Part A: Current-State Diagnosis — C4 (Cohesive Action Group)
01What is the structure?
A small, cohesive group where knowledge is still emerging through shared action — and a mediator emerges to help turn that knowledge into action. Cohesion is the engine. The mediator is the gearbox.
02The structural vector
Connectivity
Low
Centrality
Medium (5)
Clustering
High (8)
03What does it maximize?
1.Responsiveness and shared knowledge creation — the group processes information collectively and learns through shared action.
2.Learning — integration of tacit knowledge through shared action, while preserving group learning.
3.Coordinated execution toward shared goals while preserving group learning.
04How does it work?
Knowledge is still created collectively through shared action. The mediator does not control processing — they synthesize and broadcast. The knowledge remains essentially tacit, but becomes explicit through the mediator's synthesis. The group still collaborates directly — the mediator is not a controller but a broadcast mechanism.
05What breaks first?
1.The mediator starts replacing peer coordination instead of complementing it — cohesion erodes.
2.The mediator becomes an authority — the clan character erodes.
3.The mediator's relational capacity is exceeded — coordination overhead returns.
06Structural shift indicators — checklist
Mark the phenomena that actually appear — they signal the need for a structural shift.
☐Is the mediator shifting from synchronization to decision-making — is the collective voice weakening?
☐Is dependence on the mediator growing — could it become a structural single point of failure?
☐Is knowledge stabilizing and no longer requiring collective learning — is the mediator starting to filter rather than only broadcast?
☐Can the mediator no longer replace norms and trust — loss of cohesion?
07The structural profile
Five structural categories — with a network visualization under each.
Learning & Knowledge
Knowledge still develops through shared action — but an internal mediator helps turn it into action. The mediator does not control processing — they synthesize and broadcast. The processing center stays in the cohesive group; the mediator is the synthesis point, not the source of knowledge.
Maximizes: Depth of collective knowledge + the ability to turn it into action.
Limit: When the mediator starts filtering rather than only broadcasting — collective knowledge distorts.
Learning
Coordination & Collaboration
Coordination flows through a mediator who synchronizes and broadcasts — but does not control. The mediator reduces coordination overhead and enables collective action. The group still collaborates directly.
Maximizes: Coordinated execution without dismantling the group — cohesion and execution coexist.
Limit: When dependence on the mediator exceeds their capacity to coordinate — or when the mediator shifts from a broadcast mechanism to a control point.
Coordination
Power & Dependence
Influence stems from trust and expertise — and concentrates in a mediator who serves the group. Authority emerges from and serves the group — it is not an external hierarchy.
Maximizes: Trust-based authority that serves collective knowledge.
Limit: Dependence on the mediator grows over time — and may become a structural single point of failure.
Power
Resilience
Resilience to member loss is high, but resilience to losing the mediator is medium. The group keeps functioning without any member who is not the mediator. But losing the mediator brings back coordination overhead.
Maximizes: Group resilience that holds even under pressure.
Limit: The failure point concentrates in the mediator role.
Resilience
Scale
Growth is limited by the mediator's capacity to hold trust relationships and process coordination flow. The structure scales better than C1 — the mediator reduces coordination costs.
Maximizes: Expansion with reduced coordination.
Limit: When the mediator's relational capacity is exceeded, or when coordination demands exceed one person's throughput — the structure tends to move to multiple mediators (C2) or formalize coordination.
Scale
Speed Map — what it enables, maximizes, challenges
Speed is not a structure — it is an outcome. Each kind of speed is produced by a different weight distribution in the network (nodes or edges). One glance: where the weight sits, what it maximizes, what it challenges.
Weight in nodes — knowledge lives in people: tacit, context-dependent
Weight in edges — organizational capacity relies mainly on the coordination mechanisms among people
Throughput Velocity
Weight in nodesStrong
Maximizes
Information travels fast between any two points; short path lengths.
Challenges
High connectivity cost; noise rises with density.
Responsiveness Velocity
Weight in nodesMedium
Maximizes
The network senses change and reacts; driven by external structural holes (betweenness).
