feedback-learning

Power With vs Power Over

Also known as:

Distinguish between coercive power (power over) and collaborative power (power with). Build relationships and movements based on shared power.

Distinguish between coercive power (power over) and collaborative power (power with) to build relationships and movements based on shared power.

[!NOTE] Confidence Rating: ★★★ (Established) This pattern draws on Nonviolence Theory.

Section 1: Context

Organizations, movements, and public institutions operate within dense networks of interdependence—yet the default governance reflex remains unidirectional: the few decide, the many comply. In corporate hierarchies, this shows as top-down strategy cascades that treat frontline workers as execution machines. In government, it manifests as policy imposed without lived-experience input from affected communities. In activist movements, it surfaces as charismatic leaders whose vision overrides collective deliberation. In tech products, it appears as design decisions made in isolation, then force-fitted onto users who had no voice in shaping them.

This ecosystem is fragmenting. Trust erosion accelerates. When power flows only downward, feedback loops break. The system becomes brittle: it can execute single visions efficiently but cannot adapt when conditions shift. People disengage. Burnout spreads among those expected to enforce coercion. Innovation stalls because the very people closest to real problems are excluded from problem-solving.

Yet pockets of vitality exist where power is genuinely shared—where decisions emerge from collective sense-making, where authority is earned through listening, where influence flows multidirectionally. These systems show stronger resilience, faster learning, and deeper commitment from participants. The pattern names this distinction and makes it workable.

Section 2: Problem

The core conflict is Power vs. Over.

Power-over systems extract obedience through hierarchy, control mechanisms, or threat. They’re efficient in the short term—decisions are swift, accountability appears clear, resources concentrate. But they depend on constant enforcement. Dissent goes underground. Information distorts as it travels up chains of command. People comply only while watched.

Power-with systems distribute decision-making authority, make space for genuine disagreement, and require slower, messier deliberation. They feel inefficient at first. But they generate commitment, surface hidden knowledge, and build resilience through distributed learning.

The tension is real: shared power requires more time and cognitive load upfront. Centralized power offers speed. Both feel necessary.

What breaks when this tension stays unresolved: Coercive power-over systems calcify. They become deaf to early warning signs. People develop learned helplessness or quiet rage. When crisis hits, the system has no distributed adaptive capacity and collapses. Power-with systems, without naming and intentionally practicing the distinction, drift toward performative inclusion—the appearance of shared power while decisions still concentrate. People feel gaslit. Trust dies faster than in openly hierarchical systems.

The pattern’s work is to make the distinction visceral and actionable, so practitioners can consciously choose which power form fits which situation—and commit fully to the consequences of that choice rather than claiming both.

Section 3: Solution

Therefore, make the power architecture explicit in your value-creation system: audit where power actually lives, name the decision-rights you’re willing to share, and redesign feedback loops so that those most affected by decisions can influence them.

This pattern operates by shifting power from a hidden asset into visible infrastructure.

In power-over systems, authority is opaque: rules exist, but their origins and flexibility are unclear. People internalize coercion as natural law rather than choice. In power-with systems, authority is transparent and regularly renegotiated. This shift changes everything downstream.

When you name “here, this decision is truly shared” and “here, this decision is centralized because Y,” you create clarity. People can then give genuine consent or consciously withdraw it. This is more honest than pretending all power is shared when it isn’t. The honesty itself—the willingness to be vulnerable about where power actually lives—becomes the root system that holds collaborative capacity.

Nonviolence Theory calls this “structural honesty.” Gandhi, in India’s salt campaigns, made power dynamics visible through disciplined, public action. He didn’t hide British coercion; he revealed it so plainly that moderate publics began withdrawing consent. In organizations, structural honesty works similarly: when managers explicitly say “I’m deciding this, and here’s why I’m not including you in this choice,” people can adjust their trust calibration accordingly. When they say “we’re deciding this together,” shared accountability becomes real.

The mechanism: transparency about power creates choice. Choice generates genuine commitment—or honest withdrawal. Either way, the system is no longer rotting from hidden resentment. Energy can redirect toward actual collaboration where it exists, and toward realistic negotiation where it doesn’t.

