My Flipped Learning · Published 2022-12-21

The basics of systems thinking 2023

privateReview: unreviewedConfidence: high
systems thinkingcomplexityboundariesmethodologiesFeedback loopssystems-thinkingfeedback-loopsEmergenceviable-systems-modelsoft-systems-methodologycritical-systems-heuristics
Open on YouTube ↗

Summary

Overview

  • Speaker:
  • Channel: My Flipped Learning
  • Main topic: Systems Thinking Basics
  • Purpose: To provide viewers with a foundational understanding of systems thinking, its key concepts, and common methodologies used to analyze and address complex systems. An introduction to systems thinking, exploring its core concepts such as complexity, situations of interest, boundaries, interventions, and key methodologies like Causal Loop Diagrams, Viable Systems Model, System Dynamics Modelling, Soft Systems Methodology, Strategic Options Development and Analysis, and Critical Systems Heuristics.

Topic Map

Introduction to Systems Thinking

  • Explanation: Defines systems thinking as a holistic approach to understanding complex systems and their interconnected parts.
  • Key claims:
    • Systems thinking looks at the big picture and interconnectedness.
    • Holistic approach considers things as a whole rather than isolated parts.
    • Small changes in one part can have cascading effects.
  • Examples:
    • Healthcare approaches considering physical, social, emotional, and environmental factors.
    • Food chain population changes having ripple effects across species.
  • Terminology:
    • Systems thinking
    • Holistic approach
    • Interconnectedness
    • Emergent behavior
  • Why it matters: It enables comprehensive understanding of complex problems and leads to more effective, personalized solutions.

Key Concepts in Systems Thinking

  • Explanation: Explores core concepts including complexity, situations of interest, and boundaries.
  • Key claims:
    • Complexity refers to non-linear interactions among many interconnected parts.
    • Situations of interest are specific problems or issues being analyzed.
    • Boundaries define limits or edges of a system, separating what is included and excluded.
  • Examples:
    • Climate change, poverty, and healthcare reform as complex situations of interest.
    • Physical walls of a building vs. abstract organizational responsibilities as boundaries.
  • Terminology:
    • Complexity
    • Situations of interest
    • Boundaries
    • Boundary judgments
  • Why it matters: Understanding these core concepts is essential for framing problems correctly and designing valid interventions.

Methodologies in Systems Thinking

  • Explanation: Outlines specific frameworks and tools used to analyze complex systems.
  • Key claims:
    • Causal Loop Diagrams map variables, arrows, and feedback loops.
    • Viable Systems Model (VSM) uses five key functions for organizational viability.
    • Soft Systems Methodology (SSM) addresses messy, human-centric problems flexibly.
  • Examples:
    • Using VSM for goal-setting, resource allocation, decision making, monitoring, and coordination.
    • Using System Dynamics Modelling for computer-based simulation over time.
  • Terminology:
    • Causal Loop Diagrams
    • Viable Systems Model
    • System Dynamics Modelling
    • Soft Systems Methodology
    • Strategic Options Development and Analysis
    • Critical Systems Heuristics
    • Design Thinking
    • Action Research
  • Why it matters: Provides structured tools for practitioners to model, analyze, and intervene in complex real-world systems.

Key Points

Holistic Perspective

  • Explanation: Systems cannot be fully understood in isolation; their relationships and interactions must be examined.
  • Evidence: Ecosystems and healthcare examples showing multi-factor dependencies.
  • Practical implication: Avoid reductionist thinking when tackling multi-faceted societal or organizational problems.

Emergent Behavior

  • Explanation: Behaviors arise from interactions among system parts rather than individual parts alone.
  • Evidence: Flock of birds behavior emerging from individual interactions.
  • Practical implication: Focus on improving interactions and relationships within systems to alter outcomes.

Boundary Judgments

  • Explanation: Deciding what is inside or outside a system is subjective and impacts analysis.
  • Evidence: Discussion on physical and abstract boundaries and potential biases.
  • Practical implication: Be mindful of subjectivity and assumptions when drawing system boundaries.

Frameworks, Models & Processes

Viable Systems Model (VSM)

  • How it works: Helps understand how complex systems function and how they can be made more effective through five key functions.
  • Components:
    • Goal setting
    • Resource allocation
    • Decision making
    • Monitoring
    • Coordination
  • When to use: When analyzing organizational viability, adaptability, and systemic alignment.