Challenges
Discovery must flow back inward; without it, signals are caught but not acted on.
Decision Velocity
Weight in nodesMedium
Maximizes
Decisions enter and exit fast; requires distributed authority (low-moderate centrality).
Challenges
A mediator node becomes a serial bottleneck if it turns into a controller.
Output Velocity
Weight in edgesWeak
Maximizes
High-volume output against a fixed target; knowledge embedded in edges (processes), parallelizable.
Challenges
Freezes knowledge; stiffens the network and cuts collective-learning capacity.
Bottom line: responsiveness and output pull in opposite directions — the first strengthens with weight in nodes, the second with weight in edges. Choosing between them is a real structural decision, not a tuning.
Tiles · Self Organization Studio — Part A: Current-State Diagnosis (C4)
Part B: Desired State — Two Strategic Options
Summary
Given that the user requested to raise performance, two structural alternatives were proposed. The choice between them depends on what matters more — throughput and processing volume, or responsiveness and agility:
Option 1 — M2 (Collaborative Hub): an edge-weight architecture that keeps a central direction and maximizes throughput and processing volume. Suited when the goal is to process a high volume of flow efficiently, with continued central direction. The tradeoff: a tendency to concentrate dependence on central nodes.
Option 2 — T1 (Agile Network): an agile architecture that maximizes responsiveness — fast signal sensing and fast processing simultaneously. Suited for fast teams that need short build-test-fix loops. The tradeoff: sensitivity to scale — the structure collapses beyond 15 people.
Full expansion — Option 1: M2 (Collaborative Hub)
01What is the structure?
The Hub weakens slightly. First lateral ties appear between the Spokes. Local cohesion begins. The Hub becomes a formal coordinator with exclusive decision-routing — and first lateral ties carry the knowledge flow.
Diagnostic note
The C4→M2 transition is a move from a single informal mediator to one formal hub with exclusive decision-routing and first lateral ties for knowledge flow. The goal: reduce load and build partial resilience — without abandoning the central direction.
02What does it maximize?
1.Hub load reduction — first lateral ties take part of the flow, so the Hub no longer mediates everything.
2.Partial resilience and adaptability — redundancy emerges: connected Spoke pairs can communicate even if the Hub is hit — without abandoning the central direction.
3.Exclusive decision-routing — decisions and conflicts route through one formal hub, while first lateral ties carry day-to-day knowledge flow.
4.Initial conditions for tacit knowledge — explicit knowledge is still dominant, but repeated interaction between connected Spokes begins to generate tacit knowledge.
5.Local peer coordination — connected Spoke pairs can co-create at the local level.
03How does it work?
Signal source
From the Hub (explicit knowledge) + repeated interaction between Spokes (emerging tacit knowledge)
Processing center
Shifts — from a single informal mediator to one formal hub; first lateral ties begin to carry initial local processing
Coordination mechanism
Shifts — the Hub still directs, but part of the flow moves to first lateral ties
Knowledge type
Explicit knowledge dominant; tacit knowledge begins to form through repeated interaction
Power source
Concentrated in the single formal hub; well-connected Spokes accumulate informal influence that must be contained
04When does it fit?
1.Hub load reduction is needed — the structure begins to share load through first lateral ties.
2.The Hub is still the directing center — but units begin to develop a shared understanding of peer work.
3.Explicit knowledge is still dominant — but initial conditions for tacit knowledge emerge.
4.Spokes are capable of local peer coordination — connected Spoke pairs can co-create at the local level.
5.Partial redundancy emerges — several Spoke pairs can maintain communication if the Hub is hit or disrupted.
05What does it enable?
1.Hub load reduction while keeping a central direction.
2.Partial resilience and adaptability without abandoning the direction.
3.Processing continuity through clear routing — with partial redundancy at the edges.
4.Local peer coordination and the beginning of tacit knowledge creation.
06What breaks first?
1.Lateral ties grow quadratically — if they multiply too fast, the structure loses its Hub-and-Spoke character and moves toward M4.
2.The Hub remains a central dependency — removing it still severely damages the network.
3.Spokes with many lateral ties accumulate informal influence — a parallel power structure begins to form.
