Institutional similarity drives cultural similarity among online communities

How communities and organizations develop depends greatly on their institutions and cultures. Culture provides community members with a group identity and behavioral guidance, while institutions constitute a set of structuring rules that constrain members’ behaviors. Yet, it is difficult to disentangle the effects of institutions and cultures on a community’s evolution due to their mutual effects on each other. For example, Alesina, Cozzi, and Mantovan show that various preindustrial institutions can lead to long-lasting cultural differences in people’s perceptions of poverty and wealth¹. These differences in cultural values in turn influence policy choices and their effects today. Thus, to understand how institutions and cultures affect community development over time, we have to investigate the relationship between these two factors. Indeed, the importance of the relationship between institutions and culture has been widely recognized in economics^2,3,4, sociology^5,6,7,8, anthropology^9,10,11, political science¹² and communication^13,14,15.

Cultural effects are typically more difficult to detect than institutional effects because “top-down” institutional changes work more directly and are easier to operationalize and observe than changes in culture. Individuals in most institutional or organizational settings do not have any power over those institutions, making it more difficult for their cultural preferences to directly influence institutional development. However, there is one domain in which the tangled relationship between culture and institutions is more direct, observable, and mutual: online communities. Online communities are an ideal laboratory for understanding the mutual effects of institutions and culture for three reasons: (1) they are similar to real-world communities in that institutions directly regulate community members’ behaviors, and members can internalize the rules into their cultural values and preferences^4,16,17,18; (2) they are different from real-world communities in that members of online worlds can choose to migrate between communities at low cost, offering them more bargaining power over institutional structure. Community members can then select against or directly shape the institution’s development towards their own cultural values and preferences. (3) Users who self-select into the same communities share a sense of identity and preferences that form the community group culture, making it possible to infer similar cultural preferences from communities that share overlapping group membership. This observable pattern provides a unique lens into the effects of culture on a community’s formal rules.

Player traffic between customized self-governing Minecraft servers

The video game Minecraft provides a useful context for discussing the relationship between subcommunities and their rules. Minecraft is a massive multiplayer online game that allows for various autonomous user activities, including building with blocks, exploring a virtual world, gathering resources, exchanging goods, and engaging in game combat. Importantly, in Minecraft, users can establish and manage their own private servers for playing the game. These servers function both as spaces other users can explore and communities that users can engage with¹⁹. Someone who sets up a server, the system administrator, takes the responsibility of governing it. To achieve success at building a community of players around their server, administrators have to recruit and retain repeat visitors who can and do migrate between servers at a low cost. Administrators also face constraints in physical resources (e.g., RAM, CPU, bandwidth, monthly server fees) and virtual resources (e.g., software-based currency, reputation systems), all of which must be carefully managed to provide the membership with a quality game experience. Minecraft has been used in science for education^20,21, design^22,23, and the study of self-governance^19,24.

Rules in Minecraft

In the Minecraft ecosystem, administrators who run private servers rely on custom software-based institutions to manage limited resources and solve collective action problems. These software “plugins” are modular programs that administrators can install on their servers to automatically implement rules and other political-economic constructs. Plugins can allow for certain behaviors or activities, or improve the experience of them. For example, “factions” is a plugin that allows administrators to socially subdivide their community. Others prohibit or punish rule violations. For example, the “AntiCheat” plugin prohibits cheating behavior in the game, while “Combatlog” is used topunish unwelcome aggression. By “mixing and matching” plugins and fine-tuning their settings, server administrators craft highly customized formal institutions and implement a social structure that can solve problems and achieve governing goals.

As of 2016, when our data collection ended, the Minecraft community has developed almost 20,000 plugins categorized under 16 types by the Minecraft developer community. (The plugin categories are Admin Tools, Anti-griefing Tools, Chat Related, Developer Tools, Economy, Fixes, Fun, General, Informational, Mechanics, Role Playing, Teleportation, Website Administration, World Editing and Management, World Generators, and Miscellaneous. See https://www.curseforge.com/minecraft/bukkit-plugins/world-editing-and-management.) Among those, Frey and Sumner¹⁹ identified four types that directly related to governance: top-down administration, communication, economy, and information.