Soft Systems Methodology (SSM)

  • How it works: Systematic approach to problem solving for complex, messy situations involving human values and differing stakeholder perspectives.
  • Components:
    • Defining problem in human values terms
    • Identifying stakeholders
    • Analyzing from multiple viewpoints
    • Resolving contradictions
    • Generating solutions
    • Implementing and testing
  • When to use: In ambiguous situations with multiple stakeholders holding conflicting views.

Critical Systems Heuristics (CSH)

  • How it works: Multidisciplinary approach to understanding and addressing complex systems by questioning boundary choices and stakeholder perspectives.
  • Components:
    • Holism
    • Pluralism
    • Ethics
    • Participatory democracy
    • Reflection
  • When to use: When analyzing social and environmental problems with diverse stakeholder values.

Examples & Case Studies

Healthcare analyzed through a holistic lens rather than symptom-only focus.

  • Illustrates: Interconnectedness of physical, social, emotional, and environmental factors.
  • Lesson: Broadening the system boundary leads to more effective and personalized treatment plans.

Examining a food chain where population decreases in one species ripple across others.

  • Illustrates: Cascading effects and feedback loops in simple systems.
  • Lesson: Small changes in one part of a system trigger widespread systemic consequences.

Actionable Takeaways

  • Immediate:
    • Define situations of interest using clear structured formats.
    • Map out system boundaries mindfully, considering potential biases.
  • Strategic:
    • Select appropriate methodologies matching the problem complexity and available resources.
    • Involve key stakeholders throughout the intervention and decision-making process.
  • Questions to investigate:
    • What are the key variables and feedback loops influencing the system?
    • Who are all the relevant stakeholders, and what are their diverse perspectives?

Claims Worth Verifying

  • Stafford Beer developed the Viable Systems Model (VSM). (historical fact)
  • Peter Checkland developed Soft Systems Methodology (SSM). (historical fact)

Notable Quotes

"Systems thinking is a way of understanding and analyzing complex systems and the relationships between their various parts."

Compressed Summary

  • Systems thinking takes a holistic approach, viewing systems as interconnected wholes rather than isolated parts.
  • Key concepts include complexity, situations of interest, and boundary judgments.
  • Methodologies such as VSM, SSM, CSD, and Design Thinking offer structured toolsets for problem solving.
  • Keywords: systems thinking, complexity, boundaries, methodologies, feedback loops
  • Core insight: Systems thinking provides holistic mental models and methodologies to navigate complexity, uncover root causes, and design effective interventions.

Core insights

6
Mental Modelmedium noveltymoderate evidence

Boundary definition is itself the primary design act in complex-system work: what is placed inside versus outside the system is a subjective judgment (a 'boundary judgment') that determines what the subsequent analysis can even see, and it carries unexamined bias.

Why it matters

In agentic systems the analogous boundary is the context window, tool surface, accessible memory scope, and which signals count as telemetry. These are usually set implicitly by defaults rather than chosen deliberately, which silently excludes the factors most likely to explain failures. Treating the boundary as an explicit, reviewable artifact moves it from an accident of implementation into a governed design decision.

Generalization

Any system-definition step (context scoping, tool exposure, observability coverage, evaluation set construction) is a value-laden inclusion/exclusion choice; making it explicit and reversible is a general reliability practice, not a systems-theory specialty.

Deciding what is inside or outside a system is subjective and impacts analysis.
Open source video
Be mindful of subjectivity and assumptions when drawing system boundaries.
Open source video
Boundaries define limits or edges of a system, separating what is included and excluded.
Open source video
Mechanismmedium noveltymoderate evidence

Emergent behavior is attributed to interactions among components rather than to the components themselves, which implies the correct intervention target is the interaction topology, not the individual part.

Why it matters

Most agent debugging targets the individual component (swap the model, rewrite the prompt, tune a tool). If the behavior is generated by the interaction layer, component-level fixes cannot be reliably effective; the leverage sits in routing, handoff contracts, and message schemas between agents/tools. This predicts that teams instrumenting interactions will debug faster than teams instrumenting components.

Generalization

For any multi-component system exhibiting unexpected global behavior, attribute-first debugging should start at the coupling between parts rather than inside them.