07Structural shift indicators — checklist
Mark the phenomena that actually appear — they signal the need for a structural shift.
☐Are lateral ties multiplying — Spokes starting to coordinate directly and bypass the Hub?
☐Are high-connectivity Spokes accumulating informal influence — a parallel power structure taking shape?
☐Is Hub load dropping but not translating into faster decisions?
☐Are lateral ties becoming dense enough to make the Hub redundant?
08The structural profile
Five structural categories — with a network visualization under each.
Learning & Knowledge
The processing center shifts — from a single informal mediator to one formal hub. The hub still processes and distributes most of the knowledge, but pairs of connected Spokes begin to process repeated interactions locally. Explicit knowledge still flows from the hub — and the initial conditions for tacit knowledge form between connected Spokes.
Maximizes: Hub load reduction through initial local processing — without losing the central direction.
Limit: When most knowledge is created in distributed interactions and can no longer converge for processing at the hub.
Learning
Coordination & Collaboration
The coordination mechanism shifts — routing of decisions and conflicts becomes exclusive to the formal hub. First lateral ties between Spokes carry day-to-day knowledge coordination — but not authority.
Maximizes: Authority clarity — one exclusive routing point, without choking knowledge flow at the edges.
Limit: When the volume of decisions reaching the hub exceeds its capacity, or when lateral ties grow too dense and render the hub redundant.
Coordination
Power & Dependence
Power concentrates in the single formal hub — and well-connected Spokes accumulate informal influence that must be contained. Control of the flow stays with the hub.
Maximizes: Decision consistency through a single authority point — provided the informal influence at the edges is addressed.
Limit: Dependence on the single hub remains a structural failure point; a parallel informal power structure may take shape at the edges.
Power
Resilience
Resilience rests on partial redundancy at the edges — connected Spoke pairs can keep communicating if the hub is hit — but dependence on the hub remains high.
Maximizes: Partial redundancy improves the edges compared to a pure star.
Limit: Dependence on the hub remains a structural failure point — a disruption there paralyzes a significant portion of the network.
Resilience
Scale
Growth happens by reducing hub load — exclusive decision-routing and shifting knowledge flow to lateral ties — not by adding hubs to a core.
Maximizes: Growth is possible without overloading the hub — as long as authority stays routed.
Limit: Lateral ties grow quadratically — if they multiply too fast, the structure loses its hub-and-spoke character and drifts toward M4.
Scale
Speed Map — what it enables, maximizes, challenges
Speed is not a structure — it is an outcome. Each kind of speed is produced by a different weight distribution in the network (nodes or edges). One glance: where the weight sits, what it maximizes, what it challenges.
Weight in nodes — knowledge lives in people: tacit, context-dependent
Weight in edges — organizational capacity relies mainly on the coordination mechanisms among people
Throughput Velocity
Weight in nodesMedium
Maximizes
Information travels fast between any two points; short path lengths.
Challenges
High connectivity cost; noise rises with density.
Responsiveness Velocity
Weight in nodesStrong
Maximizes
The network senses change and reacts; driven by external structural holes (betweenness).
Challenges
Discovery must flow back inward; without it, signals are caught but not acted on.
Decision Velocity
Weight in nodesWeak
Maximizes
Decisions enter and exit fast; requires distributed authority (low-moderate centrality).
Challenges
A mediator node becomes a serial bottleneck if it turns into a controller.
Output Velocity
Weight in nodesStrong
Maximizes
High-volume output via parallel autonomous hubs; knowledge stays in people (tacit, local) — not formalized into processes.
Challenges
Systemic ceiling = cross-cluster broker capacity; as clusters multiply, systemic output converges to that capacity.
Bottom line: high output here stems from parallel autonomous hubs — not formalization. Knowledge stays in people (tacit, local); the systemic ceiling is cross-cluster broker capacity.
Full expansion — Option 2: T1 (Agile Network)
01What is the structure?
Cluster Triple: Adhocracy + Clan + Market. Three typologies from three different quadrants converge to the same structural space. The vector: (7, 3, 7) — high connectivity, low centrality, high clustering.
