Giles Hindle · Published 2021-05-11

Systems Thinking in Practice

privateReview: unreviewedConfidence: high
systems thinkingviable system modelproblem structuring methodssystem dynamicsboundary judgementssystems-thinkingemergent-behaviorviable-system-modelproblem-structuringboundary-judgementsmulti-agent-architectureevaluation-strategyexperiential-learning
Open on YouTube ↗

Summary

Overview

  • Speaker: Dr. Giles A. Hindle
  • Channel: Giles Hindle
  • Main topic: Applied systems thinking for managers, analysts, and policy makers
  • Purpose: To educate managers, analysts, and policy makers on how to apply systems thinking and problem-structuring methods to complex organizational and societal situations. An introductory lecture on applied systems thinking in practice, covering reductionism versus systemic properties, definitions of systems, organizational applications, system dynamics, viable system models, and soft operational research (problem structuring methods).

Topic Map

Introduction to Systems Thinking

  • Explanation: Overview of systems thinking from the perspective of a single interconnected world and science, contrasting with reductionism.
  • Key claims:
    • In reality, there is only one world and therefore one science.
    • Reductionism studies objects in isolation by taking them apart from the bottom up.
  • Examples:
    • Studying a duck mechanically from the bottom up versus biological systems.
  • Terminology:
    • systems thinking
    • applied systems thinking
    • reductionism
    • systemic properties
  • Why it matters: Provides foundational understanding of why traditional scientific reductionism fails to capture systemic behavior.

Defining Systems and Their Varieties

  • Explanation: Exploration of how the term 'system' has been used across physical, biological, designed, abstract, human, and ecological domains.
  • Key claims:
    • Systems are assemblies of things interconnected in some way to create a whole.
    • Different types of systems include physical, biological, designed physical, designed abstract, human (social), and natural systems.
  • Examples:
    • Solar system (physical), central nervous system (biological), heating system (designed physical), decimal system (designed abstract), education system (human/social), ecological system (natural).
  • Terminology:
    • physical system
    • biological system
    • designed system
    • abstract system
    • human system
    • natural system
  • Why it matters: Clarifies the broad applicability and diverse referents of the word 'system'.

Organizations and Society Viewed as Systems

  • Explanation: How theorists and applied systems researchers like Ackoff, Beer, and Checkland viewed organizations and society.
  • Key claims:
    • Theorists began to view society and organizations as systems.
    • Applied systems researchers asked what systems thinking meant for policy making, management, and analysis.
  • Examples:
    • City skylines representing complex societal systems.
  • Terminology:
    • system dynamics
    • viable system model
    • problem structuring methods
  • Why it matters: Bridges academic systems theory with managerial and policy practice.

The Viable System Model (VSM)

  • Explanation: Stafford Beer's model of the five necessary functions for an organization to be viable and adapt over time.
  • Key claims:
    • Successful systems must be able to adapt and survive over time.
    • Identified five necessary functions: System 5 (policy/identity), System 4 (intelligence/environment), System 3 (control/overview), System 2 (stability/coordination), and System 1 (sub-systems/operations).
  • Examples:
    • Modelling organizations and prison services using VSM.
  • Terminology:
    • viable system model
    • VSM
    • sub-system
    • environment
    • systemic properties
  • Why it matters: Offers a rigorous framework for designing and diagnosing organizational structure and viability.

Soft Operational Research and Problem Structuring Methods (PSMs)

  • Explanation: The emergence of soft OR in the 1970s and 1980s to help groups think, learn, and address unstructured complex situations.
  • Key claims:
    • PSMs offer a way of representing situations to enable participants to clarify predicaments, converge on actionable issues, and agree on commitments.
    • Focuses on experiential learning rather than system design.
  • Examples:
    • Rich pictures, oval mapping, and stakeholder engagement workshops for Trussell Trust and Network Rail.
  • Terminology:
    • soft operational research
    • problem structuring methods
    • PSMs
    • rich picture
    • experiential learning
  • Why it matters: Provides flexible facilitation tools for tackling unstructured, messy, multi-stakeholder problems.

Critical Systems Thinking and Boundary Judgements

  • Explanation: Exploration of multi-methodology, critical systems thinking, and how boundary judgements define what is included in a system of concern.
  • Key claims:
    • How we draw the line between the system and the environment is a matter of judgement.
    • Boundary judgements distinguish the system of concern from its physical and social environment.
  • Examples:
    • Classifying participant relationships as unitary, pluralist, or coercive.
  • Terminology:
    • critical systems thinking
    • multi-methodology
    • boundary judgements
    • unitary
    • pluralist
    • coercive
  • Why it matters: Highlights the subjective and political nature of defining system boundaries and stakeholder relationships.