Behaviors arise from interactions among system parts rather than individual parts alone.
Open source video
Focus on improving interactions and relationships within systems to alter outcomes.
Open source video
Architecturemedium noveltymoderate evidence

The Viable Systems Model decomposes organizational viability into five separable functions — goal setting, resource allocation, decision making, monitoring, and coordination — positioning monitoring and coordination as distinct first-class functions rather than byproducts of decision making.

Why it matters

This is an architectural separation-of-concerns template for autonomous systems: runtimes that fold monitoring into the decision path lose the ability to detect their own drift, and runtimes that fold coordination into goal setting lose the ability to reconcile conflicting subagents. Naming these five as separate components gives a concrete checklist for what must be independently inspectable.

Generalization

Control architectures for autonomous systems should separate policy (decision making), resource budgeting, objective management, observation/monitoring, and inter-unit coordination so each can be tested and replaced independently.

Helps understand how complex systems function and how they can be made more effective through five key functions.
Open source video
Goal setting - Resource allocation - Decision making - Monitoring - Coordination
Open source video
Tradeoffmedium noveltymoderate evidence

Methodology selection is explicitly conditioned on problem type and available resources: soft, human-centric methodologies (SSM, CSH, SODA) are designated for ambiguous situations with conflicting stakeholder perspectives, while quantitative approaches (System Dynamics Modelling) are designated for computable simulation over time. There is no universal method.

Why it matters

This argues against a single harness or evaluation regime for all tasks. Tasks with objective measurable dynamics justify simulation/quantitative modeling; tasks dominated by contested values and ambiguity justify structured elicitation and viewpoint reconciliation instead. Committing to one formalism for both classes produces either unfalsifiable narrative or models that ignore the actual conflict.

Generalization

Choose the analysis/engineering formalism to match the epistemic character of the problem (computable vs value-contested), and treat the choice itself as a resource-bounded decision.

Select appropriate methodologies matching the problem complexity and available resources.
Open source video
Systematic approach to problem solving for complex, messy situations involving human values and differing stakeholder perspectives.
Open source video
In ambiguous situations with multiple stakeholders holding conflicting views.
Open source video
Mental Modelmedium noveltymoderate evidence

Critical Systems Heuristics treats boundary choices and stakeholder inclusion as an ethical and participatory act — it questions who is affected by the boundary but left outside it, combining holism, pluralism, ethics, participatory democracy, and reflection.

Why it matters

For deployed AI systems, the affected-but-excluded set (users not represented in the eval set, downstream consumers of outputs, operators) is exactly the population that generates surprise failures and incident reports. CSH provides a named procedure for surfacing them at design time rather than after deployment.

Generalization

Every system boundary creates an affected-but-unrepresented population; making that population an explicit design input is a general safety and requirements-elicitation practice.

Multidisciplinary approach to understanding and addressing complex systems by questioning boundary choices and stakeholder perspectives.
Open source video
Holism - Pluralism - Ethics - Participatory democracy - Reflection
Open source video
When analyzing social and environmental problems with diverse stakeholder values.
Open source video
Mechanismlow noveltyweak evidence

Small local changes propagate non-locally through cascading effects and feedback loops, and Causal Loop Diagrams are the designated instrument for mapping variables, arrows, and loops before intervening.

Why it matters

This is the argument for causal/loop mapping as a pre-change artifact rather than post-incident forensics. It implies that reliability work should enumerate loop structure (reinforcing vs balancing paths) so that a local change can be traced to its second-order effects before rollout, which is directly analogous to change-impact analysis in agent tool/prompt modification.

Generalization

In any tightly coupled system, local modifications should be preceded by an explicit loop map identifying amplifying and dampening paths.

Small changes in one part of a system trigger widespread systemic consequences.
Open source video
Causal Loop Diagrams map variables, arrows, and feedback loops.
Open source video
Small changes in one part can have cascading effects.
Open source video

Deep dives

4

Scope manifests as versioned boundary judgments for AI systems

Research question

Can boundary judgments be encoded as an explicit, versioned, machine-checkable artifact (included tools, data, signals, memory scope, and excluded items with rationale) such that a scope diff mechanically triggers re-evaluation of the affected eval classes?

Why

Boundary choice is currently an accident of implementation defaults (whatever context, tools, or data happened to be wired in), which silently excludes the factors most likely to explain failures. If scope can be represented, diffed, and tied to eval suites, then an exclusion becomes a reviewable assumption with change control instead of an invisible blind spot.