Diagnostic note
The C4→T1 transition is a move from a cohesive group with a single mediator to an agile network combining group cohesion, diverse signal access, and light coordination. This is a choice for structural agility — simultaneous sensing and processing.
How reactivity happens — the network mechanics
In T1 the processing center is local and parallel at the edges — within the cohesive group — rather than at a central node. The triple dimension produces simultaneous sensing-and-processing: (1) Input — bees and boundary ties bring diverse signals from outside (connectivity=7). (2) Processing — the cohesive group (clustering=7) processes simultaneously in short build-test-fix loops; low centrality (3) ensures no central bottleneck waits for processing. (3) Action-reversal — the light coordinator synchronizes the collective output and does not process it themselves; short path lengths distribute the response to all nodes.
Network actions required to reach it from C4
1.Lower centrality — turn the single mediator into a light coordinator: from central processing to synchronization only, to avoid a bottleneck.
2.Keep high clustering — preserve internal cohesion, since it enables the collective parallel processing at the edges.
3.Raise connectivity — build boundary ties and organizational bees for diverse external input and short path lengths that distribute signal and response.
4.Keep small scale (5-15) — so the triple stays stable and does not collapse into a simpler structure.
02What does it maximize?
1.Fast signal sensing and fast processing simultaneously — the network combines adhocracy's diverse signal access with the clan's collective processing speed.
2.Agile-like work patterns with a structural basis — short feedback loops, collective learning, iterative execution.
3.Parallel collective processing — the team processes information collectively in short build-test-fix loops.
4.Access to diverse knowledge — structural holes at the boundaries provide access to diverse information outside the group.
5.Light coordination (not authority) — a coordinator synchronizes the collective output against external demands without replacing the collective intelligence.
03How does it work?
Signal source
From within the cohesive group (collective sensing) and also from outside through boundary ties
Processing center
Shifts — collective and parallel within the group, in short build-test-fix loops
Coordination mechanism
Shifts — group norms (clan) + light internal synchronization (market) + external boundary ties (adhocracy)
Knowledge type
Internal tacit knowledge + diverse external signals; convergence through light coordination
Power source
Distributed collectively with one light coordination role (the coordinator)
04When does it fit?
1.Fast signal sensing and fast processing — the network combines diverse signal access with collective processing speed — simultaneous sensing and processing.
2.Internal cohesion enables parallel processing — the team is tight enough to process information collectively in short build-test-fix loops.
3.External signal diversity — structural holes at the boundaries provide access to diverse information outside the group.
4.Present execution coordination — a mediator/coordinator (not a controlling manager) helps synchronize the collective output into executable decisions.
5.Agile-like work patterns — short feedback loops, collective learning, iterative execution.
05What does it enable?
1.Simultaneous fast sensing and processing.
2.Agile-like work patterns with a structural basis.
3.Short feedback loops and collective learning.
06What breaks first?
1.Triple collapse — one of the three dimensions starts to dominate and the structure reduces to a pair.
2.Scale — the cognitive cost of maintaining all three dimensions simultaneously exceeds the group's capacity.
07Structural shift indicators — checklist
Mark the phenomena that actually appear — they signal the need for a structural shift.
☐Is the coordinator becoming an authority — the internal coordinator shifts from synchronization to decision-making — the collective voice weakens?
☐Is there a cohesion-execution conflict — the group's learning pace and execution pace diverge?
☐Is there external signal flooding — the mediator receives more external input than the group can process collectively?
☐Is scale dissolving the triple — the structure too complex to maintain at a larger scale?
08The structural profile
Five structural categories — with a network visualization under each.
Learning & Knowledge
Signals arrive from within the cohesive group (collective sensing) and also from outside through boundary ties. The processing center shifts — collective and parallel within the group, in short build-test-fix loops. The coordinator helps bring the parallel processing to convergence. The coordination mechanism shifts — tacit knowledge is created internally through cohesion (clustering=7), and diverse signals are absorbed through structural holes at the boundary.
Maximizes: Fast internal loops (clan), external input through boundary ties (adhocracy), convergence through light coordination (market) — simultaneous sensing and processing.
