Systems Thinking for Leaders: Designing Solutions That Work

Main ideas, concepts, and lessons

• Systems thinking is necessary because real-world problems show “policy resistance.”

  • People propose well-researched solutions that work initially, but later the original problem returns (or worsens).
  • The cause is usually feedback effects and unintended consequences, not simply bad intentions or lack of effort.

• Move beyond slogans (“everything is connected”) to operational tools leaders can use.

  • The goal is to give leaders concrete methods they can apply even in short executive programs.

• Policy resistance is driven by a mismatch between mental models and reality.

  • Many organizations use an open-loop mental model: identify → analyze → choose optimal → implement → problem solved.
  • Real systems behave more like a closed-loop feedback world:
    • decisions change the system,
    • that creates new information,
    • which conditions subsequent decisions,
    • producing ongoing cycles.

• Causal diagramming helps avoid policy resistance.

  • A causal diagram represents:
    • a goal (desired state),
    • the current system state,
    • and the causal links created by decisions.
  • It helps leaders anticipate that actions create multiple effects, including reinforcing and undermining loops.

• “Side effects” are reframed as real effects caused by incomplete mental models.

  • There is no such thing as a true “side effect”; there are simply effects that weren’t correctly modeled or anticipated.
  • When leaders pull the system toward their goals, other actors (customers, employees, markets, communities) also act based on their own goals, often pulling things away—creating complexity and unintended outcomes.

• A key modeling technique: treat your model like an “iceberg.”

  • Above the waterline: what managers see (intended outcomes and balancing loops).
  • Below the waterline: what managers don’t directly observe (delayed, reinforcing, harmful feedbacks).
  • Leaders should challenge the boundaries of their model to discover what lies “under the waterline.”

---

Instructional / methodology content (detailed bullets)

1) How to use systems thinking to design solutions that work

• Clarify the goal

  • Example: gain market share (desired state vs. actual state).

• List possible decisions leaders might take to move toward the goal.

• Map causal relationships using a causal diagram

  • Identify:
    • direct intended effects,
    • indirect effects,
    • feedback loops (reinforcing vs. balancing),
    • and time delays.

• Reinterpret “side effects” as missing causal effects

  • Assume harmful outcomes come from:
    • incomplete mental models,
    • feedback pathways that weren’t represented,
    • or other actors’ responses.

• Include multiple stakeholder goals

  • Model how employees, customers, regulators, markets, suppliers, and communities react.
  • Account for how those reactions generate additional consequences.

• Use “iceberg thinking” to test the model boundary

  • Ask what’s above waterline (what you measure and manage directly).
  • Ask what might be below waterline (delayed outcomes, indirect drivers, reinforcing cycles).

• Look specifically for policy resistance patterns

  • Solutions that appear effective short-term but fail long-term.
  • Systems that keep reintroducing the original problem.

---

2) How to diagnose “policy resistance” in a healthcare policy example (prior approval, etc.)

• Identify the policies intended to reduce costs

  • Prior approval
  • Preferred drug lists (formularies)
  • Step therapy

• Identify the visible balancing loop (intended effect)

  • Cost pressure → tighten rules → lower unit drug/procedure costs → lower total costs.

• Identify the hidden reinforcing loops (unintended consequences)

  • Tight rules → worse timeliness/appropriateness of care → patients get sicker for longer.
  • Worse health → more follow-up visits/tests/treatments → total costs rise.
  • Higher costs → more rule tightening → further declines in care quality/timeliness.

• Expand to second-order effects

  • Administrative burden and denial/appeals
  • Emergency department visits, hospitalizations, readmissions
  • Litigation/malpractice risk
  • Patient plan switching (revenue impact) if outcomes deteriorate

---

3) How “management flight simulators” are used as a learning methodology

• Purpose

  • Help people build better mental models by letting them experiment safely.
  • Avoid the ineffectiveness of “telling people the right answer” with lectures/data alone.