Key Points

Reductionism vs. Systems Thinking

  • Explanation: Reductionism takes objects apart in isolation to understand them from the bottom up, while systems thinking studies the whole object because systemic properties only appear at the system level.
  • Evidence: Biological and behavioural examples of dogs and owners.
  • Practical implication: Managers must study organizations and social situations as wholes rather than isolated components.

The Five Necessary Functions of Viable Systems

  • Explanation: Stafford Beer established that any viable system requires five distinct functions covering operations, coordination, control, intelligence, and policy.
  • Evidence: The Viable System Model (VSM) framework.
  • Practical implication: Organizational audits can use VSM to identify structural weaknesses in management and control.

Soft OR and Experiential Learning

  • Explanation: Problem Structuring Methods (PSMs) prioritize experiential learning, holistic situation views, and process consulting over expert system design.
  • Evidence: Rich pictures and stakeholder workshops in public sector projects.
  • Practical implication: Facilitators should engage stakeholders directly in mapping and modelling complex, ambiguous problem situations.

Frameworks, Models & Processes

Viable System Model (VSM)

  • How it works: Diagnoses or designs organizational structure based on five interacting management and operational functions operating within an environment.
  • Components:
    • System 1: Operational sub-systems
    • System 2: Stability and coordination
    • System 3: Internal control and overview
    • System 4: Environmental scanning and intelligence
    • System 5: Policy, identity, and ultimate authority
  • When to use: When evaluating organizational structure, autonomy, and long-term viability.

Problem Structuring Methods (PSMs) Logic

  • How it works: Takes a complex situation involving multiple stakeholders and viewpoints, maps it using conceptual tools, conducts modelling and discussion, and leads to action plans.
  • Components:
    • Complex situation
    • Stakeholder groups & conflicting views
    • Mapping (e.g., rich pictures)
    • Modelling & discussion
    • Improvements & action plans
  • When to use: When dealing with unstructured problems, multiple stakeholders, conflicting interests, and significant intangibles.

Examples & Case Studies

Modelling prison sentences using system dynamics and influence diagrams with Dr. E.A.J.A. Rouwette.

  • Illustrates: How system dynamics maps feedback loops and time delays in public services.
  • Lesson: System dynamics helps simulate and understand unintended consequences in policy management.

Child protection system modelling in the UK by Professor David Lane based on the Munro Review.

  • Illustrates: Applying causal loop diagrams to complex social safety services.
  • Lesson: Modelling compliance, rules, and professional judgment improves service understanding.

Rich picture workshops with senior managers at the Trussell Trust and Network Rail.

  • Illustrates: Participatory problem structuring and stakeholder engagement.
  • Lesson: Participants leading the drawing of rich pictures builds shared understanding of strategic situations.

Actionable Takeaways

  • Immediate:
    • Recognize that systemic properties disappear when systems are broken into isolated parts.
    • Use rich pictures to capture stakeholder perspectives on complex problems.
  • Strategic:
    • Apply the Viable System Model to ensure organizations have all five necessary management functions.
    • Embrace process consulting and experiential learning when tackling unstructured strategic situations.
  • Questions to investigate:
    • Where are the boundary judgements being drawn in our organizational problem definitions?
    • Are our stakeholder relationships best characterized as unitary, pluralist, or coercive?

Claims Worth Verifying

  • Stafford Beer developed the Viable System Model (VSM) based on principles of successful surviving systems. (historical fact)
  • Problem Structuring Methods emerged in the 1970s and 1980s through researchers like Ackoff, Churchman, Checkland, Rosenhead, and Mingers. (historical fact)

Notable Quotes

"In order to study biological phenomena, we need to study the object as a whole." (at 2:04) "Systemic properties only apparent at the level of the system. And if we take the biological object to pieces, the systemic properties disappear." (at 2:10) "Beer wanted to know if there were generic features which ALL successful systems shared. If so, organisations must have these features." (at 12:00) "PSMs offer a way of representing the situation that will enable participants to clarify their predicaments, converge on a potentially actionable mutual problem or issue within it, and agree on commitments that will at least partially resolve it." (at 16:30) "Overall purpose of the approach is experiential learning, rather than the design of a system." (at 17:30)

