Synpraxis (Greek: syn + praxis = “acting together”) is the study of how independent agents achieve outcomes together across varying degrees of integration — and how those arrangements fail, are exploited, or turn adversarial. The domain encompasses coordination, cooperation, and collaboration as points on a spectrum, not as interchangeable synonyms, along with their anti-patterns: defection, free-riding, capture, coercion, and parasitism.
The term is coined to name what no existing term covers cleanly. “Cooperation” and “collaboration” each name one region of the spectrum. “Collective action” (Olson) is narrowed to public goods and free-rider problems. “Game theory” covers the formal mathematics but not the institutional, biological, or creative dimensions. “Interdependence” names the condition but not the dynamics. Synpraxis names the full territory: the constructive patterns, the failure modes, and the structural conditions that favor one over the other.
Covers: The coordination-cooperation-collaboration spectrum and its orthogonal dimensions; preconditions for authentic joint action (privacy, agency, identity sovereignty, fair incentives, trust); mechanisms of joint action (consensus — human and machine, governance, incentives, protocols); dynamics under interdependence (reciprocity, defection, coopetition, arms races); anti-patterns and failure modes; institutional design for collective outcomes (commons governance, cooperative structures); biological cooperation (mutualism, symbiosis, kin selection); computational cooperation (consensus algorithms, Byzantine fault tolerance, multi-agent systems, collaborative editing); creative collaboration (co-creation, emergent output, shared intentionality); cooperative play and game design; pattern languages for group process.
Does not cover: Individual cognition or solo performance. Competition and conflict are in scope only where they interact with cooperative dynamics (coopetition, competitive escalation, defection within cooperative frames). Pure market competition without cooperative structure belongs elsewhere. Personal history and case studies of synpraxis in practice are separate forms (future cases), not part of the domain definition.
The cooperation-collaboration inquiry identified persistent axes across all domains examined. The spectrum has three positions — coordination, cooperation, collaboration — but many orthogonal dimensions that do not all move together. A given arrangement can show high trust with low identity-merging, or shared goals with no shared governance. The three positions are useful landmarks, not rigid categories; real joint action often occupies the spaces between them or combines elements from different levels.
| Dimension | Coordination | Cooperation | Collaboration |
|---|---|---|---|
| Division of labor | Align timing, avoid conflict | Split tasks, assemble results | Work intertwined throughout |
| Goals | Avoid mutual harm | Compatible but separate | Genuinely shared |
| Output | Deconflicted actions | Sum of parts | Greater than sum of parts |
| Structure | Rules, protocols | Roles, incentives | Self-organizing, emergent |
| Authority | Central or protocol-based | Each party autonomous | Joint governance |
| Accountability | Each for own compliance | Each for own deliverable | Joint, mutual |
| Information sharing | Signals, minimal | Relevant information exchanged | Open, full transparency |
| Dimension | Coordination | Cooperation | Collaboration |
|---|---|---|---|
| Agency required | Procedural | Behavioral | Intentional, shared cognition |
| Trust required | Predictability | Good faith | Vulnerability, psychological safety |
| Power symmetry | Can be imposed top-down | Accommodates asymmetry | Implies rough equality of agency |
| Identity | Separate, unaffected | Separate, acknowledged | Partially merged (“we” for this purpose) |
| Dimension | Coordination | Cooperation | Collaboration |
|---|---|---|---|
| Risk | None assumed | Minimal, individual | Shared, mutual |
| Exit cost | Trivial | Moderate switching cost | High — entangled work, shared IP/liability |
| Coercion vulnerability | Low (exit is cheap) | Medium (can be compelled) | High (identity is entangled) |
| Conflict resolution | Exit or escalate | Negotiate | Internal governance |
Some mechanisms are not spectrum positions but operate across the spectrum, taking different forms at each level.
Consensus is the most important cross-cutting mechanism. It operates at three distinct levels:
Machine consensus (coordination level) — Algorithms that enable distributed nodes to agree on state without a trusted central authority. The Byzantine generals problem (Lamport, 1982) formalized the challenge; proof-of-work (Nakamoto, 2008) demonstrated a practical solution. Machine consensus is deterministic or probabilistic, operates on data, and requires no shared understanding — only protocol compliance.
Negotiated consensus (cooperation level) — Parties with compatible but separate goals reach agreement through structured process. Parliamentary procedure, standards bodies, treaty negotiation. Each party retains autonomy; the consensus is an agreed outcome, not a shared mental state. Groupware (in its original definition: “intentional group processes and software to support them”) was designed to facilitate this level.
