Skip to content

Communication Pattern Review

Diagnostic review — instantiates Organization–Artifact Topology Alignment

Reads the collective's real communication network — who actually coordinates with whom — from its meetings, messages, and handoffs, so the social topology can be compared against the architecture it will imprint.

Before you can align an organization with the artifact it builds, you have to know what the organization's communication graph actually is — not what the org chart claims. Communication Pattern Review infers that graph from behavioural traces: who meets whom, whose pull requests get reviewed by whom, which chat channels bridge which groups, where tickets get reassigned. Its defining move is that it maps the social side of the mirror from observed interaction rather than reporting lines, so that the network Conway's law says the artifact will imprint is made visible and legible. It maps the collective; it does not touch the artifact, and it does not yet compare the two.

Example

A game studio keeps stalling a new co-op mode at the same place: the handoff from the art team to the engine team. A Communication Pattern Review pulls six weeks of aggregated signals — meeting attendance, PR review threads, cross-posted chat, ticket reassignments — and draws the coordination graph. It shows that art and engine almost never interact directly; nearly everything between them routes through a single overloaded tech-artist. The org chart draws the two teams as peers with a clean seam. The real network shows a one-person bridge carrying the entire load, and that bridge is exactly where the mode keeps getting stuck. That map — a load-bearing broker the chart never showed — is the input the rest of the alignment work builds on.

How it works

What distinguishes it from a plain org-chart reading is that every edge is earned from behaviour:

  • Infer edges from traces, not titles — frequency and latency of real interactions (reviews, threads, meetings, reassignments) define who coordinates with whom.
  • Weight the edges by volume or latency, so "who talks a lot" and "where things wait" are distinguishable.
  • Find the topology features that matter — bridges, bottlenecks, and structural holes where a needed conversation isn't happening.
  • Aggregate to the team level before anyone reads it, so the output is a map of the collective, not a dossier on individuals.

Tuning parameters

  • Signal set — which channels count (meetings, PRs, chat, tickets). More channels give a fuller picture but add noise and widen privacy exposure.
  • Aggregation grain — individual, team, or department. Finer grain surfaces the key broker but raises the surveillance risk the privacy guard exists to contain.
  • Time window — a snapshot versus a trailing average. Longer smooths noise but blurs the effect of a recent reorg.
  • Edge weighting — frequency versus latency versus volume; changes whether the map foregrounds busy ties or slow ones.
  • Coordination threshold — how much interaction counts as a real edge. Too low and the graph is a hairball; too high and weak-but-critical ties disappear.

When it helps, and when it misleads

Its strength is that it recovers the real network when the official chart has drifted from it — and the single most useful thing it finds is usually the undocumented broker whose overload is invisible on paper. It is the social-side input every downstream alignment move depends on.

Its central caution is that communication data is a proxy: an absent edge is ambiguous. It can mean two groups are healthily independent, or that they should be coordinating and aren't — the graph alone cannot tell you which. The classic misuse is turning a coordination map into an instrument for ranking or surveilling individuals, or running it backwards to "prove" a reorganization already chosen. The discipline that guards against both is to aggregate to the team level, treat every edge as a hypothesis to confirm with the people in it, and keep the review diagnostic rather than evaluative — which is precisely why the privacy guard is part of the mechanism, not an afterthought.[1]

How it implements the components

Communication Pattern Review fills the social-mapping side of the archetype — only the components an observer of the collective can produce:

  • collective_communication_topology_map — its primary output: the behaviourally-derived graph of who actually coordinates with whom.
  • informal_network_privacy_guard — the aggregation-and-anonymization discipline it runs because the raw who-talks-to-whom data is sensitive by nature.

It does not build the artifact_topology_map — that is the artifact-side counterpart Architecture Dependency Graph Review — and it does not perform the seam_congruence_assessment that compares the two maps, which belongs to the Organization–Artifact Topology Overlay.

Notes

An absent edge is the review's most easily misread finding: silence between two groups can be healthy autonomy or a missing necessary link, and the communication graph cannot resolve the ambiguity on its own. Interpreting it requires the artifact and domain maps alongside — a seam that is quiet and carries a heavy dependency is a problem; a quiet seam with no dependency is fine.

References

[1] Conway's law — the observation (Melvin Conway, 1968) that a system's structure tends to mirror the communication structure of the organization that designs it. The review reads the communication structure precisely because it predicts, and often precedes, the artifact's structure. ↩