and literature, past examples and experiences, etc.. The result: a base set of factors and dynamics relevant to creating new player-player social connections. First, as a common language, then as a common abstraction for guiding designs and evaluations of new features, improvements, and thought experiments.

Each element in our model describes factors key to forming new friendships in-game:

  • Meaningful communication—ability to sharing thoughts and feelings
  • Humanize—foster empathy, understanding, and patience for others
  • Proximity—shared spaces, with opportunities for social recognition and interaction without planning
  • Shared experience—interpersonal and group identity through common experience
  • Group reliance—“Is the group more efficient than the individual?”; opportunities to help one another
  • Status expressions—ability to demonstrate and recognize individual status, and exchange accordingly

Below, each of these factors will be discussed in more detail, along with example approaches, using this model as a guide.

Meaningful Communication

Communication is essential to forming and maintaining relationships. While the modes may vary, the more meaningful a dialogue can be, the stronger the possible connections will be produced as a result.

In multiplayer games, the first job of communication is the coordination of game action. The greater a game’s need has for coordination, the more informative the communication supporting the game must be. Hands down, the most popular mode of communication, historically and presently, is text chat. It’s simple, universal, and practically boundless, though thoroughly abusable. More recently, text chat is often supplemented, or replaced, by voice chat. Either mode can bear fruits we would rather avoid, such as misunderstanding, isolation, toxicity, bullying, etc. Attempts to overcome the risks and limitations must struggle to balance limiting the opportunity for abuse, with necessary levels of expressiveness. The potential for communication can be described as though on a simple continuum. At one extreme, realtime video chat. At the other end of our spectrum, only the barest, abstract form of stimulus-response communication, beeps (as in Journey) or Yo! Somewhere between the extremes are sophisticated systems, such as Toontown Online’s speed chat, that provide some expressiveness, while curating utterances to those that are safe and positive.

Yet all systems that employ limited communication systems, place a burden on player expression. For example, Hearthstone, limits player communications to selecting from among emotes, “Thanks,” “Well played,” “Greetings,” “Sorry,” “Oops,” and “Threaten.” As reactions to game situations, these might be sufficient. But even within these contexts, as one would expect, players have evolved conventions, such as greeting players to a new game with “Threaten,” or repeatedly using “Greetings” to hurry another player that has been taking too long on their turn. Even with such a limited system, Hearthstone still includes a “squelch” feature, so that obnoxious users can be silenced. So, even while struggling to share deeper meaning, players can still be disturbed by abuse of this limited system. Is it worth being unable to building friendships with others?


Players can’t make friends from dehumanized strangers. Before a person can be seen as a peer, or judged worthy of our time, they must be seen as socially responsive and relatable. While safe, the interactions of most people tend toward friendliness. But the nature of game systems often make this a tricky factor, in practice.

For example, competitive games may cast other players as obstacles to one’s goals, even when not actual foes, such as with unskilled teammates in League of Legends. With intense focus on the split-second reactions to gameplay situations and cues, players may not have any attention left for recognizing or empathizing with the players behind the other avatars. Stripped of their humanity, people can see each other in rather black and white terms; a player who has defeated me five times is a monster, and the teammate that failed to help is an idiot. These effects are exacerbated by short time limits, or bursts of intense gameplay where single mistakes can all but seal a team’s fate.

Attempting to interject additional social mechanics into tight, frenzied play environments might actually worsen the gameplay experience. They may introduce communication opportunities at precisely those moments when players are feeling the least supportive. Or the may simply be ineffective, such as as if player goals involve filtering out non-essential information, additional social contexts are likely to be filtered too.

Proximity & Repeat Encounters

The classic idea of a “third place,” as cornerstone of community, upholds the notion that proximity and repeated encounters are key indicators and predictors of an environment’s ability to support creating and deepening friendships. Contrary to curmudgeonly opinions like Apuleius’s “familiarity breeds contempt,” familiarity actually tends to breed fondness. The more exposure we have to a stimulus—be it object, person, or what have you—the more we will tend to like it. Keeping people near enough to one another that they can be recognized and communicate, over extended periods of time, encourages affection while maximizing opportunities for meaningful interaction.

Certainly, external spaces and communities are popular instruments for socializing. From game-focused forums, to Twitter, or Twitch streaming and Youtube ‘let’s play’ videos, compelling gathering places can be built outside of the game experience. But these needs can also be addressed from within games, through practices like level design.

For example, World of Warcraft, begins new players in low-level ‘newbie zones’, secluded from all larger threats, with the intent being the teaching of the game’s fundamentals. By the time they’ve travelled to their first major city, players can be assumed to have the basic skills needed for success. Once passed, those areas no longer play a role in the game, and are practically never revisited by that character.

