The current approach to dealing with player toxicity is like a quarantine: take the affected population and keep them from infecting the rest of the population. But what if the metaphor of a contagious disease is what was keeping us from seeing the true nature of the problem? What if it’s closer to an instinct, and in the case of online team-coop games like Overwatch, design elements are baked in the games themselves that prime players’ psyches to view each other with hostility?

I’m going to talk about online team-coop games using Overwatch as the primary example, but I believe the principles apply to many of the other notably toxic games out there like League of Legends, DotA 2, Counter Strike: Global Offensive, Call of Duty, etc. In many ways, Overwatch is typical for the space, but is worth specific examination because Blizzard has previously signaled an intent to be broadly conscientious, and because they’re willing to revise significant portions of their game’s designs when they feel like it is warranted. To me, this means Overwatch, as it exists today is the way that it is in spite of presumed efforts make the culture “good”, and is able change if need be.

So what is a toxic player? To paraphrase Lyte (Jeffrey Lin, League of Legends (LoL) Lead Social Systems Designer who worked on LoL from 2011 to 2016, and was responsible for halving toxic behavior during that period), a toxic player is one whose negativity can impact and infect other players. The current strategy developed by Lin to combat player toxicity is to permit players to address the symptoms to minimize incidences of negative actions, thereby enabling player feedback to separate the toxic population from the majority of non-toxic players. This approach assumes that player toxicity is a behavioral extension of trolling. As some have noted, online trolling behavior may the symptom of individuals enforcing perceived digital social norms, as much as it is about wanton hate spewing. In that sense, LoL’s approach and resultant success makes a lot of sense.  This approach leverages the player population as much as it relies on algorithms to identify and separate the problem population and discriminately punish them according to commensurate degrees of toxicity, rate of incidence, or other metrics. It also leaves less room for trolls to justify their actions based on individually presumed norms by codifying positive expected behaviors. Also by creating affordances for players to police their own population, the non-toxic majority can express and manifest said codified norm in the form of feedback to the toxic minority. Overall, various means of encouraging positive and discouraging negative behavior seem to be having a desirable effect in reducing the volume of abuse so its efficacy isn’t in question, but, to examine how design exacerbates the problem, it is helpful to treat it as behavior and not as demographic issue.

What the current strategy may obfuscate is the fact that the origin of this type of behavior may be a consequence of the design itself. With respect to the design of a game’s rules, Salen and Zimmerman talk about three kinds of rules: constituative, operational, and implicit. As players internalize the consituative and operational rules of the game, they will individually titrate out their own understanding of the implicit rules of the game; the resultant combination of a player community’s interpretations going on to form the base of that game’s local cultural norms. From another angle, Morton Deutsch, an American social psychologist and researcher of conflict resolution explains in his “Crude Law of Social Relations” that, “the characteristic processes and effects elicited by a given type of social relationship also tend to elicit that type of social relationship.” What Deutsch found was that the types of conditions, whether cooperative or competitive, were born of circumstances that instigated those types of behaviors in the first place like a self-reinforcing feedback loop. So if a game’s culture becomes toxic, it might have started with something communicated by the design, intentionally or unintentionally, through its rules.

A possible source of the toxicity in Overwatch may be a conflict in the interpretations of the game, with some players focusing on the obvious and advertised team combat-based matches, while others maybe playing another game that the former may not be acknowledging, and the developers don’t want to speak to explicitly. Looking at the design of the matches, players developing a team-centric and pro-social implicit rule framework logically follows. Afterall, what is a team but a collection of interdependent individuals working to achieve individual and common goals. Games involving teams will often feature challenges that test the team as a unit along its operational characteristics, such as role definition, communication, coordination; and its social characteristics, such as trust, motivation, and stress tolerance. Typically team games’ challenges are external to the group so as to allow everyone on the team to work together to overcome the problem together. Overwatch does more than most to reinforce this paradigm by framing objectives as team goals, facilitating team formation by categorizing heroes into classes, and having notifications for when a team composition is missing a necessary role. But a game isn’t constrained to just the parts that focus on obvious conflicts and challenges, and the design of systems external to the action do as much to frame the players’ experiences as much as what happens in the moments of action. When you expand the definition and scope of what the game is beyond the matches, the makings of intra-team conflict become evident as the implicit team-centric framework comes into conflict with the other games and forms of play cohabitating within the title.

Transactional Approach

Transactional Approach

Situational Approach

Situational Approach

Brian Upton’s situational approach to game design talks about how play can be thought of as the entire mentally simulated experience within the context of a game, where the game is the space that encloses the rules, and interactive space, and the play of a game. Extending this concept, in Overwatch, the rules of the game can be further broken down into different interrelated systems of rules. There’s the action portion where the teams shoot at each other, but there’s also the part of the game before the doors open and the shooting starts when you can briefly socialize with your team and admire their cosmetic accomplishments, the part before the map even loads when players pick their characters in coordination with their teammates, and the even larger non-social meta space in which players open loot boxes, visit forums for strategies and try to optimize their efforts to get the cosmetics they want as quickly and efficiently as possible. In Overwatch, the team-centric social expectations are undermined by a supporting design that emphasizes a mentality that is counterproductive in encouraging good team dynamics, and advantages players willing to look out for themselves at the expense of others.

