Practitioner field guide
System Canvas · Practitioner field guide
Provisional method, not yet validated
Read the system spatially
Use the interactive spatial diagram as an orientation before completing the worksheet.
block columns 3 space G["PURPOSE & GOVERNANCE<br/>What directs the system?"] space C["CAPACITY<br/>People · resources · dependencies"] O["OPERATIONS<br/>Activities · decisions · flows"] R["OUTCOMES<br/>Recipients · benefits · harms"] space F["OBSERVATION & FEEDBACK<br/>What signals return?"] space space V["EVALUATION & REFINEMENT<br/>What persists, fails, or changes?"] space C --> O O --> R G -.-> O R -.-> F F -.-> O R --> V classDef capacity fill:#143c36,stroke:#67c1a6,color:#e9faf4 classDef operation fill:#183e54,stroke:#79bfe5,color:#eef8fd classDef outcomes fill:#313450,stroke:#aeb6fa,color:#f5f4ff classDef governance fill:#423820,stroke:#e3c276,color:#fff6df classDef feedback fill:#302a49,stroke:#ad9bea,color:#f6f1ff classDef evaluation fill:#353624,stroke:#d0bd7e,color:#fcf8eb class C capacity class O operation class R outcomes class G governance class F feedback class V evaluation
The arrows are questions about relationships, not claims that particular flows or feedback are already verified. Ask where resources accumulate; where they move; what delays intervene; how participants learn of outcomes; who may adjust rules; and whose viability is affected.
When to reveal another diagram
| If the overview cannot explain… | Reveal… | Check before trusting it |
|---|---|---|
| A resource shortage or bottleneck | Stock/flow, throughput and delay diagram | Units, rates, inventories, delays |
| A persistent pattern or adaptation | Causal-loop or state-transition view | Rival mechanisms and empirical evidence |
| Disputed observation or misinformation | Evidence/provenance map | Sources, corrections, unknowns |
| Inability to change a rule | Institutional authority / IAD action-situation view | Actual rule enforcement and appeal access |
| Conflicting ends | Goal-tension / 4EM view | Distinguish declared goals from incentivized behavior |
| Undesirable side effects | Outcome and impact evaluation view | Counterfactual, affected parties and time horizon |
Leverage overlay: when selecting a possible intervention, check stocks and buffers, rates and delays, feedback loops, information flows, rules, and goals. The type of lever does not prove its effect. Record responsible authority, baseline, expected mechanism, side effects and revision criteria.
The default canvas: minimum viable diagnosis
Start here before choosing any theoretical lens. The canvas is a discussion and decision aid, not a complete organizational map.
| Field | Plain-language prompt |
|---|---|
| Problem | What undesirable or puzzling outcome keeps recurring? |
| People and boundary | Who participates, who is affected, and what lies outside this sketch? |
| Mechanism | What rules, dependencies, incentives, information gaps or feedback might reproduce it? |
| Evidence | What do we know, what do we infer, and what would disprove the explanation? |
| Judgment | For whom is the result good or bad, under which purposes and criteria? |
| Change | Who could change which lever, and what consequences should we monitor? |
A novice should be able to explain the sketch in ordinary language. Add detail only when it would change an inference, decision or evaluation.
Four practical modes
- Analyze — establish a sufficiently useful picture. Expand Structure or relevant enterprise-architecture viewpoints only as needed.
- Diagnose — explain persistence, show a feedback or strategic mechanism, and write rival hypotheses.
- Evaluate — distinguish declared aims, operating metrics, effects, distributional costs, viability and legitimacy; record uncertainty.
- Refine — choose an intervention, implementation authority, baseline, safeguards and next observation.
The existing five moves are a suggested facilitation sequence, while these modes are capabilities that can be revisited in any order.
When to open a deeper lens
| Symptom | First additional view | Established theory to consult if needed |
|---|---|---|
| Coordination, duplication, brittle dependencies | Structure / Behavior | Enterprise architecture, VSM |
| A rule creates perverse incentives | Authority / Behavior | IAD, institutional economics |
| Disputes over what is true | Knowledge | Evidence evaluation, epistemology |
| Nobody can change an obviously defective rule | Authority / Purpose | IAD, second-order cybernetics |
| Reported success conceals harm to another group | Outcomes / Purpose | Program evaluation, ecological economics |
| Separate institutions reinforce one another’s failures | Behavior / Structure | Compositional games, network analysis |
Escalate model complexity only if the simpler view cannot answer a consequential question. Stop adding diagrams when the next analytical action is to obtain evidence, consult affected parties, or test a change.
Usability is not just aesthetics
Review whether independent practitioners can reproduce the reasoning and avoid false causal conclusions. Record completion time, points of confusion, omissions, unwarranted causal claims, and intervention quality. A visually elegant canvas that encourages unjustified inference is not accessible in the relevant sense.
Five moves
Frame → Map → Explain → Redesign → Test and adapt.
Frame
Name the problematic outcome, affected parties, explicit boundary, time horizon, stated purpose, actual operational objectives and available evidence.
Map: use only the necessary lenses
| Lens | Ask | Draw |
|---|---|---|
| Structural | Who controls resources, depends on whom and crosses boundaries? | Actor/dependency map |
| Behavioral | What recurring actions, incentives and feedback sustain outcomes? | Process or causal loop |
| Epistemic | Who can observe, verify, dispute and correct? | Evidence/information map |
| Deontic | Who decides, must act, enforces and hears appeals? | Rights/obligations matrix |
| Teleological | Which goals are stated, rewarded and contested? | Goal-tension diagram |
| Evaluative | What changes, for whom, and how do we know? | Outcomes and evidence table |
Cross-cutting checks: time, environment, power, strategic spillovers, information, legitimacy.
Explain
Because [rule / dependency / information gap / incentive], actors repeatedly [behavior], producing [outcome], reinforced by [feedback].
Tag each causal link: O directly observed; S source-supported; H hypothesis; U unknown. Include competing explanations and ways to distinguish them.
Redesign
If [authorized actor] changes [lever], [behavior] should change and [outcome] improve because [mechanism].
Compare leverage, feasibility, consent, costs, distributional effects, unintended harms, gaming risks and reversibility. Sometimes better outcomes require redistribution of authority or resources, not refined indicators.
Test and adapt
Specify baseline, leading measures, outcome indicators, side effects, time window, counterfactual where feasible, stopping rule and who revises the intervention.
Reflexivity check
Who selected the goal and metric? What does the regulator know? What incentives shape the regulator? Who can audit or replace it? Expand the boundary if these questions materially change the diagnosis.
Reusable case worksheet
Case, date, source links, method version:
Problem, boundary, time horizon, affected parties:
Declared purposes / operational targets / observed effects:
Structural / Behavioral / Epistemic:
Deontic / Teleological / Evaluative:
Mechanism (mark evidence O/S/H/U):
Alternative explanations:
Overlapping strategic games:
Power and legitimacy:
Intervention A and B:
Who can implement / consent / appeal:
Predicted effects and side effects:
Baseline / indicators / stopping criteria:
Regulator-of-regulator question:
Unknowns and evidence to collect:Avoid
Do not infer causality from a diagram; treat a stated rule as enforced; equate engagement with welfare; equate perceived credibility with truth; assume common stakeholder goals; treat power as universally zero sum; or report proposed interventions as demonstrated improvements.