Limit: When the coordinator receives more external input than the group can process collectively (flooding), or when scale forces specialization that dissolves the triple.
Learning
Coordination & Collaboration
The coordination mechanism shifts — internal coordination is light via a coordinator, and external coordination via boundary ties. Alignment is high — shared goals and mutual dependence are maintained. Collaboration is strong — the group still co-creates. The coordinator enables convergence without replacing the joint creation.
Maximizes: Light, fast coordination — cohesion enables coordination without hierarchy, and connectivity enables flexibility. Short feedback loops and iterative execution.
Limit: When the coordinator becomes an authority — the collective voice weakens, or when learning pace and execution pace diverge (cohesion-execution conflict).
Coordination
Power & Dependence
Power is distributed collectively with one light coordination role (the coordinator). Group members depend on each other symmetrically, with light dependence on the coordinator for synchronization. The coordinator is not a controlling manager — they help synchronize the collective output and manage the interface with external demands.
Maximizes: Power is distributed — there is no single control point, and influence stems from trust and structural position, not authority.
Limit: When the coordinator role hardens into authority — the clan character erodes, and the triple collapses into a simpler structure.
Power
Resilience
Internal resilience is high — but the structure is vulnerable to triple collapse under scale pressures. Clustering=7 provides internal redundancy. Resilience to removing the coordinator is medium — the group can reorganize around coordination.
Maximizes: Internal resilience stems from cohesion — the group keeps functioning under pressure, and the coordinator is not a critical failure point.
Limit: Under scale pressures — when the cognitive cost of maintaining all three dimensions exceeds the group's capacity, one dimension dominates and the triple collapses.
Resilience
Scale
The structure works best at a small-team scale (5-15 people) where all three dimensions can be maintained. The structure scales with difficulty — as scale grows, the triple demands more explicit and rigid management. Scale forces specialization — the coordinator becomes a manager, and the triple collapses into a simpler structure.
Maximizes: The structure is optimal for small teams — short feedback loops and collective learning are possible at this scale.
Limit: When scale forces specialization — one of the three dimensions starts to dominate, the coordinator becomes a manager, and the triple collapses into a simpler structure (C1, C3/C4, or A1/A2).
Scale
Speed Map — what it enables, maximizes, challenges
Speed is not a structure — it is an outcome. Each kind of speed is produced by a different weight distribution in the network (nodes or edges). One glance: where the weight sits, what it maximizes, what it challenges.
Weight in nodes — knowledge lives in people: tacit, context-dependent
Weight in edges — organizational capacity relies mainly on the coordination mechanisms among people
Throughput Velocity
Weight in nodesStrong
Maximizes
Information travels fast between any two points; short path lengths.
Challenges
High connectivity cost; noise rises with density.
Responsiveness Velocity
Weight in nodesStrong
Maximizes
The network senses change and reacts; driven by external structural holes (betweenness).
Challenges
Discovery must flow back inward; without it, signals are caught but not acted on.
Decision Velocity
Weight in nodesMedium
Maximizes
Decisions enter and exit fast; requires distributed authority (low-moderate centrality).
Challenges
A mediator node becomes a serial bottleneck if it turns into a controller.
Output Velocity
Weight in edgesMedium
Maximizes
High-volume output against a fixed target; knowledge embedded in edges (processes), parallelizable.
Challenges
Freezes knowledge; stiffens the network and cuts collective-learning capacity.
Bottom line: responsiveness and output pull in opposite directions — the first strengthens with weight in nodes, the second with weight in edges. Choosing between them is a real structural decision, not a tuning.
Tiles · Self Organization Studio — Part B: Desired State
Part C: Structural Strategies
AI tags: AI promoter — the strategy turns knowledge into formulated regularity an agent can act on (or perform). AI inhibitor — the strategy preserves tacit knowledge AI cannot read. Not a critique — sometimes holding tacitness is the right call.
Strategies for the C4 → M2 transition (processing volume / throughput)
What drives the transition?