Emergent shared understanding (collaboration level) — Meaning converges through joint work. Not a vote or a protocol outcome but an alignment of mental models that emerges from sustained interaction. Jazz improvisation converging on a feel. A writing partnership arriving at a shared voice. This cannot be algorithmed — it requires the full stack of preconditions (agency, trust, psychological safety).
Whether these three levels are one mechanism at different scales or three distinct mechanisms that share a label is an open question (see [[Is Consensus One Mechanism or Three]]↑). The levels do inform each other in both directions. Machine consensus proved that distributed agreement is achievable without trusted third parties — a result that reshapes what’s thinkable about human coordination. Conversely, groupware research demonstrated that computational solutions for human consensus fail without human preconditions: people will not use collaborative software if they believe it is a panopticon. Privacy is not a feature of consensus tools; it is a precondition for the cooperation those tools are meant to support.
Pattern languages are another cross-cutting mechanism — structured collections of recurring solutions that can operate at any spectrum level. The Group Works Project produced 91 patterns for group process spanning facilitation, decision-making, and collective intelligence. Meeples Together produced patterns for cooperative game design. Alexander’s original architectural pattern language was itself a collaborative artifact. Pattern languages are both a product of synpraxis and a tool for improving it.
The spectrum is not just descriptive — each position has structural preconditions. Remove a precondition and you don’t get a lesser form of joint action; you get an anti-pattern. This dependency chain is a thesis, not established fact; the strength of each link varies and some are contested (see [[Precondition Dependencies in Joint Action]]↑).
| Precondition | What it enables | What breaks without it |
|---|---|---|
| Predictability | Coordination | Collision, deadlock |
| Privacy | Coercion-resistance → agency | Surveillance chills authentic participation |
| Agency | Genuine (not performative) cooperation | Coerced cooperation, compliance theater |
| Incentive compatibility | Sustained cooperation | Rational defection, free-riding |
| Identity sovereignty | Full participation as an agent, not a subject | Capture, exploitation |
| Exit rights | Voluntary participation at every level | Coercion disguised as cooperation |
| Trust (built on the above) | Collaboration | Groupthink (false trust) or paralysis (no trust) |
The dependency runs upward: privacy enables coercion-resistance, which enables agency, which enables genuine cooperation, which (with incentive compatibility and identity sovereignty) builds the trust required for collaboration. Each layer depends on those below it. This is why security infrastructure, identity systems, and incentive design are synpraxis problems — they are preconditions for authentic joint action, not separate concerns.
Two notes on contested preconditions. “Fair incentive distribution” is better framed as incentive compatibility — a system-design property where individual rationality aligns with collective benefit. Bitcoin demonstrated this: the system works because defection is expensive, not because a central authority distributes rewards “fairly.” Fairness is a value judgment; incentive compatibility is a structural property. Second, exit rights deserve explicit status as a precondition rather than being implicit in “agency.” The ability to leave a cooperative arrangement without catastrophic cost is what keeps cooperation voluntary. Markets work as coordination mechanisms precisely because exit is cheap. When exit becomes prohibitively expensive, cooperation shades into coercion regardless of how it started.
The deepest pattern across all domains: cooperation describes behavioral outcomes regardless of cognitive sophistication (bacteria cooperate, OS processes cooperate, nations cooperate). Collaboration implies shared intentional creation — it requires minds that can hold shared representations and produce emergent novelty. But collaboration also requires the most preconditions, which is why it is the most fragile and the most rewarding — and why collaboration is not always the right choice (see [[Is Collaboration Always Superior to Cooperation]]↑).
Understanding how joint action breaks down is not optional — it is half the domain. Anti-patterns are not simply failure modes to avoid; they are dynamics to understand because the same behavior can be pathological or adaptive depending on the system it occurs within. Schneier argues in Liars and Outliers that while defectors threaten trust (liars, criminals, cheaters), certain forms of rule-breaking — whistleblowing, civil disobedience, resistance to unjust laws — help societies adapt and correct flawed norms. The goal is not eliminating defection but managing it. A system with zero defection is either totalitarian or fictional.
This has a design consequence: you cannot build resilient cooperative systems without modeling the predator. Understanding the incentives and motivations of defectors, parasites, and coercers is prerequisite to designing systems that withstand them — or that become antifragile (Taleb), gaining strength from stressors rather than merely surviving them. Security infrastructure, incentive design, and governance all require threat models, and threat models require understanding anti-patterns from the inside.
These dynamics operate at different levels of analysis — individual moves (defection), group dynamics (groupthink), institutional structures (capture) — which is a feature of the domain’s breadth, not a flaw in the taxonomy.
These questions are substantial enough to warrant their own inquiry forms:
This domain page currently carries content that belongs in typed forms. The sections above will be trimmed as forms are extracted. Content remains here until extracted — this is a living document, not a finished index.