Compressed Summary

  • Reductionism studies parts in isolation; systems thinking studies wholes and systemic properties.
  • Systems span physical, biological, designed, abstract, human, and natural referents.
  • Stafford Beer's Viable System Model identifies five necessary functions for organizational survival.
  • Soft OR and Problem Structuring Methods facilitate experiential learning for unstructured complex problems.
  • Boundary judgements and stakeholder relationship types (unitary, pluralist, coercive) shape systems practice.
  • Keywords: systems thinking, viable system model, problem structuring methods, system dynamics, boundary judgements
  • Core insight: Effective systems thinking in practice requires shifting from reductionist decomposition to holistic stakeholder engagement, experiential learning, and structural viability.

Core insights

5
Mental Modelmedium noveltymoderate evidence

Because systemic properties appear only at the whole-system level, evaluations and optimizations that isolate individual components (prompts, models, tools) cannot be trusted to improve or even predict the behavior of complete agent systems; system-level emergent outcomes need to be evaluated as a whole.

Why it matters

Many agent pipelines are debugged and benchmarked component-by-component, yet whole-system failure can arise from interactions that component-level metrics never expose. Whole-agent scenarios, global traces, and emergent-outcome probes become necessary, not optional.

Generalization

Any artifact comprised of interacting software agents must be instrumented and evaluated at the ensemble level before micro-optimizing parts.

systemic properties only appear at the system level
Open source video
Reductionism studies objects in isolation by taking them apart from the bottom up
Open source video
Architecturehigh noveltymoderate evidence

VSM distinguishes five structurally separate functions that must coexist for an organization to be viable: operations, coordination/stability, control/oversight, environmental intelligence, and policy/identity. A viable agent system likely needs the same architectural separation rather than fusing all roles into one orchestration loop.

Why it matters

Autonomous agent deployments need to survive changes in task distributions and environment shifts. VSM gives a concrete checklist: is there a scanning loop for external change, an internal control loop, and a policy loop separate from day-to-day execution, and do these systems recurse at each level?

Generalization

Any self-management capability—human or machine—can be audited by asking which of the five viability functions exist and how they are connected.

Successful systems must be able to adapt and survive over time.
Open source video
Identified five necessary functions: System 5 (policy/identity), System 4 (intelligence/environment), System 3 (control/overview), System 2 (stability/coordination), and System 1 (sub-systems/operations).
Open source video
Mechanismhigh noveltymoderate evidence

Soft OR / problem structuring methods such as rich pictures and mapping are not about designing a solution to an assumed problem; they are a process for representing a messy situation, eliciting conflicting stakeholder views, and converging on joint commitments. The hard part is structuring the problem, not solving it.

Why it matters

For ill-defined goals, an agent that immediately optimizes against its first model of the user request will often solve the wrong problem. The process itself—building a shared map, making assumptions explicit, testing commitments—is a legitimate first-class agent workflow.

Generalization

Human-agent and multi-agent collaboration on unstructured objectives requires a negotiation and elicitation phase whose output is a structured issue and agreed constraint set, not merely a recommendation.

PSMs offer a way of representing situations to enable participants to clarify predicaments, converge on actionable issues, and agree on commitments.
Open source video
Focuses on experiential learning rather than system design.
Open source video
Mental Modelhigh noveltymoderate evidence

Deciding what is inside the system of concern versus the environment is a boundary judgement—context-dependent, subjective, and often political—not an objective fact discovered from data.

Why it matters

An agent's context, memory, tool access, and permitted scope function as boundary judgements. They determine what the agent can see, whose interests count, and what will be considered success. Leaving them implicit encodes one set of values into the system without stakeholder discussion.

Generalization

Boundary-setting interfaces and governance are as architecturally important as the agent's reasoning engine; them should be explicit, revisable, and auditable.

How we draw the line between the system and the environment is a matter of judgement.
Open source video
Boundary judgements distinguish the system of concern from its physical and social environment.
Open source video
Practicemedium noveltymoderate evidence

A mature strand of applied systems practice explicitly chooses experiential learning over system design as its primary deliverable—meaning analysts/facilitators are process consultants who improve the stakeholders' capability to think together, not technicians who produce an optimal artifact.

Why it matters

For AI-assisted decision making, this shifts the objective from 'generate the perfect plan' to 'enable the group to learn, disagree productively, and commit'. An agentic system optimized only for final answer quality can be counterproductive when the real gap is collective cognition.

