# Controlling the Effects of Faction in a Large Online Community ## Decision and recommendation The community should adopt a pluralist moderation structure designed to control the effects of faction rather than attempt to eliminate faction itself. The structure should distinguish disagreement from conduct that threatens other members’ ability to participate; apply rules to majority and minority factions alike; distribute interpretive authority rather than concentrating it in one moderator or one dominant bloc; and make interventions legible through consistent standards and reasons. This recommendation follows from a basic constraint: disagreement cannot be removed without either destroying the liberty that makes a community worth preserving or forcing its members into uniform opinions, passions, and interests. Neither remedy is acceptable or practicable. The relevant question is therefore not how to create consensus, but how to prevent factional power from becoming oppressive. The proposed design has four elements. First, define faction behaviorally: a faction is not merely a group with a shared interpretation, but any majority or minority group whose coordinated passion or interest becomes hostile to other members’ rights or to the community’s continuing interests. Second, protect disagreement as such while intervening against conduct that produces intimidation, pile-ons, coordinated obstruction, or other conditions under which people cannot participate. Third, use representative and distributed decision processes so that no single local majority, highly motivated minority, or moderator faction can settle contested questions solely by force of numbers or coordination. Fourth, require transparent, reviewable decisions so that legitimacy does not depend on everyone agreeing with the outcome. This is not a promise that conflict will disappear. It is a recommendation to make conflict governable. ## The problem: faction is a permanent condition, not an occasional failure A large community contains people with different opinions, passions, interests, attachments, and understandings of the community’s purpose. Those differences naturally produce parties. Some factions will form around interpretations of the rules or the community’s identity. Others will form around personalities, preferred projects, or competing expectations about what participation should feel like. The particular subject will change, but the underlying condition will remain. That fact matters because it changes what moderation can reasonably promise. A rule that treats faction itself as the problem will eventually classify ordinary pluralism as misconduct. It will pressure members to share the same opinions, passions, or interests. Even if such uniformity were desirable, it would be impracticable: people differ, and those differences are part of the community’s life. Liberty supplies the conditions for disagreement just as air supplies the conditions for fire. Removing the condition would also destroy something essential. The alternative failure is to treat every factional conflict as protected expression regardless of its effects. That approach mistakes freedom from enforced agreement for freedom from coercive group behavior. A faction can be a majority or a minority. Its size does not determine whether it is dangerous. What matters is whether its organized force is being used against other members’ ability to participate or against the community’s continuing interests. The harms of unmanaged faction are therefore not limited to disagreement. They include instability in decisions, distrust of commitments, confusion about applicable standards, disregard for the broader good, and threats to the rights of less powerful members. In a community, these harms appear when members cannot predict how rules will be applied, when a popular interpretation becomes effectively mandatory despite formal pluralism, when a highly motivated minority dominates every discussion, or when coordinated pressure makes participation intolerable for people outside the faction. The central moderation problem is consequently the protection of both sides of a legitimate commitment: energetic disagreement must remain possible, but no group should be able to convert numerical strength, coordination, or persistence into a license to exclude others in practice. ## Two remedies, and why one must be rejected There are two broad ways to address faction: remove its causes or control its effects. The first option is attractive because it appears simpler. If disagreement produces conflict, perhaps the community can define one correct purpose, suppress rival interpretations, or require members to adopt a common view of acceptable participation. That remedy would be worse than the disease. Destroying liberty in order to eliminate faction would destroy the condition that makes a voluntary community meaningful. It would also place moderators in the position of deciding which opinions, passions, and interests are sufficiently uniform to count as legitimate. The intervention would not end faction; it would make the authority to define faction the central object of factional struggle. The second version of removal—forcing uniformity without openly suppressing liberty—is no more realistic. Human reason is fallible, interests differ, and members will continue to attach themselves to different interpretations, leaders, and causes. A community can state a purpose, but it cannot make every member experience that purpose in exactly the same way. Attempts to do so will tend to convert ordinary disagreement into suspicion and