The Φ-Framework

Organizational coherence analysis through the See / Spec / Split lens

Every queue exists because someone upstream doesn’t know what someone downstream needs.

The Problem

Most organizational failures trace to the same root: information that didn’t reach the right place at the right time. A developer who doesn’t know the spec is ready. A team that can’t see the queue forming at the next station. A decision sitting in someone’s inbox for three days.

Slow shipping, rework, politics, burnout, meetings that accomplish nothing: all information-processing failures. The Φ-Framework makes them diagnosable.

The Channels

An organization’s coordination capacity (R_avail) is the sum of its information-processing channels. Each channel has distinct cost, persistence, and scalability. Most organizational problems are channel mismatches: expensive channels doing cheap work, or no channel at all where one is needed.

Channel	What It Does	Cost	Examples
Φ_surface	Makes state visible without conversation (stigmergy)	Lowest: persists, scales, self-serve	Kanban boards, CI dashboards, pheromone trails, shared databases, version control
Φ_formal	Encodes expectations in documented, management-controllable protocols	Medium: upfront design cost, then runs unsupervised	Handoff specs, checklists, API contracts, standard work, decision criteria
Φ_tacit	Learned routines, habituated patterns, institutional memory distributed across people	High to build (accumulates with time), fragile (destroyed by reorgs and turnover)	Who-actually-calls-whom, undocumented workarounds, tribal knowledge, “how things really work”
Φ_comm	Real-time dialogue that generates responses to unanticipated situations	Highest: doesn’t persist, doesn’t scale, vulnerable to social capture	Meetings, 1:1s, Slack threads, persuasion, face-to-face negotiation

Φ_formal and Φ_tacit together compose Φ_rule, the total protocol capacity. In mature organizations, Φ_tacit dominates Φ_formal because most coordination patterns were never documented. This is why reorganizations hurt: a reorg can reset Φ_formal overnight, but Φ_tacit decays on its own schedule, creating a valley where total capacity drops below requirements.

What the channels must carry

The coordination requirement (R_req) decomposes into two parts:

Requirement	What Generates It	Which Channels Can Handle It
R_routine	Coordination demands resolvable by pre-specified responses	Any channel: Φ_surface, Φ_formal, Φ_tacit, or Φ_comm
R_novel	Situations requiring real-time human judgment to generate a response that doesn’t yet exist in any protocol	Φ_comm only. Protocols and substrates cannot handle what they weren’t designed for.

R_novel further splits into uncertainty (missing data, any channel can supply it) and equivocality (conflicting interpretations, requires rich face-to-face Φ_comm with social cues and meaning negotiation). This is why Toyota keeps team leaders despite massive Φ_formal investment: standardized work handles routine variance, but exceptions need a human who can judge.

The viability condition: R_avail ≥ R_req, with the additional constraint that Φ_comm ≥ R_novel. When capacity falls below requirements, agents act on stale or missing information. The symptoms are predictable: duplicated effort, conflicting decisions, quality defects, escalating rework.

Three Moves

When you find a bottleneck, there are three moves. Apply them recursively to every queue you can find. You stop when the remaining wait time is in genuinely novel work that requires human judgment. That residual cost is the compression floor.

See the Queue

Make work visible. Where does it wait? For how long? If you can’t see the queue, you can’t fix it. Build Φ_surface.

Spec the Handoff

Define “ready for the next station.” Write it down. If the upstream team doesn’t know what downstream needs, every handoff generates rework. Build Φ_rule.

Split the Traffic

Not everything needs the bottleneck. Route by complexity. Auto-approve what’s low-risk. Reserve the expensive channel for what actually requires it.

The General Solution Three words applied recursively to every queue: See. Spec. Split. The content changes per bottleneck; the pattern holds. Fixing one queue reveals the next. You keep going until you hit the compression floor, where the remaining coordination cost is genuinely irreducible for your task complexity.

The Diagnosis

A Φ-Framework report analyzes a company’s organizational architecture through this lens. For each of the four channels, across every domain (engineering, hiring, handoffs, decisions, team formation, power dynamics), we ask: is this channel strong, weak, absent, or misallocated?

Then we find the queues. Where does time die? What waits, and why? The answers stay concrete: this handoff has no spec, this queue has no visibility, this decision routes through a bottleneck that doesn’t need it.

Every company has a paradox: spectacular success in one dimension coexisting with structural failure in another. The framework explains why both exist simultaneously, and what to do about the failure without destroying the success.

Reports

Origin

The Φ-Framework grew out of research on information-processing constraints in organizations. Effective teams everywhere converge on similar sizes, spans, and coordination patterns, because these reflect a universal tradeoff between information-processing requirements and capacity.

The academic version has propositions, proofs, and falsifiers. The operational version has three channels and three moves. This site is about the operational version.

Intellectual debts: Galbraith’s information-processing view (1974), Thompson’s interdependence typology (1967), March’s exploration/exploitation tradeoff (1991). The framework reduces their insights to operational diagnostics.

Φ_rule = Φ_formal + Φ_tacit. In a 20-year-old organization, Φ_tacit typically dwarfs Φ_formal because most coordination patterns were never written down. This is measurable: organizations with high tacit-to-formal ratios experience deeper performance valleys during reorganizations (Prediction 4 in the paper).

The anti-pattern: add people, add meetings, add tools. Adding people widens one station but doesn’t fix handoff specs. Adding meetings adds Φ_comm to a problem that may need Φ_formal. Adding tools builds Φ_surface for a path you haven’t mapped. Diagnosis first, capacity second.

The compression floor is borrowed from information theory. Below a certain density, every word removed loses essential meaning. Below a certain coordination cost, every process removed loses essential coordination. The goal is to reach the floor, not to go below it.

One metric: end-to-end delivery time for your most common work type. Elapsed wall-clock time from “someone asked for this” to “they have it.” Ignore velocity, throughput, and story points. If that number halves, you won.

Recurring patterns across reports:

• Protocol-mediated systems (high Φ_surface + Φ_rule) always outperform conversation-mediated systems (pure Φ_comm) at scale.

• Financial success subsidizes coordination failures. The crisis arrives when margins compress, not when the dysfunction starts.

• Every founder-led company shares the same deepest risk: the org architecture is the founder.