Collaboration in Pokémon GO

Participant observations, survey with 510 people, and interviews with 25 players of Pokemon GO on coordination online & in-real-life.

Published in the conference of Human Factors and Computing Systems (CHI) 2019: Full paper PDF

Research team

Arpita , Travis W. Windleharth, Rio Anthony Ishii, Ivy M. Acevedo, Dr. Cecilia R. Aragon, Dr. Julie A. Kientz, Dr. Jason C. Yip, Dr. Jin Ha Lee

My role

I led our research team in designing study protocols, conducting interviews, surveys, and data analysis, and writing.

Motivation

Raiding is format of game play in online games in which multiple players have to collaborate to get to a virtual location in the game and defeat a powerful enemy, called the "Raid Boss". Through raiding, players earn in-game rewards and items that otherwise cannot be earned individually. In 2017, Niantic released the raiding feature in their location based game, Pokémon GO. The design of raiding was intended to encourage socializing in-real-life (IRL). Raid bosses would spawn at different physical locations in the game (called Gyms), and players had to coordinate to get a group of 2-20 people together at specific times. We started participating in these raids to understand how strangers, friends, and family members were coordinating and how this design was effecting social dynamics of players IRL and online.

People gathered in parking lots, university campus, and other public areas to raid in Pokemon GO.

Research Questions

  1. How do groups coordinate and interact online and in real life for raiding in Pokémon GO?
  2. What factors contribute to participation, coordination, and social interactions in raids in Pokémon GO?

Methods

Participant observations: Our research team collectively participated in over 1500 raid battles IRL and in online discussion groups and forums such as Discord and Facebook.

Online Surveys: We designed a survey based on these observations and obtained 510 complete surveys from Pokemon GO players in multiple states across the US and other countries from suburban (47%), urban (43%), and rural (9%) regions (gender: 54% Female, 43% Male, 2% other).

Interviews: To triangulate our observations and surveys, we conducted semi-structured interviews with 25 participants around the urban Seattle area (44% Female, 56% Male).

I inductively analyzed the qualitative data with the help of two coders (Ivy and Rio). We created a codebook, distributed the data among us three coders, iterated on the codes, and wrote and shared memos. With expertise of our research team, we reorganized our codes based on theory of small group interactions by Arrow et. al [1].

[1] Holly Arrow, Joseph E McGrath, and Jennifer L Berdahl. 2000. Small groups as complex systems: Formation, coordination, development, and adaptation. Sage Publications.

Affinity modeling of sub-themes from inductive analysis into three group dynamics explained by Arrow et. al. [1].

Findings

Global Dynamics are system level variables. In Pokemon GO these are features that are designed by game designers, for example, location of the Gym, point systems.

Local Dynamics are activities of members who constitute a group. In this game, this includes how the members coordinate across different regions. For example, the need to gather at the same location at a specific time to coordinate goals and actions in the game.

Contextual Dynamics are determined by group’s embedding context. These dynamics vary depending upon a community's cultural norms (e.g., approach towards strangers), social capital (e.g., number of people who play are much less in rural areas than in urban areas and thus, have a disadvantage in ability to defeat the raid boss), and technological preferences (e.g., preferences to use Telegram or Whatsapp over Discord or vice versa). Individual factors are also determined contextually. For example, players who did not want to coordinate with others brought 6 phones to raid with them. Similarly, players who were concerned about their safety or did not want to converse with others, raided from inside their cars.

Design Implications

Design to Bridge Global & Contextual Dynamics

Though designers have intent of encouraging collaboration, challenges and strategies vary across different contexts and cultures. The global dynamics should not be independent of these contextual differences but rather communicate with each other dynamically.

  • Rural regions can get system level bonuses or the raid boss' power can be adjusted dynamically to account for lack of social capital. Such considerations for design will help foster equity of game resources.
  • Community managers can add metadata tags on gyms to let other players know locally known information that regular players should be aware of, such as safety, space constraints, and parking tips.

Design to Scaffold Social Bridging

  • Technology can scaffold reciprocal interactions with in game rewards such as adding in-game rewards for helping out a fellow player.
  • Designers can also balance privacy and safety IRL while allowing for remote communication. For example, allowing players to see that a raid location needs help (such as an S.O.S. message) but not broadcasting the players' identities.

Slides from my CHI 2019 talk: PDF