Plugins in the administration category allow administrators to execute additional control over server states and player behavior toward preventing or remediating problem behaviors among the games anonymous and young users. Plugins like WorldGuard permit the administrator to manage vandalism by rolling parts of the world back to prior snapshots, while GroupManager and Nations help administrators distribute administrative burdens over a hierarchy of “moderator” users with elevated rights. Plugins in the communication category facilitate interpersonal communication by providing additional or higher bandwidth channels for peer-to-peer communication. For example, the popular “Dynmap” plugin renders a dynamic web-based map of the entire world that players use to coordinate their actions and find each other. Informational plugins provide more channels for broadcasting messages and regulations to the community. For example, the “AutoMessage” plugin makes it easier for administrators to send specific contextual information to users automatically in response to environmental triggers, while “LogBlock” helps players resolve conflicts on their own by encoding by making publicly accessible all prior changes to all locations of the world. Economy plugins protect private property rights and facilitate resource exchange. Plugins like “iConomy”, “ChestShop”, and “Signshop” all support market exchange, either peer-to-peer or peer-to-administrator, while plugins like “Lockette” and “Townies” implement private property rights on top of Minecraft’s common property default.

Culture and membership in Minecraft

Drawing from one element of the sociological conceptualization of culture, we understand the culture in Minecraft as a set of cultural repertoires based on shared practices and experiences, and isolate it behaviorally in terms of the server communities that players self-select into: a major part of the ecology of the game is that players can choose what community they join. A cultural repertoire is acquired through players’ experiences in different servers and interactions with other players. The communities that a user selects into thus give a sense of the range of identities that the user holds. And when many users share overlap in their range of identities, evidenced by their tendency to traffic in the same subset of servers, we infer that they share this sense of culture, operationalized here both as overlap in the set of subcultures, and frequent interaction over a set of similar communities. To be clear, we are not attempting here to define the cultural identity of users in terms of the communities they visit, nor the subculture of a community by the users who visit it: we are attempting to define a set of communities as culturally similar by the existence of a large and consistent group of users who travel together between them.

Based on our definition of Minecraft culture—repertoires of shared meaning that provide references for preference and behaviors—it is reasonable to measure the cultural similarity between servers by dual membership; servers with a high share of dual membership will share a similar cultural repertoire either because, having advertised on similar markers of identity, they attract the same types of users, or because of the knowledge and practices that their dual-membership users contribute to both group repertoires. Accordingly, servers with a low share of dual membership are less likely to share a similar group repertoire because their members are less likely to have similar identities or experiences in Minecraft.

Given the close connection between shared membership and server culture similarities, two claims become apparent. On one hand, shared membership can lead to institutional similarities. First, shared membership facilitates information transmission in institutional isomorphism, which refers to a process that drives one organization to resemble others in the same environmental condition⁵. DiMaggio and Powell⁵ provide one explanation for the institutional isomorphism in Minecraft, that is, the uncertainty and risk in the organizational environment drive the communities to mimic the institutions of each other. Second, shared membership can influence institutional development through user preferences. Server success is almost solely dependent on recruiting and retaining repeat visitors, so users indeed have the bargaining power to negotiate with administrators about institutional decisions. Through this kind of mechanism, users’ preferences can possibly influence institutional development. Thus,***

H1: Shared membership between Minecraft servers causes them to become more institutionally similar.

On the other hand, Institutional similarity can drive shared membership. Individuals’ behavioral and cultural preferences in Minecraft are cultivated and internalized through institutions. When individuals internalize an institutional logic, they may self-select to the institutions close to their previous institutional experience in Minecraft. First, individuals who learned certain behavior in their first Minecraft community might suffer social costs in institutions that do not reward those sets of behaviors. Comparing to learning new behaviors and preferences in a different environment, choosing a similar enough environment provides lower learning cost and higher average payoff. Second, individuals may normalize institutional logic and enforcement, resulting in cultural persistence^18,25. Although cultural persistence may also happen in Minecraft, given the low cost of migration between Minecraft communities, it is more likely that individuals will migrate to communities that provide similar institutional experiences. Thus,

H2: Institutional similarity between Minecraft servers causes them to exhibit more shared membership.

Although they are phrased in a way that may sound mutually exclusive, our method, grounded in network dynamics and dynamical systems, permits an approach under which both or neither may be true simultaneously, as in the case that institutional similarity and shared membership are mutually constitutive.