Deciding what is inside or outside a system is subjective and impacts analysis.
Open source video
Be mindful of subjectivity and assumptions when drawing system boundaries.
Open source video
Boundaries define limits or edges of a system, separating what is included and excluded.
Open source video
Source video

Separating monitoring and coordination from the decision path

Research question

Does decomposing an agent runtime into five independently inspectable functions (goal setting, resource allocation, decision making, monitoring, coordination) improve drift detection and conflict reconciliation relative to a single decision loop that folds them together?

Why

Runtimes that fold monitoring into the decision path lose the ability to detect their own drift, and runtimes that fold coordination into goal setting lose the ability to reconcile conflicting subagents. Naming the five functions as separate components yields a concrete checklist of what must be independently testable and replaceable.

Helps understand how complex systems function and how they can be made more effective through five key functions.
Open source video
Goal setting - Resource allocation - Decision making - Monitoring - Coordination
Open source video
Source video

Interaction-level telemetry as a replacement for component-level scoring

Research question

If emergent behavior originates in interactions rather than components, do measurable properties of interactions (handoff contracts, message schemas, retry/return paths, routing topology) correlate with task failures better than per-component scores?

Why

Most agent debugging targets the individual component: swap the model, rewrite the prompt, tune a tool. If behavior is generated at the interaction layer, component-level fixes cannot be reliably effective, and the leverage sits in routing, handoff contracts, and message schemas. This predicts that teams instrumenting interactions debug faster than teams instrumenting components.

Behaviors arise from interactions among system parts rather than individual parts alone.
Open source video
Focus on improving interactions and relationships within systems to alter outcomes.
Open source video
Source video

A decision procedure for choosing the analysis formalism

Research question

Can methodology selection be operationalized as a rubric keyed on problem character (computable dynamics vs value-contested ambiguity) plus budget, with acceptable inter-rater agreement and measurable downstream outcome differences versus a single default formalism?

Why

A single harness or evaluation regime for all tasks produces either unfalsifiable narrative on value-contested problems or models that ignore the actual conflict. Making the selection itself a resource-bounded decision is the engineering consequence of the claim that no universal method exists.

Select appropriate methodologies matching the problem complexity and available resources.
Open source video
In ambiguous situations with multiple stakeholders holding conflicting views.
Open source video
Systematic approach to problem solving for complex, messy situations involving human values and differing stakeholder perspectives.
Open source video
Source video

Article ideas

4

Your Context Window Is a Boundary Judgment You Never Made

In agentic systems the boundary is the context window, tool surface, memory scope, and telemetry coverage, and because these are set by defaults rather than chosen deliberately, agent reliability work begins with making scope an explicit, reviewable, diffable artifact rather than a byproduct of implementation.

Angle

Translate the systems-thinking concept of boundary judgment into a concrete engineering artifact: a scope manifest listing included and excluded data, tools, signals, and stakeholders with rationale, plus change control that re-runs affected evals on scope diff.

Source video

Debug the Coupling, Not the Component

If system-level behavior is generated by interactions rather than parts, then swapping models, rewriting prompts, and tuning tools are component-level fixes applied to an interaction-level defect, which is why teams that instrument handoffs, message schemas, and routing topology resolve failures faster than teams that instrument components.

Angle

Reframe agent debugging as interaction-graph forensics: define measurable interaction properties, show where per-component scores are blind, and argue that attribution must start at the coupling.

Source video

One Harness to Rule Them All Is a Category Error

Committing to a single evaluation formalism for all task classes is a category error: computable, measurable dynamics justify simulation and quantitative modeling, while value-contested ambiguity justifies structured elicitation and viewpoint reconciliation, and applying one regime to both yields unfalsifiable narrative or models that ignore the real conflict.

Angle

Argue for a selection rubric over a universal harness, with the cost model (complexity, budget, stakeholder plurality) made explicit and auditable rather than left as a judgment call.

Source video

Who Is Outside Your Boundary? The Affected-but-Excluded Population as an Incident Backlog

Every deployed AI boundary creates a population that is affected but unrepresented, and that population is exactly the source of surprise failures and incident reports, so surfacing it at design time is a requirements-elicitation and safety practice rather than an ethics afterthought.

Angle

Turn stakeholder-inclusion critique into a concrete design-time checklist: enumerate downstream consumers, data subjects, and operators excluded from the eval set, then pre-register failure hypotheses against them.