Section 4: Implementation

Corporate contexts: Begin by mapping decision-rights in your organization. For each major decision category—strategy, hiring, budget, product priorities, working norms—write explicitly who has final say: the founder alone, a leadership team, a council with veto power, or all affected parties. Post this map visibly. Then pilot one high-stakes domain (e.g., product roadmap) where you shift from presentation-to-employees to genuine co-creation with users and frontline staff. Create decision-making processes that require synthesis of input before conclusions, not just comment windows that get ignored. Measure success not by speed of decision but by whether people who influenced the outcome feel heard and whether implementation quality improves from their participation.

Government contexts: Institute mandatory participatory design for policies that affect specific communities. Don’t consult after drafting; co-author from problem definition onward. Create “power audits” where citizens and staff together map where authority actually resides (often discovering it’s frozen in silos) and where it could be shared. Design appeal and revision mechanisms into every significant policy rollout—built-in acknowledgment that initial implementation will reveal blind spots and that power to amend belongs partly to those experiencing the policy. Train public servants to distinguish between “technical expertise determines this choice” (power-over is appropriate) and “values and trade-offs determine this choice” (power-with is required).

Activist contexts: Name leadership structures explicitly. If your movement has rotating spokespersons, say so and rotate visibly. If certain people hold veto power over direction, acknowledge it and explain why (e.g., “because they hold legal liability”). Create forums where challenges to strategy can surface without punishment. Institute regular power-audits within your organizing team: “Are we actually practicing what we’re advocating? Where are we replicating the hierarchies we’re fighting?” Make failure in this domain a learning prompt, not shameful silence. Document decisions and how they were made so new members can see power working, not just feel excluded from invisible processes.

Tech contexts: Design products with users, not for them. This means early, ongoing involvement of end-users in defining problems and evaluating solutions—not UX testing of fixed designs. Build feedback loops into your product roadmap so that users can see how their input shaped what you built next. Create governance structures (even advisory boards with real influence) for platform policies that affect user expression or data. Make algorithms and their trade-offs transparent; let communities decide which values matter most (speed vs. accuracy, reach vs. safety) rather than hiding these choices in code. If you’re building AI systems, distribute decisions about what those systems optimize for—don’t let engineers or executives alone decide what “better” means.

Section 5: Consequences

What flourishes: Genuine shared power generates rapid, distributed learning. When frontline staff, users, or community members influence decisions, they surface problems before they metastasize. Commitment deepens—people sustain effort not from compulsion but from knowing their voice matters. Relationships become the real organizational architecture; trust becomes the bottleneck that accelerates or slows all other work. Psychological safety spreads, allowing people to name mistakes early. Innovation accelerates not from top-down directives but from permission to experiment within shared constraints. Movements built on power-with relationships attract and retain people; those built on power-over rely on continual recruitment to replace burnout.

What risks emerge: Shared power requires slower deliberation and higher emotional labor upfront. Impatient practitioners revert to power-over when the system feels sluggish. Without active maintenance, power-with systems drift toward performative inclusion where voices are solicited but decisions proceed unchanged—this is worse than honest hierarchy because it corrodes trust. The commons assessment scores signal this risk: resilience (3.0), ownership (3.0), and autonomy (3.0) are moderate, not high. This pattern maintains existing vitality but doesn’t necessarily build new adaptive capacity. If the shared power becomes routinized—”we always consult users but never act on what they say”—the pattern becomes hollow ritual. Watch especially for “consultation fatigue,” where people stop showing up because their participation has never changed outcomes. Rigidity emerges when power-with becomes dogma: sometimes centralized decision-making is more ethical (e.g., a parent protecting a child from harm, or an expert making a technical call that protects others). The pattern’s integrity depends on matching power form to context, not universalizing either approach.

Section 6: Known Uses

Civil Rights marches and salt campaigns: The U.S. Civil Rights Movement deliberately distinguished power-with organizing from power-over tactics. In the sit-ins, marches, and voter registration campaigns, organizers trained participants in nonviolence not as weakness but as a strategic power-with stance: we’re acting together, in full visibility, with shared discipline and shared risk. Compare this to racist power-over enforcement (police violence, legal intimidation). By choosing power-with among organizers and participants, the movement revealed the moral bankruptcy of the opposition’s power-over tactics. Gandhi’s salt campaign similarly made power visible through collective action—Indians walking to the sea to make salt, not asking permission, not hiding. The British could only respond with arrest, which revealed the coercive structure to moderate publics. The pattern worked because power architecture became undeniable.

