Second Order Thinking

Second Order Thinking

When to Use

Use this skill when:

• User explicitly asks to apply second-order thinking, map ripple effects, or trace downstream consequences of a specific decision
• User says they want to think beyond the obvious impacts -- phrases like "what am I missing," "what happens next," or "what are the unintended consequences"
• User is weighing a major life or professional decision (career change, relocation, large purchase, organizational restructuring, policy change) and wants to stress-test it before committing
• User has already made a decision and wants to prepare for downstream effects -- second-order thinking applied retroactively functions as an early-warning system
• User is designing a system, policy, or incentive structure and wants to know how people will adapt their behavior in response (classic second-order territory)
• User describes a situation with competing stakeholders whose reactions will create follow-on effects -- decisions in organizations, markets, or relationships where other agents respond
• User asks about the difference between what they intend to happen and what might actually happen

Do NOT use when:

• User wants to imagine a specific failure scenario before a project begins -- use premortem-analysis, which focuses on the decision already failing rather than tracing all consequence chains
• User wants to compare multiple options with weighted criteria -- use weighted-decision-matrix, which scores trade-offs across options rather than mapping one decision's consequences forward
• User wants to build contingency plans for multiple possible futures -- use scenario-planning, which constructs parallel futures rather than one consequence chain
• User needs a formal record of a decision with rationale preserved -- use decision-journal, which documents context, not consequence maps
• User is asking about business-level strategic moves requiring stakeholder analysis, competitive dynamics, and market positioning -- use business strategy skills that handle multi-party environments explicitly
• User wants to enumerate all the ways a specific plan can fail -- premortem-analysis is the correct tool; second-order thinking maps all consequences (positive and negative), not just failure modes
• User needs to evaluate moral or ethical dimensions of a choice -- second-order thinking surfaces consequences, but ethical evaluation requires a separate ethical reasoning framework

Process

Step 1: Anchor the Decision Precisely

Before mapping anything, establish a clear, specific decision statement. Vague decisions produce useless consequence maps.

• Ask the user to state the decision as a single sentence: "I am choosing to [action]." If they cannot state it that way, help them sharpen it first.
• Distinguish between a reversible and an irreversible decision -- reversible decisions (can undo with moderate cost) warrant lighter analysis; irreversible decisions (quitting a job, selling a house, having a child, enacting a policy) demand full three-order mapping
• Establish the intended first-order outcome -- what the user believes will directly happen. This is the "official story" of the decision and the baseline against which hidden consequences are measured
• Identify all affected domains explicitly: Career, Finances, Relationships, Health/Energy, Time/Attention, Identity/Reputation, Legal/Regulatory, Market/Competitive, Organizational Culture, Environment. Not every decision touches all domains -- identify which ones are in scope
• Set the analysis time horizon: short (0-3 months), medium (3-18 months), long (18 months+). Some decisions have fast feedback loops; others take years. State the horizon and hold to it throughout analysis
• Identify the key stakeholders -- who else is affected by this decision and whose reactions will shape second-order effects? Behavioral responses from other people are the most common source of non-obvious consequences

Step 2: Map First-Order Effects With Discipline

First-order effects are the direct, immediate, intended consequences. They are what most people stop at.

• List 3-6 first-order effects. More than 6 suggests the decision has not been scoped tightly enough or multiple decisions are being conflated
• Apply the domain checklist: does this decision have a first-order effect on finances? Career? Relationships? Time budget? Health? Identity? Scan each domain before assuming no effect
• Assign polarity (+/-/neutral) and estimated timeframe to every effect. Neutral is acceptable at first order but should prompt scrutiny -- if an effect is truly neutral, question whether it belongs in the map
• Distinguish intended effects (what the decision is designed to achieve) from unintended first-order effects (direct consequences that were not the goal but are immediate). Both belong in the map
• Weight by magnitude, not just polarity. A first-order effect that is negative but small does not deserve the same analytical depth as a first-order effect that is negative and large. Use a rough magnitude scale: Minor (barely noticeable), Moderate (noticeable impact on daily life or operations), Major (significantly changes a domain), Transformative (restructures how you operate in that domain)
• Challenge the user's stated intended outcome. Ask: "Is this truly what will happen, or is this what you hope will happen?" First-order effects can themselves be optimistic assumptions