Source video

Project ideas

4

Scope Manifest with Scope-Diff Eval Triggering

beyond-evals

Scope diffs in a versioned scope manifest predict failure-class regressions in an eval suite better than version-number-only or prompt-diff tracking; systems with manifest-gated deployments ship fewer scope-induced regressions.

Proof of concept

Define a manifest schema (included data, tools, signals, memory scope, excluded items and rationale), wire it to a fixed eval suite, and replay a series of real or injected scope changes to check whether the diff identifies which eval classes degrade.

Measurement

Precision/recall of scope-diff → degraded-eval-class prediction, and delta in scope-induced regression rate between manifest-gated and ungated deployment paths.

Source video

Interaction-Graph Telemetry

movement-lab

Defects in the interaction graph (handoff contract violations, schema mismatches, unbounded retry/return paths, routing cycles) correlate with task failures more strongly than per-component quality scores.

Proof of concept

Instrument a multi-agent pipeline to emit handoffs, message schemas, and retry/return paths as a graph per task run; label failed runs and compare a graph-defect feature set against a per-component score baseline as predictors of failure.

Measurement

AUC of interaction-graph features vs. component-score baseline for predicting task failure; time-to-root-cause in a debugging trial with and without graph telemetry.

Source video

Five-Function Control Plane

gatehouse

An agent runtime that implements goal setting, resource allocation, decision making, monitoring, and coordination as separately replaceable components detects its own drift and reconciles conflicting subagents faster than a single-loop runtime with folded control, at an acceptable interface overhead.

Proof of concept

Build two runtimes over the same task family: a monolithic decision loop and a five-component control plane with its own monitoring and coordination interfaces; inject drift and subagent goal conflicts into both.

Measurement

Time-to-drift-detection, successful conflict reconciliation rate, and added latency/message overhead of the separated architecture.

Source video

Boundary-Width Sweep

new

Outcome quality rises with boundary width only up to an inflection point, after which added context and interacting variables introduce new failure modes and degrade net outcome per unit cost.

Proof of concept

On a fixed task family, run a controlled sweep of boundary width (amount of context, number of tools, scope of signals considered), scoring each width on outcome quality and resource cost.

Measurement

Outcome quality and quality-per-cost as functions of boundary width, with the location and stability of the inflection point across task families.

Source video

Architectural implications

3

Monitoring and coordination appear as named functions distinct from decision making in the Viable Systems Model rather than as implicit behaviors of the decision component.

Before

Control logic, observability, and inter-unit coordination are implemented inside the same agent/loop that makes decisions.

After

Decision making, monitoring, resource allocation, goal setting, and coordination are separable components with their own interfaces.

Consequence

Drift detection and coordination conflicts become independently testable and replaceable, at the cost of additional interfaces and message-passing overhead.

Source video

Boundary judgments are treated as subjective and bias-prone, and are recommended for explicit, mindful mapping.

Before

System scope is an implicit consequence of implementation defaults (whatever context, tools, or data happened to be wired in).

After

Scope is an explicitly documented artifact stating what is included, excluded, and why.

Consequence

Excluded factors become reviewable assumptions rather than invisible blind spots; the trade-off is added documentation and review latency on scope changes.

Source video

Intervention guidance says to broaden the boundary to reach more effective and personalized outcomes (the healthcare example) while also warning that boundary choice is subjective.

Before

Tight boundaries chosen for tractability and low cost.

After

Deliberately widened boundaries for problem classes where the excluded factors carry causal weight.

Consequence

Requires a per-problem boundary policy instead of one global default, since widening increases cost and complexity.

Source video

Tradeoffs and failure modes

4

Holistic boundary width vs. tractability

Benefit

Broadening the system boundary captures multi-factor dependencies and yields more effective, personalized interventions.

Cost or risk

Wider boundaries include more interacting variables, more stakeholders, and more nonlinearity, increasing analysis and resource cost.

Broadening the system boundary leads to more effective and personalized treatment plans.
Open source video
Source video

Boundary subjectivity as a bias channel

Benefit

Explicit boundary-setting makes problem framing and intervention design coherent and communicable.

Cost or risk

Because inclusion/exclusion is a judgment, unexamined assumptions and analyst bias silently shape conclusions and interventions.