At the heart of the intra-team conflict seems to be a combination of common design features that many other games in this genre share, namely randomized loot box rewards delivery, procedural match-making, and individualized play-driven success metrics. Firstly, players’ investment into and interpretation of the reward system can greatly change their understanding of the reward system’s implementation. Players play 5 vs 5 team matches, in which experience points (XP) are awarded for the time in the match, medals earned, completion of the match, and victory, with bonuses for consecutive matches played, voluntarily hopping in to fill a game in progress, winning your first match of the day, and for playing with friends. According to “BigAIChubbs”, after taking into account the average time in a match, the XP curve, and the number of bonuses in a game, a player is looking at between 884 and 1,064 hours of game time to earn all of the ~1200 items in the game. As Paul Tassi estimates, when pegging to real life currency, this places a value of somewhere between $1000.00 to $3000.00 -- a number that gets compounded when taking into account the rate-dependent nature of earning upgrades through play, and the time-sensitive nature of the desirable seasonal and special event rewards. As Jim Sterling points out, these rewards matter a lot to players on a psychological, meta, and social level. Given the importance of these cosmetic rewards, it might be more appropriate to frame the situational design for some players in Overwatch as one where they earn personal rewards through the social and mechanical play, than the presumed one that it is a game with mechanical and social play that merely features rewards.

Some players are as invested if not more invested in the rewards than the combat part of the game, and so the combat part of the game as well as what counts as reasonable conduct towards others in said portion of the game take on a very different light. Players that are so inclined will logically look at each match as a time investment whose rewards can be maximized by winning any given match, but also whose cost can be minimized by limiting the amount of time committed to a match that is sure to result in a loss. Through this lens, rage quitting and opposition feeding aren’t just examples of poor sportsmanship, they’re time optimization tactics.

The rewards themselves are only a part of the equation -- the individualized metrics system used to determine who gets rewarded undermines the social dynamics of the team. A team’s performance is the product of its member’s collective contributions. As Robert Huckman and Gary Pisano found in their 2005 research into cardiac surgeon performance found, no matter how exceptional a surgeon’s personal skills were, their patient outcomes weren’t notably better than baseline when they had to perform without the benefit of their normal team. This is the reason why, in the military, we culturally frame success and achievements external to individuals. Promotions, performance evaluations, and awards are obviously individualized, but even those are framed in terms of service or contribution to the unit, mission, and others. This knowledge isn’t limited to the military domain, either; as Babe Ruth famously put, “The way a team plays as a whole determines its success. You may have the greatest bunch of individual stars in the world, but if they don't play together, the club won't be worth a dime." So a system that only measures and values deterministic and individualized stats not only emphasizes the wrong thing, it totally ignores much more significant contributions to a team that a team member can provide. Some contributions can’t be measured or even correlated to deterministic outcomes; what is the quantitative measured value for coordinating the team’s efforts, analyzing and communicating information that contributes to critical situational awareness, or for identifying and remedying intra-team issues. The individualized stats are also problematic because individuation is a cerebrally antagonistic process to teamwork. Team operation fundamentals are all about best practices and techniques to maximize the synchronistic work of its members while minimizing the effort, cost, and waste in achieving said output; it’s about getting to a state where all of the barriers to acting as one have been removed or minimized. Alternatively, an emphasis and reward structure that primes players to think about themselves is not only counterproductive to good team function, but works to reframe teammates as mere obstacles to success.

For example, Overwatch’s medal system seems to be intended to encourage individual skill growth, discourage deliberate underperformance, and mitigate the externalities of unfortunate match-making, but it enables a dissociative mentality for players on a team. Players on a winning team intent on maximizing their XP earnings are incentivized to abandon their team and their common objectives to the extent that it doesn’t change the outcome. A player on a losing team could either blame their team and try to mitigate the sunk cost by playing selfishly, or become caustic to whoever they blame to get them to quit and potentially reroll a replacement player of a higher skill. Even to players that aren’t looking at their team strictly transactionally, their measure of each teammate will be influenced by a design that reduces individuals’ contributions to the team (and by extension, their culpability in their collective success or failure) to individualized performance statistics. Not only is this false, as Huckman and Pisano found, it is detrimental and disruptive in a context where players should be spending their brain cycles on ways to maximize the actions of their teammates, and working to capitalize on the opportunities that their teammates have created for them. Once the team mentality is replaced by that of individuals in this situation, how else could a player interpret their teammates other than as simple contributors and obstacles to success?

Finally, the procedural match-making approach to team formation teaches players to think of each other as expendable. The match-making system is an understandable and necessary feature in a market where player want to be able to play whenever they want with our without their irl friends. The system’s shortcomings may be inevitabile in light of said want, but still have to be acknowledge if they are to be leveraged or mitigated. Specifically here, when you pair expendability with a design that already primes players to think in terms of transactional relationships with other players, it shouldn’t be a surprise when players treat each other more like disappointed customers than invested partners.

I want to point out that this isn’t a problem with these design features in isolation, but their unsuitability for a team-coop game. For example, individual performance metrics made sense in early team-coop games like the Team Fortress mod for Quake -- in 1996 -- in a world when nascent match-making technology over Kali and GameSpy were regularly augmented by irl friends. It doesn’t in the same way in the anonymized and prolific digital culture of 2019. Cosmetic rewards are a natural fit in CRPGs and similar spaces where they are marks of progress, but become more insidious when coupled with a a FOMO approach to player engagement (and addiction). Also, it is important to realize that behaviors, like designs, carried over from contexts can result in dramatically different interpretations and results -- when a player quits because they don’t stand a chance of winning in StarCraft, their “GG” is an act of courtesy so as to not draw out a game longer than necessary, but in a team game like Overwatch and LoL, it’s rude and frustrating, so how and what designs incentivize should be considered. The point isn’t to simply criticize Overwatch; as I pointed out earlier, Overwatch has pioneered more than a few design features that work to facilitate teamwork, and that should be acknowledged, borrowed and/or adapted as we attempt to craft better cooperative digital spaces. The point is that even toxic players are an opportunity for reflection in how what we communicate through our words, actions, and designs echo through the spaces we create.