In C4 a single mediator synchronizes and broadcasts the collective knowledge. The move to M2 stems from the mediator reaching coordination capacity: the group grows beyond what one mediator can hold of the entire flow. The solution is neither dismantling cohesion nor fully turning knowledge into formal processes — but turning personal knowledge into shared regularity that no longer depends on the person who originated it, and distributing the load according to it: adding first lateral ties between members, and appointing one formal hub with exclusive routing of decisions and conflicts. The processing center shifts — from one person holding knowledge to a formal hub holding regularity; the central direction is preserved.
Recommended network strategies
1. Add first lateral ties
Network edit: Create first triangles — direct connection between Spokes/members who so far coordinated only through the mediator. Lateral ties take part of the coordination flow.
Structural impact: Reduced mediator load; their betweenness drops; short paths form without mediation.
Risk: Lateral ties grow quadratically — if they multiply too fast the structure moves toward M4 and loses its character.
לפני
←
אחרי
2. Turn the mediator into a formal hub — exclusive routing and shared regularity
Network edit: Turn the single mediator into one formal hub: formal appointment, exclusive routing of decisions and conflicts through it, and turning its personal knowledge into shared regularity that no longer depends on the person. First lateral ties carry the day-to-day knowledge flow — not authority.
Structural impact: Dependence on the person drops — knowledge exists as shared regularity and not only in the mediator; authority clarity through one exclusive routing point; hub load drops as lateral ties carry the knowledge flow; peer coordination emerges at the edges.
Risk: If the routing is not exclusive in practice — people keep bypassing the hub and the structure returns to C4; and if the regularity remains trapped in a person — the hub becomes a bottleneck.
AI promoter
The main promoter is the transition from personal knowledge to shared regularity — the necessary condition for AI adoption. An agent cannot act on tacit knowledge trapped in a person; it acts on formulated regularity. Once the regularity is codified, an AI agent can act on it in parallel to the hub — without replacing the exclusive routing. So codification is both the proof the regularity was achieved, and the infrastructure that lets an AI agent enter as another "hub" acting on the same regularity.
לפני
←
אחרי
3. Create partial redundancy
Network edit: Connect Spoke pairs so they can communicate even if the Hub is hit or disrupted. The redundancy is not full — the direction is still central.
Structural impact: Partial resilience and adaptability without abandoning the direction; dependence on the mediator drops but is not eliminated.
Risk: If redundancy grows too fast it makes the mediator redundant — this is a drift toward M4, not a stabilization of M2.
לפני
←
אחרי
4. Manage lateral-tie growth
Network edit: Intentional control of the lateral-tie creation rate — gradual, not quadratic, addition. Prevent multiplication that loses the Hub-and-Spoke character.
Structural impact: Maintaining the balance between load reduction and preserving the central direction. M2 is an intermediate state — keep it as such.
Risk: Without rate management, the structure drifts to M4 (full decentralization) or returns to C4 (single mediator).
לפני
←
אחרי
5. Prevent a parallel power structure
Network edit: Prevent a situation where Spokes with many lateral ties accumulate informal influence that creates a parallel power structure to the coordination core.
Structural impact: Maintaining systemic control in the coordination core; preventing authority split.
Risk: If Spokes accumulate unbalanced influence, a parallel power structure forms that bypasses the core and harms coordination stability.
לפני
←
אחרי
Bottom line
To increase output volume, the structure must enable more work with less dependence on the informal mediator — through exclusive decision-routing, lateral ties for knowledge flow, and controlled redundancy at critical points. The challenge is calibration: enough lateral flow to reduce load, but not so much that the structure collapses into full decentralization or an informal power center forms. M2 is a stable intermediate state — with central direction and controlled lateral flow, not decentralization.
Strategies for the C4 → T1 transition (responsiveness / agility)
What drives the transition?
To react fast to change, information cannot wait to pass through one person. The structure must allow people to sense, process, and act close to where the signal is created. In C4 a cohesive group has a single mediator who synchronizes and broadcasts knowledge; the move to T1 stems from a need for simultaneous fast sensing and processing — when response speed to the environment matters more than central processing volume, and when the group must combine collective learning with access to diverse external knowledge. The coordination mechanism shifts — the single mediator is replaced by a light coordinator who is not a controller, and external boundary ties are added alongside internal cohesion. This is a choice for structural agility — not full decentralization.