make enforcement look like preference for one faction’s understanding. The practical conclusion is that the causes of faction cannot be removed. Relief must be sought in controlling its effects. For moderation, this means accepting that disagreement, rival interpretations, and organized advocacy will recur. The design task is to keep those forces from becoming a mechanism by which one group can deprive another of meaningful participation. ## A working definition of factional harm A useful definition must be narrower than “a group I dislike” and broader than “a group that violates an explicit rule.” Faction should be understood as a group, whether a majority or minority, united by a passion or interest that becomes hostile to other members’ rights or to the community’s enduring aggregate interests. Applied to moderation, this definition creates an important distinction. A group advocating an unpopular interpretation is not necessarily factional in the harmful sense. A group disagreeing intensely with moderators is not necessarily factional in the harmful sense. A group that is numerically dominant is not automatically abusive merely because it wins a vote or receives more support. Conversely, a small group is not automatically protected from scrutiny merely because it is outnumbered. The relevant question is whether the group’s behavior is directed toward persuasion and participation, or toward making participation by others impossible, unsafe, or practically futile. The distinction should be evaluated through observable conduct and effects rather than through political labels or the popularity of the underlying view. This standard preserves room for vigorous disagreement while recognizing that rights can be threatened without formal exclusion. A faction may use superior force to decide matters against a minority, or a minority may use persistent coordination to obstruct ordinary administration and disturb the community. Both patterns matter, but they are not symmetrical in every respect. A majority faction is more dangerous because it can retain the forms of popular decision-making while sacrificing the rights and interests of the minority. A minority faction may be disruptive and difficult to manage, yet ordinary voting or representative procedures can often defeat its preferred measure. The moderation system should therefore ask at least three questions before intervention. Is the conduct disagreement about meaning, policy, or purpose? Is it coordinated in a way that threatens other members’ ability to participate rather than merely seeking support? And does the response protect the community’s continuing interests, or does it simply impose the preferred interpretation of the currently powerful group? These questions do not eliminate judgment, but they make the judgment more principled and more contestable. ## Majority power is the primary institutional risk The community should not assume that a decision is legitimate merely because many members support it. Numerical strength can be used against justice and the rights of a minority. A majority can preserve the appearance of popular government while treating its own interest as the whole community’s interest. This is the online analogue of the most important institutional danger in the source argument: a majority faction has both the motivation and the capacity to define normality, shape enforcement, and characterize dissent as illegitimate. If moderators simply follow the loudest or largest bloc, they may reproduce the problem while believing they are being democratic. This does not mean that majority preferences should never prevail. A community must make decisions, and a minority cannot receive a veto over every collective choice. It means that majority rule is a decision procedure, not a complete theory of legitimacy. The rights of minority participants and the community’s longer-term ability to host disagreement must remain constraints on what a majority may do. The same principle applies to popular moderators. A moderator may be trusted by one faction and distrusted by another. A decision can therefore be correct in substance yet still require explanation and review because affected members need to know that the rule was not simply an expression of factional preference. The aim is not to make every intervention immune from criticism. It is to make criticism possible without allowing the criticized group to seize the process. ## Representation and distributed judgment A large community cannot resolve every contested question through direct, simultaneous participation by everyone. Direct aggregation may appear maximally democratic, but a small or highly coordinated majority can communicate and act together against weaker parties. The existence of a vote does not itself prevent factional harm. Representation offers a different mechanism. A chosen or delegated body can refine and enlarge the community’s views by considering broader interests rather than responding only to the immediate passion of an assembled group. For moderation, representation does not require creating a permanent political class. It requires creating a decision process in which contested questions can be considered by people who are not all drawn from the same faction, and in which the reasoning behind a decision can be examined beyond the moment of conflict. Representation is not automatically safe. Representatives can be factious, locally prejudiced, corrupt, or self-interested. A moderator panel can become a faction of its own. A trusted appeal group can betray the interests it was meant to protect. The