Mozilla’s governance of open standards: Mozilla’s design processes for Firefox and web standards shift between power-with and transparent power-over. Developers worldwide contribute code, but decisions about direction—which proposals advance, how security trade-offs get weighed—involve community voices explicitly. When Mozilla decided to deprecate certain APIs or prioritize privacy features, they published reasoning and gave community members real opportunities to shape implementation. This attracted and retained talent far beyond what a proprietary browser could. The pattern’s weakness: when Mozilla made decisions perceived as top-down (e.g., removing certain privacy tools under corporate pressure), the distributed community couldn’t override them. But the infrastructure was transparent enough that people could choose exit or voice—they weren’t surprised by hidden agendas.

Participatory budgeting in NYC: Starting in 2009, New York’s ranked-choice voting process for neighborhood budgets distributed real decision-making power to residents about how to spend city capital. Residents research, debate, and vote on projects. The consequence: projects reflect lived experience (fixing subway stairs instead of building new monuments), and participation builds civic engagement across neighborhoods. The pattern’s limits: participatory budgeting touches only a fraction of city spending; larger budgetary decisions remain centralized. But where it operates, it’s proven that people make thoughtful, generative choices when given real power and real information.

Section 7: Cognitive Era

AI and algorithmic systems are reshaping this pattern in three ways.

First, opacity deepens the power-over risk. When decisions are made by opaque machine-learning systems, humans lose even the partial visibility of human hierarchy. You can’t appeal to an algorithm’s conscience. Practitioners must now demand not just that decisions be shared but that the decision-making process itself—what the model optimizes for, what data trained it, what trade-offs it embeds—be transparent and contestable. Power-with becomes: communities co-determine what fairness means before training begins, not after.

Second, distributed intelligence creates new leverage for power-with. If AI systems can surface patterns in massive datasets, communities can use them to reveal hidden power structures. Workers can analyze wage data to negotiate collectively. Communities can audit whether algorithmic systems are biased before deployment. The tech becomes a commons engineering tool if governed through power-with, or a surveillance tool if locked in power-over.

Third, AI at scale tempts centralization. Training large language models or vision systems concentrates power in whoever controls the training process. Practitioners building with AI must consciously distribute this power: open-source the models, share training data, let communities fine-tune systems for their contexts. Otherwise, the pattern disappears entirely—replaced by benevolent-seeming AI that has no mechanism for human contest or redirection.

The tech context translation becomes urgent: How do you build products with AI such that users retain power-with the system, not just power-over its outputs? This is an unsolved design problem. Most current AI products concentrate power in the vendor. The pattern points toward federated learning, transparent model cards, community fine-tuning, and economic models that let users own their data. Without these, AI products are the purest form of power-over yet built.

Section 8: Vitality

Signs of life:

People unprompted share feedback about decisions and see it surface in the next iteration. This is the visceral signal that power actually moves.
Dissent emerges visibly in meetings and forums—not whispered in parking lots afterward. Psychological safety is high enough that disagreement feels safe.
New people entering the system can articulate where power lives and why. The architecture is legible.
Decision-making speed is slower for high-stakes choices but faster for implementation because people have commitment rather than compliance burden.

Signs of decay:

Consultation processes exist but never produce visible changes. The system performs inclusion while decisions proceed unchanged.
People stop showing up to forums and meetings. Participation fatigue signals that power-with is theater.
Conflicts that should surface—disagreements about values or direction—stay hidden. Compliance masks resentment.
New members feel confused about authority: who actually decides? The architecture is opaque again, having reverted to hidden hierarchy.
Leadership becomes a rotating cast because holders of real power remain invisible and those in visible roles become targets for resentment.

When to replant: Redesign this practice when you detect decay (usually at 8–12 month intervals). Don’t wait for breakdown. Small audits—”map actual power again; identify where it’s drifted from stated structure”—prevent hollow rituals. Replant after significant membership turnover, after a major decision that revealed your power architecture was misaligned with your values, or when you notice psychological safety dropping. The right moment is when the system is still functional enough to reflect without crisis urgency, so you can redesign with genuine power-with participation rather than emergency top-down fixes that prove the pattern never took root.