Recommended network strategies
1. Turn the mediator into a light coordinator
Network edit: Shift the mediator role from "synthesize and broadcast" to "synchronize only". The coordinator is not a controlling manager — they help synchronize the collective output and manage the interface with external demands, without replacing the collective intelligence.
Structural impact: Centrality stays low (3); the collective voice is preserved; cohesion remains the engine.
Risk: If the coordinator starts making decisions instead of synchronizing — the collective voice weakens and the structure collapses to C3 (Focused Community).
לפני
←
אחרי
2. Build external boundary ties
Network edit: Create weak ties outward — members sitting at the boundary between the group and external worlds access diverse knowledge. Structural holes at the boundaries provide access to diverse information outside the group.
Structural impact: Connectivity rises to 7; signal diversity grows; the group accesses external knowledge without dismantling internal cohesion.
Risk: If boundary members receive more external input than the group can process collectively (flooding) — the structure collapses.
לפני
←
אחרי
3. Preserve internal cohesion
Network edit: Maintain high clustering (7-8) — the group stays tight enough to process information collectively in short build-test-fix loops. Cohesion is the engine that enables fast processing.
Structural impact: Fast collective processing capacity; tacit knowledge is created internally; coordination is light because norms hold.
Risk: If cohesion erodes in favor of external ties — the group loses collective processing capacity and the structure moves toward A1/A2 (open network).
AI inhibitor
Cohesion relies on tacit norms AI cannot read — preserving tacitness, not a critique.
לפני
←
אחרי
4. Organizational bees — knowledge pollinators between inside and outside
Network edit: Defined boundary roles — members who "fly" out to external networks, gather signals and knowledge, and pollinate them inward. Like bees transferring pollen between flowers, organizational bees transfer knowledge between the group and the environment. They are not gatekeepers who filter — they are pollinators who distribute.
Structural impact: Fast signal sensing — the group accesses a wide range of external environments without dismantling internal cohesion; external knowledge is absorbed and pollinated inward continuously; boundary connectivity fills the role of an environmental radar.
Risk: If the bees become gatekeepers — they filter instead of pollinate and the group loses signal diversity; if they are absorbed outside — internal cohesion erodes and the structure moves to A1/A2.
AI promoter · agent
The pollination function itself can be performed by an agent — gathering formulated external signals and distributing them inward.
לפני
←
אחרי
5. Keep small scale
Network edit: Keep the group at a scale of 5-15 people — the range where all three dimensions (cohesion, connectivity, light coordination) can be maintained simultaneously.
Structural impact: Stable triple; full agile capacity; short feedback loops.
Risk: Beyond 15 people — scale forces specialization, the coordinator becomes a manager, and the triple collapses into a simpler structure (C1, C3/C4, or A1/A2).
לפני
←
אחרי
6. Prevent triple collapse
Network edit: Continuous monitoring of the three dimensions — cohesion (clan), connectivity/boundary ties (adhocracy), and light coordination (market). Prevent a situation where one dimension dominates at the others' expense.
Structural impact: Stable triple; full structural agility.
Risk: If cohesion wins → C1 (pure clan). If execution wins → C3/C4 (coordinated community). If external brokering wins → A1/A2 (open network).
לפני
←
אחרי
Bottom line
The C4→T1 transition is turning the mediator into a light coordinator + building external boundary ties + preserving internal cohesion. The central challenge: keep the three dimensions in balance — cohesion, connectivity, and light coordination — without one dominating. The art is to keep the coordinator as a synchronizer rather than an authority, and the group small enough to maintain the triple — so agility does not collapse into a simpler structure.
Tiles · Self Organization Studio — Part C: Structural Strategies
Example | How the model approaches a structural dilemma
The business question
How do we enable people to make autonomous, focused decisions — that is, to identify where to act quickly — when knowledge is still in the discovery phase?