Generalization

In multi-stakeholder or governance domains, the measured value of an interactive agent may lie more in process quality (clarifications, learning gains, commitment quality) than in solution artifact quality.

Focuses on experiential learning rather than system design.
Open source video
Provides flexible facilitation tools for tackling unstructured, messy, multi-stakeholder problems.
Open source video

Deep dives

5

Whole-system evaluation for multi-agent systems

Research question

What metrics and harness designs can reliably detect emergent failures in multi-agent systems that component-level benchmarks miss?

Why

Agent pipelines are debugged and benchmarked module-by-module, but failures often arise from interactions among components; evaluation must measure the ensemble before optimizing parts.

systemic properties only appear at the system level
Open source video
Reductionism studies objects in isolation by taking them apart from the bottom up
Open source video
Source video

Viable System Model as a diagnostic audit for agent architectures

Research question

In long-running autonomous agent deployments, does the absence of any VSM function (operations, coordination, control, intelligence, policy) predict specific classes of production failure?

Why

Agent systems that fuse execution, coordination, auditing, environment scanning, and policy into a single loop may be unable to adapt to environmental changes; VSM gives a grounded set of functions to inspect.

Identified five necessary functions: System 5 (policy/identity), System 4 (intelligence/environment), System 3 (control/overview), System 2 (stability/coordination), and System 1 (sub-systems/operations).
Open source video
Successful systems must be able to adapt and survive over time.
Open source video
Source video

Boundary judgement governance for agent context and tooling

Research question

How should decisions about what an agent can see, remember, and do be formalized so that boundary exclusions become explicit, contestable, and revisable by affected stakeholders?

Why

Agent context, memory, tool access, and permitted scope are boundary judgements that encode values and determine whose interests count; leaving them implicit bakes those choices in without discussion.

How we draw the line between the system and the environment is a matter of judgement.
Open source video
Source video

Problem structuring as a first-class agent workflow

Research question

Which problem structuring method steps can be automated in an agent, and under what conditions does structuring-before-solving reduce solving-the-wrong-problem failures?

Why

For ill-defined or multi-stakeholder tasks, an agent that immediately optimizes against its first model of the request will often solve the wrong problem; the representation-building, elicitation, and commitment phases need explicit support.

PSMs offer a way of representing situations to enable participants to clarify predicaments, converge on actionable issues, and agree on commitments.
Open source video
Applied systems researchers asked what systems thinking meant for policy making, management, and analysis.
Open source video
Source video

Process-quality metrics for experiential-learning agents

Research question

How can the quality of an agent's facilitation be measured when its intended contribution is stakeholder learning, shared understanding, and commitment rather than delivery of a final artifact?

Why

In governance and multi-stakeholder situations, optimizing only for solution quality can be counterproductive if the actual gap is collective cognition; process outcomes need to be tracked.

Focuses on experiential learning rather than system design.
Open source video
Source video

Article ideas

4

Systemic Metrics Before Micro-Optimization: Why Your Agent Benchmarks Are Lying

Because systemic properties only emerge from interactions, any agent evaluation that averages component scores will mislead you; the primary quality gate must be whole-system scenarios and measured emergent outcomes.

Angle

A critique of component-level eval culture with a practical mandate for whole-session telemetry and global failure-mode tracking.

Source video

Does Your Agent Company Have Five Systems? Architecting Autonomous Agents with Beer's Viable System Model

Long-lived agent systems should separate operations, coordination, control, intelligence, and policy into distinct subsystems rather than stuffing all functions into one orchestration loop, because viability requires structural variety.

Angle

Applying management cybernetics to software architecture.

Source video

Structure First, Solve Second: The Soft OR Phase Your Agent Is Missing

When tasks are messy or multi-stakeholder, an agent should produce a shared representation of the situation before any proposed solution; problem structuring is productive work, not an interaction tax.

Angle

Making problem structuring methods (PSMs) concrete for AI product designers.

Source video

What Your Agent Can't See Is a Political Decision

The boundary of an agent's context, tools, and permissions is a subjective judgement that determines whose interests it serves; hiding that boundary in implementation makes it no less value-laden and far less accountable.

Angle

Linking AI ethics to architecture through boundary judgements.

Source video

Project ideas

4

Emergence Harness: Whole-Session Agent Evaluation

beyond-evals

Adding a whole-session evaluation harness that scores final, emergent, ecosystem-level outcomes will catch a class of cascade failures that component-level metrics miss in multi-agent orchestration.