Contrast this implementation with the ‘newbie zone’ of EverQuest, which surrounds most major cities with the low-level zone. In this case, players from all experience levels must travel through this zone to reach the city, and new players are routinely exposed to this traffic. Additionally, adjacent zones are typically much more difficult, constraining en masse new-player movement, and gathering the traffic of players, in similar situations, for longer periods of time.

Seeding desirable higher-level goals in otherwise lower-level zones might also help to increase lower to higher level interactions. For lower-level players, these tougher goals would provide a sense of aspiration, and for the higher-level players, there would be an opportunity for education.

Shared experience

Shared game experiences offer people material with which to identify and relate to one another. The currency of communitas, they are key to social group identity and are the protoplasm of cultural life. Catalysts for deeper social interaction and relation, and evidenced by ritual practices around the world, shared experiences are intertwined with all   understandings of each other, and the understandings of one’s self too.

Sociopsychological expressions of this concept can been seen in the real world through activities like initiation rituals, training boot camps, or other indoctrination events. Afterward, shared experiences become artifacts of identification, common to all members of the group.

Games, as entertainment experiences, are already shared among players, but how might systems be added or improved to help build group identity and recognition with shared experiences?

An example: encounter difficulty and cadence as catalyst for communication and deeper social interaction. An example would be a required but atypically difficult or unique quest that every member of a certain class would have to complete to advance. Having a common touchstone of experience would provide a fun avenue for guidance, commiseration or empathy toward other members of that same class.

Psychological expressions of this concept can been seen in the real world through activities like hazing, boot camp, or other indoctrination events. Afterward, that shared experience can be a way to identify with other members of the same group, but the application of the event could be traumatic or otherwise unhealthy if used without consideration.

Could splitting groups into smaller subgroups further increase the bonds felt between subgroup members? Limiting group size would certainly focus the number of others one might identify with an experience. An example for Hearthstone: class-based epic match that unlocked a legendary card for the current character. Afterward, the players who have succeeded against that challenge might feel incentivized to share their experiences, tips, or perspectives with players who have yet to encounter that challenge.

Group reliance

People, as social animals, readily understand and identify groups; they’re a familiar and useful structure to games as they band together cooperative player effort. As in the real world, our roles in groups plays a part in determining how we behave and view ourselves, and how we imagine ourselves to be viewed. People with a strong sense of care and belonging from groups, also feel worthy of the same, and are more likely to act prosocially within that group. In a game, by default, a player’s value to any other player comes from their potential to lend a hand and reach objectives. It’s within this context that our social skills are usually operating. The nature of the dependencies between players will directly influence players’ utility assessments. Some games, by virtue of their gameplay, make life rough for noobs.

Groups are a fundamental aspect of team games like League of Legends, but practical matchmaking constraints can lead to game-determined partnerships with non-ideal members, such as low skill and low social aptitude. This can be a frustrating experience for competitive players.

In multiplayer games featuring unit differentiation (such as with class-based RPG character systems), there often exists tension between designers and players on the issue of solo-play viability. Many players wish to avoid reliance on other players as a prerequisite for success, and most players prefer the reduced logistics and downtime that comes with needing little coordination with others. But the costs of allowing players to be more self-sufficient include a reduction of grouping and social interaction.

In newer games such as Archeage, a player can learn abilities from across the entire spectrum of play styles, leading to a proliferation of very well-rounded and self-sufficient players. Grouping in systems like this can be infrequent and might feel necessary only from a critical-mass perspective (where a certain amount of damage might be needed in a timeframe) rather than from a class-dependent system.

Unlike MMORPGs, in 1v1 games like Hearthstone, there is no grouping in the basic game structure. Nearly all socializing happens outside of the game session, and is represented through guilds or forum membership.

A classic example of a design that elicits group collaboration, EverQuest character classes are very role-focused and less broad. Individual players have little ability to play in a solo state for long, without suffering significant difficulty and time tradeoffs. In groups, a balance of classes is desirable for flexibility, and is more or less required for attempting the more difficult fights and dungeons.

A potential way to introduce group reliance in games like Hearthstone might be an additional game mode that requires two players in tandem to face off against the computer or another team. Rather than being a natural grouping (seen in the Everquest system), this would be a requirement to join these games.

Example: Memory Aids

Our attitudes toward each other are the products of our histories—memories of past interactions inform expectations of future experiences. While games typically keep good records of game performances—levels, scores, wins, losses, etc—little attention is usually given to histories of social behavior. The burden of recalling past interactions is left, solely, for the players to manage. But, with the anonymized nature of player identity, retaining any significant memory of each other is difficult. Since players can only recognize or reward past kindnesses when those earlier acts can be remembered, tools to aid and strengthen memories—at least of positive impressions—should help improve friendship rates.

Examples of passive friendship systems include monitoring chat for positive phrases, while active systems could allow players to create notes or tags to apply to each other. A Yelp-like system of simple tagging could label other players with human qualities—“funny,” “helpful,” “excited,” etc.. Then, seeing an avatar tagged with ‘funny’ or ‘helpful’ might help subtly reinforce that player’s humanity. An ideal interaction would include factoring those tags into communications with that player, rather than generically treating that player as an impediment to your in-game goals.