value of representation therefore depends on safeguards: varied membership, defined authority, transparent reasons, and some means of review or replacement. Scale creates a tradeoff. If decision-makers are too distant from members, they lose local knowledge and may misunderstand the specific context of a dispute. If decision-makers are too close to one discussion, faction, or personality, they may become excessively attached to local interests. The correct design is not “as much centralization as possible” or “as much local control as possible.” It is a mean between the two. A practical structure would assign broad and aggregate questions—such as the community’s general participation standard or the meaning of a recurring category of harmful conduct—to a wider representative process, while leaving context-specific questions to moderators closer to the relevant discussion. The broader process would establish the governing principle; local moderators would apply it to particular cases. This division preserves local knowledge without allowing every local conflict to rewrite the community’s standards. ## Extending the sphere to reduce coordinated domination The strongest institutional insight is that a larger sphere can contain faction by increasing the variety of parties and interests within the decision-making body. When more distinct interests are present, it becomes less likely that a majority will share the same motive to oppress a minority. Even where an oppressive motive exists, dispersed supporters have more difficulty discovering their strength and acting in unison. For a large online community, this does not mean assuming that size alone creates safety. A very large population can still be captured by a highly coordinated bloc, and communication tools can reduce the practical distance between members. The relevant lesson is narrower: do not allow one discussion, one local cluster, one influential personality, or one temporary surge of participation to stand in for the whole community. Decision processes should expose contested questions to a genuine variety of perspectives before treating a local consensus as a community-wide mandate. This is especially important when a factional movement is concentrated in one part of the community. A local “fire” may be intense without representing the whole. The system should contain that conflict long enough to determine whether it reflects a broader principle or merely a concentrated interest. The same reasoning applies to brigading and pile-ons. Their danger is not simply that many people express the same opinion. It is that coordination can transform a numerical advantage into pressure that a targeted member cannot reasonably withstand. The appropriate response is not to suppress the underlying disagreement automatically. It is to control the coordination mechanism when it is being used to deprive others of meaningful participation. This also explains why interventions should be based on behavior and process rather than factional identity. A popular faction and an unpopular faction must be subject to the same stated standard. Otherwise, the rule becomes a device for preserving whichever group currently controls attention. Equal application does not require identical outcomes in every context, but it does require that the same kind of participation-threatening conduct be treated as the same kind of institutional problem. ## Operating logic for moderation decisions The recommended operating logic begins with a presumption in favor of disagreement. Members should be able to dispute the community’s interpretation, challenge moderators, advocate for unpopular positions, and organize support for change. These activities may be frustrating, repetitive, or politically charged without being factional harm in the relevant sense. The presumption changes when conduct threatens participation or the community’s continuing ability to govern itself. At that point, moderators should identify the specific effect being controlled. Is the issue an attempt to silence a minority through coordinated pressure? Is a minority using repeated obstruction to prevent ordinary administration? Is a group treating its own interpretation as permission to disregard the rights of others? Is a decision being made by a concentrated interest that is simultaneously a party to the dispute and the body judging it? That last question is especially important. When moderators are both judges and parties in a conflict, their incentives can distort the decision. They may be personally invested in one interpretation, responsible for an earlier action, or under pressure from a faction they favor. The system should therefore make room for a second view when the moderator’s own conduct or factional alignment is central to the dispute. The purpose of review is not to guarantee reversal. It is to reduce the risk that a single interested party can define the facts, the rule, and the remedy simultaneously. The greater the potential effect on participation, the stronger the reason to expose the decision to a broader and more varied judgment. Explanations should state the distinction being applied: disagreement remains permissible; the intervention addresses a conduct or coordination pattern that threatens other members’ ability to participate or the community’s aggregate interests. This language will not prevent accusations of favoritism or censorship. No institutional design can guarantee that every faction accepts an adverse decision. It can, however, make the decision intelligible and give members a basis for evaluating whether the same