Step 3: Generate Second-Order Effects Using Structured Prompts

Second-order effects are the consequences of the consequences. They are where most value and most danger hide. For each first-order effect, apply these prompts systematically:

• Behavioral adaptation prompt: "How will other people change their behavior in response to this first-order effect?" Human behavioral response is the dominant engine of second-order effects -- incentives change, norms shift, relationships adjust
• Resource reallocation prompt: "What does this first-order effect consume or free up? Time, money, attention, energy, political capital?" Resources redirected by first-order effects create second-order consequences in every domain that was drawing on those resources
• Perception and signaling prompt: "What does this first-order effect signal to others? How will it change how you are perceived, or how you perceive yourself?" Reputation and identity shifts are systematically underestimated
• Dependency and coupling prompt: "What systems, plans, or relationships depended on the pre-decision state that will now be disrupted?" Second-order effects often emerge from broken dependencies rather than the decision itself
• Compounding and accumulation prompt: "If this first-order effect continues or accumulates over 12 months, what does the cumulative state look like?" Many second-order effects are just first-order effects running forward in time
• Each first-order effect should generate 1-3 second-order effects. If you cannot find a second-order effect for a first-order item, push harder with the behavioral adaptation and resource reallocation prompts -- they almost never fail
• Cross-domain effects are the prize: a decision that has a career first-order effect very often has financial, relational, and health second-order effects. Trace across domain lines explicitly

Step 4: Generate Third-Order Effects for High-Magnitude Chains

Third-order effects require selectivity. Apply full third-order analysis only to second-order effects rated Moderate or higher in magnitude.

• Apply the same structured prompts from Step 3, but now directed at the second-order effect as the new input
• Third-order effects are inherently more speculative. Calibrate confidence explicitly: High confidence (structurally forced by the earlier effects), Medium confidence (likely given typical human and system behavior), Low confidence (possible but dependent on contingencies)
• Watch for tipping points and phase transitions at third order. These are the most valuable findings: a gradual negative second-order effect that, by third order, has crossed a threshold into a qualitative change. Examples: a financial strain becoming insolvency, a team dissatisfaction becoming attrition, a habit change becoming a new identity
• Watch for convergence: when multiple independent second-order chains produce the same third-order effect, that effect has much higher probability and severity than if only one chain leads there. Flag convergent third-order effects explicitly
• Track third-order effects that loop back to affect the original decision domain -- these are feedback loops in formation
• Limit third-order chains to the 3-5 most impactful paths. Do not trace every second-order effect to third order -- depth on important chains beats exhaustive breadth

Step 5: Identify Feedback Loops

Feedback loops are among the most powerful and most overlooked findings of second-order analysis. A feedback loop is a consequence chain that circles back to reinforce or suppress the original effect.

• Virtuous loops (positive reinforcement): an initial positive effect generates second-order effects that amplify the original positive effect. Example: improved performance leads to recognition, which leads to more responsibility, which builds skills, which improves performance further. These are often the real compounding value of a decision
• Vicious loops (negative reinforcement): an initial negative effect generates second-order effects that amplify the original negative effect. Example: budget pressure leads to cutting staff, which increases workload on remaining staff, which increases attrition, which increases budget pressure. These are often the hidden catastrophes of decisions
• Balancing loops: effects that dampen themselves over time. Example: a pay raise reduces financial stress, which reduces the urgency to seek additional income, which stabilizes spending. Balancing loops are why some effects matter less in the long run than they appear to at first order
• To identify loops, look for any effect in your map where the "and then what?" answer is "the original decision condition" or "the starting first-order effect." That circularity is your loop
• Every vicious loop found should be immediately flagged as a high-severity risk regardless of its order of origin

Step 6: Identify Irreversibility and Optionality

Not all consequences are equal -- the most important distinction is reversibility.