Proof of concept

Build a small multi-agent system with interchangeable subagents; run a suite of end-to-end scenarios with injected perturbations; compare aggregated component scores against whole-system outcome scores and list cases where rankings invert.

Measurement

Detection rate and precision of whole-session vs component-only metrics; number of failure modes visible only at system level; runtime overhead.

Source video

VSM-Check: Agentic Viability Audit

new

Mapping an existing agent architecture onto VSM and adding a separate environment-scanning/intelligence loop (System 4) will reduce the frequency of surprise failures when task distributions shift, compared to a monolithic baseline.

Proof of concept

Instrument a deployed agent system to identify VSM functions, add a low-frequency scanning subsystem that watches environment changes, and run long-duration simulations with variable task distributions.

Measurement

Surprise failure rate, recovery time after distribution shifts, system survival rate; audit coverage score.

Source video

GateHouse: Boundary Board for Agent Scopes

gatehouse

Making an agent's context and tool scope an explicit, versioned boundary register that stakeholders can review yields fewer harmful exclusions than an implementation-chosen, static scope, while adding acceptable latency.

Proof of concept

Implement an agent that loads tools and context sources from an auditable scope file; present scenarios where key stakeholders or data sources are outside the default scope; compare explicit review vs static default.

Measurement

Rate of relevant stakeholders/sources missing from final output; false inclusion rate; setup/review time.

Source video

RichPicture Agent: Structure-Before-Solve Planner

movement-lab

For ambiguous, multi-stakeholder task requests, an agent that first elicits and checks a rich picture/causal map will increase participant commitment and reduce plan rework versus an agent that solves from the raw request.

Proof of concept

Build two variants of a planner for unstructured tasks; one converts the request directly into a plan, the other runs a mapping and elicitation dialogue first; conduct A/B tests with users.

Measurement

User-rated shared understanding and commitment score, rework rate, final task success, dialogue overhead.

Source video

Architectural implications

4

Systemic properties only appear at the whole-system level, but agent systems are typically built and validated by improving individual modules such as prompts, retrieval, and tool calls.

Before

Evaluation gates compare isolated prompt/model outputs and then average component metrics into a score.

After

Define end-to-end task scenes and evaluate learned, emergent, ecosystem-level outcomes—collaboration quality, cascade failures, policy adherence—before attributing performance to individual components.

Consequence

New evaluation harnesses, whole-session telemetry, and failure-mode taxonomies are required; component optimization alone becomes suspect.

Source video

VSM separates operations, coordination, control, intelligence, and policy into distinct mutually connected functions rather than one monolithic controller.

Before

One large model prompt or agent loop is expected to execute work, coordinate subagents, scan the environment, ensure stability, and decide policy simultaneously.

After

Introduce structurally different subsystems: operational workers, a coordination/stabilization layer, an internal audit/control layer, an external scanning/intelligence layer, and a policy/identity layer, each with its own loop and handoffs.

Consequence

Agent platforms become more complex, but gain an explicit mechanism for long-term adaptation and diagnosability of systemic failures.

Source video

PSM workflows treat complex situations by representing them, eliciting conflicting views, and converging on issues before commitments—rather than assuming a stationary, well-defined user goal.

Before

Planning starts with a natural-language request converted directly to a task list with no joint representation of stakeholders or conflicts.

After

The agent's early loop explicitly builds a rich picture or causal map, checks it with stakeholders, reframes the problem boundary, and only then selects a solving strategy.

Consequence

Better outcomes for messy sociotechnical tasks, but interaction cost and latency rise; agent designers must choose when problem structuring is worth the overhead.

Source video

Boundary judgements determine what is included in the system of concern, yet content selection, memory scoping, and user authorization are usually implementation choices rather than reviewed design decisions.

Before

A developer selects available tools and context window contents based on convenience, and these choices remain invisible and static.

After

Context and tools are treated as explicitly recorded boundary judgements with rationale, ownership, and possible alternatives that can be reviewed by affected stakeholders.

Consequence

Changing the agent's scope becomes a deliberate act; wrong-scope and exclusion issues can be surfaced earlier, at the tradeoff of more governance overhead.

Source video

Tradeoffs and failure modes

4

Whole-level evaluation vs. component optimization

Benefit

Whole-level evaluation captures properties that only exist when components interact.

Cost or risk

Whole-system measurements are harder to localize, slower, more expensive, and make incremental component progress harder to read.

systemic properties only appear at the system level
Open source video
Source video

Problem structuring vs. direct problem solving

Benefit

Structuring methods handle unstructured, multi-stakeholder problems by creating truthful representations and shared commitments.