Be mindful of subjectivity and assumptions when drawing system boundaries.
Open source video
Source video

Flexible human-centric methods vs. rigor

Benefit

Soft Systems Methodology handles messy situations with conflicting stakeholder values that rigid formal methods cannot represent.

Cost or risk

Flexibility trades away the falsifiability and repeatability that computable simulation methods provide, and depends on resolving contradictions across viewpoints.

In ambiguous situations with multiple stakeholders holding conflicting views.
Open source video
Source video

Nonlinear interaction as the source of complexity

Benefit

Treating complexity as non-linear interaction among many parts correctly locates causality in the couplings.

Cost or risk

Nonlinear couplings make outcomes hard to predict and small local changes can trigger widespread cascading consequences.

Complexity refers to non-linear interactions among many interconnected parts.
Open source video
Source video

Open questions

4

Can boundary judgments be encoded as an explicit, versioned, machine-checkable artifact (scope manifest) for an AI system, such that changes in scope trigger re-evaluation?

Why unresolved

The summary asserts boundaries are subjective and should be chosen mindfully, but provides no representation, review process, or change-control mechanism.

Research direction

Define a scope-manifest schema (included data, tools, signals, stakeholders; excluded items and rationale) and test whether scope diffs predict failure-class regressions in an eval suite.

Source video

If emergent behavior comes from interactions rather than components, what interaction-level metrics should replace or supplement component-level evaluation?

Why unresolved

The summary asserts the interaction locus of emergence and recommends improving interactions, but names no measurable property of an interaction.

Research direction

Prototype interaction-graph telemetry (handoffs, message schemas, retry/return paths) and test whether interaction-graph defects correlate with task failures better than per-component scores.

Source video

How should the choice among competing methodologies be operationalized as a decision procedure rather than a judgment call?

Why unresolved

The summary states selection should match problem complexity and available resources but gives no threshold, rubric, or cost model for making the match.

Research direction

Build a decision rubric keyed on problem character (computable dynamics vs value-contested ambiguity) and budget, then test inter-rater agreement and downstream outcome quality.

Source video

Does deliberately widening a system boundary improve outcomes monotonically, or does it introduce new failure modes past some width?

Why unresolved

The summary presents a strong case for boundary broadening in healthcare but never states a limit or cost function for it.

Research direction

Run a boundary-width sweep on a fixed task family, measuring outcome quality against context/complexity cost to locate the inflection point.

Source video

Key claims

7
factualVerification needed

Complexity consists of non-linear interactions among many interconnected parts.

Evidence

Complexity refers to non-linear interactions among many interconnected parts.

Question

Is non-linearity among parts a necessary condition for the systems-thinking definition of complexity, or only one of several accepted criteria?

Source video
causalVerification needed

Deciding what lies inside or outside a system is subjective and materially affects the resulting analysis.

Evidence

Deciding what is inside or outside a system is subjective and impacts analysis.

Question

Can two analysts given the same problem produce different boundaries that lead to measurably different conclusions and interventions?

Source video
causalVerification needed

System-level behaviors emerge from interactions among parts rather than from the parts individually.

Evidence

Behaviors arise from interactions among system parts rather than individual parts alone.

Question

What experimental setup isolates interaction effects from component effects in a multi-agent system when predicting a target outcome?

Source video
factualVerification needed

The Viable Systems Model is built on five key functions: goal setting, resource allocation, decision making, monitoring, and coordination.

Evidence

Goal setting - Resource allocation - Decision making - Monitoring - Coordination

Question

Is this five-function formulation an accurate representation of the Viable Systems Model as originally specified?

Source video
causalVerification needed

Soft Systems Methodology is applicable to messy, human-centric problems involving human values and differing stakeholder perspectives.

Evidence

Systematic approach to problem solving for complex, messy situations involving human values and differing stakeholder perspectives.

Question

Does SSM produce outcomes superior to formal quantitative methods on ambiguous, multi-stakeholder problems, and under what measurement?

Source video
factualVerification needed

Stafford Beer developed the Viable Systems Model (VSM).

Evidence

Stafford Beer developed the Viable Systems Model (VSM).

Question

Is Stafford Beer the correct originator of VSM, and what is the primary source?

Source video
factualVerification needed

Peter Checkland developed Soft Systems Methodology (SSM).

Evidence

Peter Checkland developed Soft Systems Methodology (SSM).

Question

Is Peter Checkland the correct originator of SSM, and what is the primary source?

Source video

Connections

4