principle would apply to their own faction. ## Risks and tradeoffs The first risk is overreach. A broad concern for participation can become a pretext for suppressing difficult, unpopular, or merely offensive disagreement. This would recreate the rejected remedy of eliminating faction by destroying liberty. The mitigation is to require an observable connection between the intervention and participation-threatening conduct, rather than treating controversy or opposition as sufficient. The second risk is under-enforcement. In an effort to protect energetic disagreement, moderators may tolerate coordination that makes participation practically impossible for others. This would allow a faction to exercise coercive power while claiming the protection of free expression. The mitigation is to evaluate effects as well as stated intentions and to recognize that a minority can also create institutional harm. The third risk is representative capture. A moderation panel or review process may become a new faction, particularly if its members share the same local interests or understandings of the community’s purpose. The mitigation is not to abandon representation, but to preserve variety, separate broad standards from local application, and make decisions reviewable. The fourth risk is excessive scale. Distributing every decision across a large process may sacrifice local knowledge, slow intervention, and make members feel unseen. The mitigation is a layered structure: local decisions for immediate context, broader review for recurring or community-wide principles, and a clear distinction between the two. The fifth risk is procedural distrust. If explanations are vague or standards shift with the identity of the faction involved, every decision will be interpreted as favoritism. Transparency cannot remove disagreement, but it can reduce the suspicion that the process is merely an instrument of whichever group currently has superior force. Finally, there is a fundamental tradeoff between speed and legitimacy. A rapidly coordinated faction may require timely action, while a durable decision may require broader consideration. The community should not pretend that one value always defeats the other. It should reserve the fastest interventions for immediate participation threats and use more representative review for questions that establish precedent or change the meaning of community-wide rules. ## Open questions Several design questions require judgment rather than a universal answer. How much local authority should moderators retain before a question becomes community-wide? What size and composition would make a review body sufficiently varied without making it unable to decide? Which kinds of coordination count as ordinary advocacy, and which cross the line into participation-threatening pressure? How should a community distinguish a temporary conflict from a durable factional pattern? The source argument provides principles, not a complete operating manual. It supports controlling effects, protecting minority rights, using representation, balancing scale with local knowledge, and resisting dependence on unusually wise leaders. It does not establish a single threshold for every intervention. Those thresholds must be developed transparently and revised when experience shows that a rule systematically favors one faction. The community should also remain alert to the possibility that its formal structure is working while its practical structure is not. A rule may be neutral in wording but factional in application. A process may permit participation formally while coordinated behavior makes participation unrealistic. The relevant test is therefore not only whether members possess the same nominal rights, but whether those rights remain usable across rival groups. ## Recommendation and next steps Adopt the pluralist moderation model as the community’s governing approach. State explicitly that the purpose of moderation is not to eliminate disagreement or enforce one interpretation of the community’s identity. Its purpose is to control factional effects that threaten participation, minority rights, fair administration, or the community’s continuing interests. Then establish a written distinction between protected disagreement and participation-threatening conduct. Apply it to majority and minority factions alike. Create a layered representative process in which local moderators handle context-specific matters while broader, more varied review addresses community-wide standards and contested precedents. Require decision-makers to explain the principle and effect they are addressing, especially when they are personally involved in the dispute. Finally, review the system for unequal effects across rival groups and revise it when a faction can repeatedly convert popularity, persistence, or coordination into effective control. The objective is not a community without factions. Such a community would either be impossible or achieved only by sacrificing the liberty that gives disagreement value. The objective is a community in which factions can compete without making one another’s participation impossible; in which majority power is constrained by the rights of minorities; in which minority activity can be managed without being mistaken for illegitimacy; and in which the structure of decisions, rather than the virtue of whichever moderator happens to be present, supplies the durable protection. That is the appropriate standard for a large pluralist community: preserve the energy of disagreement, but prevent any faction from turning its energy into domination.