Cost or risk

PSM processes trade away crispness and solution-design outputs; because the focus is on experiential learning, the deliverable can feel less concrete and more time-consuming.

Focuses on experiential learning rather than system design.
Open source video
Source video

Boundary judgement scope

Benefit

Choosing a narrower boundary yields a tractable system of concern and makes analysis and control easier.

Cost or risk

A narrow boundary can silently exclude important stakeholders, externalities, and causal feedback, causing a technically valid solution to solve the socially wrong problem.

How we draw the line between the system and the environment is a matter of judgement.
Open source video
Source video

Separation of control functions in VSM

Benefit

Distinct functions give a system explicit mechanisms for stability, internal control, environmental scanning, and policy—improving viability over time.

Cost or risk

The resulting architecture has many interlocking sub-systems; the overhead can be unjustified for small or stable agent systems and can create coordination lag between the layers.

Successful systems must be able to adapt and survive over time.
Open source video
Source video

Open questions

4

What does an operationalized boundary-judgement process look like for an agentic system: who chooses the context/knowledge/tool boundary, how is it contested, and how often is it revisited?

Why unresolved

The source only establishes that boundary lines are subjective judgements, not objective features, so there is no algorithmic mechanism supplied for making the choice.

Research direction

Design explicit boundary-setting interfaces, log boundary decisions as metadata, and experimentally compare fixed vs. revisable context scopes.

Source video

Can VSM be used as a diagnostic audit for existing agentic architectures, and does the absence of any of the five functions correlate with observed failure modes in production?

Why unresolved

VSM evidence in the source comes from modeling organizations and public-sector services, not from generative-agent deployments.

Research direction

Map deployed agent systems onto VSM, identify missing functions, and prospectively test whether adding, say, an explicit environment-scanning loop improves long-horizon task reliability.

Source video

Which metrics capture systemic properties of agent systems so they can be tracked across changes?

Why unresolved

The source contrasts component-level reductionism with whole-level systemic properties but does not define operational system-level metrics.

Research direction

Create canonical multi-agent scenario benchmarks with whole-system outcomes (e.g., robustness to perturbation, coordination efficiency, intent preservation) and study their sensitivity to individual component changes.

Source video

Which PSMs can be translated into agent workflows, and what is the human-AI division of labor when running them?

Why unresolved

The summary describes rich pictures and workshops as social facilitation techniques but says nothing about automating them through AI agents.

Research direction

Prototype an agent that elicits a rich picture from stakeholders and compares output quality and participant learning against traditional facilitated workshops.

Source video

Key claims

7
factualVerification not requested

Systems are assemblies of things interconnected in some way to create a whole.

Evidence

Systems are assemblies of things interconnected in some way to create a whole.

Source video
comparativeVerification not requested

Reductionism studies objects in isolation by taking them apart from the bottom up, while systems thinking studies the whole object because systemic properties only appear at the system level.

Evidence

Reductionism studies objects in isolation by taking them apart from the bottom up

Source video
factualVerification needed

Stafford Beer's VSM identifies five necessary functions covering operations, coordination, control, intelligence, and policy that must be present for an organization to be viable.

Evidence

Identified five necessary functions: System 5 (policy/identity), System 4 (intelligence/environment), System 3 (control/overview), System 2 (stability/coordination), and System 1 (sub-systems/operations).

Question

Does Beer's model indeed posit all five functions as necessary for viability across organizational forms?

Source video
causalVerification needed

PSMs are an effective approach for helping participants clarify predicaments, converge on actionable issues, and agree on commitments in unstructured situations.

Evidence

PSMs offer a way of representing situations to enable participants to clarify predicaments, converge on actionable issues, and agree on commitments.

Question

What empirical evidence from soft OR practice supports the claimed outcomes of PSMs?

Source video
comparativeVerification needed

The appropriate focus of soft operational research is experiential learning rather than system design.

Evidence

Focuses on experiential learning rather than system design.

Question

Is this distinction between learning-focused and design-focused paradigms consistently maintained in the PSM literature?

Source video
opinionVerification not requested

Drawing the boundary between a system and its environment is a matter of judgement.

Evidence

How we draw the line between the system and the environment is a matter of judgement.

Source video
opinionVerification not requested

Successful systems must be able to adapt and survive over time.

Evidence

Successful systems must be able to adapt and survive over time.

Source video

Connections

5