Example Systems / Techniques

  • A new player, given help in a time of need, can repay with thanks many times over in the future
  • Player performs an interesting act—skillful, unexpected, etc.—that can be jointly remembered
  • Key, or useful items as memento artifacts, personally marked
  • Player-player recommendations ala LinkedIn
  • Karma system—analyze chat, raising the karma score for positive comments; subtly reward players with “luck”
  • Rivals systems—marking out sporting rivalries

Systems aiding recognition and recall of past player interactions will, naturally, magnify the long-term effects of individual interactions. So care should be taken to bias memories towards positive behaviors, tending toward forgiveness or simply forgetting bad behaviors much more quickly.

As described, this system fits most naturally into high-intensity quick-session games like League of Legends or Hearthstone. The tags could be viewed on a player’s bio or win/loss record (since that information is typically sought out), and might lightly provide additional context for that player.

In MMOs, such as World of Warcraft, the long-form play sessions and additional down-time potentially allow for expanded information. This could look like a comment field or more robust tagging system visible through player bios or inspect screens.

Example: Gifting

Giving items or other valuable resources can set the stage for positive feedback loops between players. Items must possess some value and intent, both to the giver and the receiver; a limited resource, not arbitrarily given. This can be very beneficial to one party, as in a ‘good samaritan’ situation where little is expected in return, or it can be mutually beneficial, used to express thanks or admiration of another player, who then might reciprocate the gifting process.

A concept we discussed that might apply to many genres of social games is that of a ‘jackpot bonus’. In this case, a player would have a slowly-growing pool of a valuable resource that they could not directly access, but could gift it out to other players. When gifted out, every player in the group, chat room, or other social container would receive the bonus, and the gifting player would reset their pool to zero. In this case, the value from the gifting player comes from the fact that the pool must refill, and that their appreciation of the group was ‘worth’ the temporary setback.

From the receiving players’ perspective, the generosity of the giving player is a very strong catalyst for increased communication, thanks, or other social interactions, and might also be a model for altruistic behavior for that player in the future.

Example: Cross-Session Chat

Games with rapid-in:rapid-out session structures have, traditionally, featured persistent lobbies—Counter-Strike, Unreal Tournament, etc. Many new games, with similar session structures, have streamlined and automated the matchmaking process to improve the speed and quality of the game experience, but these approaches tend to lack persistent lobby spaces. Without extended or recurring downtime, spent in a shared space, where proximity and repeat encounters can enable opportunities for friendship, players are able to befriend strangers only through the use of external spaces, such as forums or social media platforms. Requiring this level of effort on the part of players, ensures that few lasting friendships will arise directly from being engaged in the game experience, and those limited to the most hardcore members of the audience.

Can we enhance these optimized matchmaking schemes such that, within the in-game context, players can loiter and dialogue with each other?

For example: chatroom containers between games could allow queued and idling players to space for communication, expressing ideas and feelings. Signifiers, such as themes, could help kick-start these flash conversations. With titles like ‘Boston Players’, ‘Star Trek Fans’,’Actually It’s About…’ the matchmaking and lobby systems could provide the streamlined game functions, while the third-place-in-miniature chat spaces offer proximity within the context and experience of the game itself. We can find some nascent examples of these ideas in session “MMOs” like War Thunder, which provide general chat areas.

Example: Trader Proximity

Resource allocation and the need for trade can drive player interactions. Even without third place spaces, like marketplace zones or staging areas, repeated interactions can be patterned around exchanges.

For example, a low-level resource refined for use in higher-level goods, may be readily found as part of early play, but be inaccessible to more experienced characters. Perhaps the resource is time-consuming to collect, and found in areas populated by low-level players. Higher-level players create demand for the resource, yet without sufficient incentive to dedicate themselves to gathering. Supplied by lower-level players, the resulting trade practice and interface could lead to further interactions. Perhaps partnerships based on resource and crafting systems can create a space for the development of friendships.

Items from the brainstorming lists that the group thought were worth reporting:

  • Player supplied in-game services, such as a player-run hair salon, where time spent in service is used to socialize
    • Mechanics can keep bringing people together—hair styles wear off, new styles become available
  • Big Fish Games online slots, where jackpot bonuses can be shared, prompting recipients to thank the winner
  • Tagging crafted items with the creator’s name
    • Asheron’s Call allowed manual crafted item tagging
  • Explicit friendship systems, with varying levels, permissions, and tools
  • Encourage players to share expertise

Other reference material:

Dan Cook’s GDC presentation on Governance https://www.dropbox.com/s/3i0bw2zdjfqegq6/Government%20-%20GDC%202014.pptx?dl=0

Report: What Moves a Player to Meaning in Their Play Experience, Project Horseshoe 2013