• Core method

  • Place learners in a simulated system where they can try different strategies at low/no real-world risk.

• Simulation-based learning logic

  • If a strategy fails, “bankruptcy” in the simulation is not real harm.
  • If a strategy performs well, learners can test robustness under alternative scenarios.

• Instructor/learner affordances

  • Some simulators allow changing assumptions and exploring uncertainty.

• Example demonstrated: project management simulator

  • Learners act as project manager across phases (high-level design → detailed design).
  • They make trade-offs involving:
    • staffing levels vs. budget constraints,
    • overlapping phases vs. coordination risk,
    • scope changes,
    • management pressure for progress/quality,
    • responses to QA/engineering/HR/senior management feedback.
  • The simulation shows typical outcomes (late, over budget, low quality), demonstrating why simple planning models fail.

---

4) Qualitative systems thinking when you don’t build a full simulator

• Use causal mapping

  • Especially when formal simulation isn’t feasible.

• Use group modeling

  • Bring a diverse group into the room (“system in the room”).
  • Include key actors from multiple subsystems, possibly including external adversaries (e.g., community groups, advocacy groups).

• Expected benefits

  • Participants identify side-effect pathways and feedback loops that insiders miss.
  • The organization builds systems thinking capability:
    • listening before concluding,
    • collaboration,
    • focusing on learning and evidence rather than authority (“sage on the stage” vs. “guide on the side”).

---

Examples and cases mentioned

• Traffic congestion

  • Policy: build more roads
  • Short-term effect: more driving
  • Long-term: congestion persists or worsens due to induced demand, reduced transit use, and longer commuting distances.

• U.S. healthcare cost control via prior approval / formulary controls

  • Administrative overhead costs are described (prior authorization cost estimates).
  • Evidence cited:
    • an older study (1996) finding these restrictions can raise medical costs via unintended effects,
    • a newer meta-review (25 studies) finding associations with delays, worse health outcomes, hospitalizations, and lower survival rates.

• Project management

  • A poll indicates that projects often (sometimes/rarely) meet timelines, budgets, quality, and customer delight.
  • Simulator demo shows failure cascades driven by:
    • overlapping phases,
    • under-hiring due to budget,
    • management pressure,
    • rework and quality problems,
    • burnout,
    • scope creep (e.g., competitor-driven feature additions),
    • ultimately: over budget, late, and poor quality (very high defect rates).

• Dialysis anemia care improvement

  • System dynamics used to create a simulation model/“flight simulator” to improve clinical care pathways for kidney dialysis patients; described as saving lives and large costs.

• Other application domains

  • Energy policy and climate change
  • Public health
  • Drug development (biopharma)
  • Construction/chemicals/automotive/high-tech and other industries
  • Mention of En-ROADS climate policy simulator (developed with Climate Interactive).

---

What the speaker emphasizes about learning and change

• Data alone may not change entrenched behavior.

  • People resist when their existing mental model is reinforced by real-world experience.

• Simulators help change mental models

  • By letting learners experience how feedback and delays operate.
  • By enabling “safe practice” rather than direct instruction of an “optimal plan.”

• Facilitation matters

  • The instructor/facilitator’s role is to catalyze learning, not act as the “sage on the stage.”
  • “System thinking capability” includes humility, listening, and shared model building.

---

Speakers or sources featured (identified)

• Professor John Sterman (MIT)
• Molly (host/moderator introducing and managing chat/Q&A)
• Charlotte (host/moderator mentioned alongside Molly)
• Chris Sanford (asks a question about simulator recommendations)
• Mark Pellerine (asks a question about externalities/business continuity)
• Daniel Schneider (asks a question about short-term vs long-term vs ultra-short-term application)
• Eric Kebshell (asks about how simulation changes entrenched mental models)
• George Box (quoted: “all models are wrong … but some are useful”)
• Climate Interactive (collaborator mentioned for the En-ROADS climate simulator)
• Sir Thomas More (quoted from Utopia about remedies provoking complications)