In the Tiles approach, this is not primarily a question of authority or delegation. It is a question of work architecture: does the structure create the conditions that allow people to recognize the right signals, process them, and decide where to act.
Questions for examination
Signal Quality
Before choosing an intervention, we examine the quality of the signals available to the network.
Is the network exposed to a broad enough surface area to detect new signals?
Do the signals come from a diverse range of sources and viewpoints?
Does the organization encounter signals directly, or only after filtering?
When knowledge is still in the discovery phase, exposure to a wide variety of signals increases the network's discovery capacity.
Signal Flow
How do signals move within the network?
Do they reach the points of action directly?
Do they pass through gatekeepers?
Are there bottlenecks?
Does signal flow support knowledge synthesis or hinder it?
Knowledge Maturity
How mature is the knowledge for decision-making?
Are there existing patterns or regularities that can be relied upon?
Is there a shared understanding that enables local decision-making?
Or does the knowledge still require group processing to identify what to focus on?
The conditions identified above lead to different structural approaches
Alternative Work Architectures
ALocal Processing
Fits when focused decisions depend on rapid interpretation of emerging signals close to where they originate.
SignalsAction nodes
Network structure: Signals → Action nodes
This approach fits when:
The network is exposed to a wide variety of signals and information sources.
There is direct access to signals, without significant mediation.
Short feedback loops exist, allowing rapid identification of patterns and regularity from experience.
How does it contribute to focus?
In such a structure, signals reach the points of action directly and processing occurs close to where they originate. When the conditions above hold, feedback loops are shorter, patterns and regularity crystallize faster, and the ability to identify where to act develops locally, without the need for central processing of all information.
Required condition
Local processing is effective only when signals reach decision points with sufficient diversity and quality. Otherwise, local learning converges on noise rather than useful patterns.
BMediated Processing
Fits when signal quality, flow, knowledge maturity, or skills still do not allow consistent local decision-making. In this case, a mediator (person or system) is added whose role is to improve the conditions for focused decisions.
The mediator does not replace local judgment, but improves access to signals, knowledge synthesis, or information flow.
Examples of structural interventions:
Boundary Spanner
When exposure to diverse signals is missing — add a boundary spanner whose role is to expand the network's surface area and channel new, diverse signals.
Bridges between knowledge hubs
When knowledge is dispersed across groups — create bridges that connect knowledge hubs directly and enable synthesis and knowledge sharing.
Shortening flow paths
When bottlenecks or unnecessary mediation exist — shorten flow paths, reposition the mediation, or reduce dependence on a single mediator.
M2 (Collaborative Hub) — Speed Map
Speed Map — what it enables, maximizes, challenges
Speed is not a structure — it is an outcome. Each kind of speed is produced by a different weight distribution in the network (nodes or edges). One glance: where the weight sits, what it maximizes, what it challenges.
Weight in nodes — knowledge lives in people: tacit, context-dependent
Weight in edges — organizational capacity relies mainly on the coordination mechanisms among people
Throughput Velocity
Weight in nodesMedium
Maximizes
Information travels fast between any two points; short path lengths.
Challenges
High connectivity cost; noise rises with density.
Responsiveness Velocity
Weight in nodesStrong
Maximizes
The network senses change and reacts; driven by external structural holes (betweenness).
Challenges
Discovery must flow back inward; without it, signals are caught but not acted on.
Decision Velocity
Weight in nodesWeak
Maximizes
Decisions enter and exit fast; requires distributed authority (low-moderate centrality).
Challenges
A mediator node becomes a serial bottleneck if it turns into a controller.
Output Velocity
Weight in nodesStrong
Maximizes
High-volume output via parallel autonomous hubs; knowledge stays in people (tacit, local) — not formalized into processes.
Challenges
Systemic ceiling = cross-cluster broker capacity; as clusters multiply, systemic output converges to that capacity.
Bottom line: high output here stems from parallel autonomous hubs — not formalization. Knowledge stays in people (tacit, local); the systemic ceiling is cross-cluster broker capacity.
These are alternative structural responses to the same business question. The appropriate choice depends on the characteristics of the signal environment rather than on a preference for centralized or decentralized decision-making.