• Irreversible negative effects deserve outsized weight in the analysis. These are consequences you cannot undo regardless of subsequent action: lost compound growth years in a retirement account, a reputation shift in a small professional community, a dissolved long-term relationship, a health consequence with permanent impact, a legal record
• Irreversible positive effects (option-destroying commitments) also matter -- some positive consequences lock in value but close off other options permanently. Having a child is a positive irreversible effect that eliminates certain freedoms
• Optionality effects are second or third-order consequences that expand or contract your future choices. Decisions that preserve optionality are worth more under uncertainty than their first-order value suggests. Decisions that destroy optionality cost more than their first-order value suggests
• Apply the "what does this foreclose?" test to every major second and third-order effect: does this effect close doors that currently exist? Which doors?
• The Nassim Taleb heuristic applies here: "if in doubt, do not" scales with irreversibility. For effects that are both irreversible and high-magnitude, the asymmetry of the downside justifies a risk premium even when the expected value is positive

Step 7: Synthesize the Assessment and Produce the Output

The analysis is only complete when it produces an actionable synthesis, not just a populated table.

• Compute a directional balance at each order: are the effects at first order net positive or net negative? Does the picture improve or worsen as you move to second and third order? The trajectory pattern matters: decisions that look bad at first order but improve at second and third are classic "short-term pain, long-term gain" structures -- they should be handled differently than decisions that look good at first order but worsen at deeper orders
• Identify the single most important non-obvious finding -- the one insight that the user would not have reached without this analysis, that most significantly changes how they should think about the decision
• Produce a concrete recommendation: proceed / proceed with specific mitigations / reconsider. The recommendation must reference the specific high-magnitude hidden risks and opportunities identified in the analysis
• If mitigations are recommended, be specific: what exact action reduces what specific risk, and at what order does that mitigation intervene?

Output Format

## Second-Order Thinking: [Decision Title]

### Decision Profile
- **Choice:** [Single-sentence statement of the decision: "I am choosing to [action]"]
- **Decision type:** [Reversible / Partially reversible / Irreversible]
- **Intended first-order outcome:** [What the user believes will directly result]
- **Domains in scope:** [List all domains identified as affected]
- **Analysis horizon:** [Time period -- e.g., "0-24 months"]
- **Key stakeholders whose reactions matter:** [People or groups whose behavioral responses generate second-order effects]

---

### Consequence Map

#### First-Order Effects (Direct, Immediate)
| ID | Effect | Domain | +/- | Magnitude | Timeframe |
|----|--------|--------|-----|-----------|-----------|
| 1A | [direct effect] | [domain] | [+/-/N] | [Minor/Moderate/Major/Transform.] | [when] |
| 1B | [direct effect] | [domain] | [+/-/N] | [Minor/Moderate/Major/Transform.] | [when] |
| 1C | [direct effect] | [domain] | [+/-/N] | [Minor/Moderate/Major/Transform.] | [when] |
| 1D | [direct effect] | [domain] | [+/-/N] | [Minor/Moderate/Major/Transform.] | [when] |

#### Second-Order Effects (Consequences of the Consequences)
| Source | ID | Effect | Domain | +/- | Magnitude | Timeframe | Confidence |
|--------|----|--------|--------|-----|-----------|-----------|------------|
| 1A --> | 2A | [consequence of 1A] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |
| 1A --> | 2B | [consequence of 1A] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |
| 1B --> | 2C | [consequence of 1B] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |
| 1C --> | 2D | [consequence of 1C] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |
| 1D --> | 2E | [consequence of 1D] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |

#### Third-Order Effects (Deep Ripples -- Selected High-Magnitude Chains)
| Source | ID | Effect | Domain | +/- | Magnitude | Timeframe | Confidence |
|--------|----|--------|--------|-----|-----------|-----------|------------|
| 2A --> | 3A | [consequence of 2A] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |
| 2C --> | 3B | [consequence of 2C] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |
| 2D --> | 3C | [consequence of 2D] | [domain] | [+/-] | [scale] | [when] | [H/M/L] |

---

### Consequence Chain Visualization

[Decision] | +--> 1A: [first-order] (+/-) | +--> 2A: [second-order] (+/-) --> 3A: [third-order] (+/-) | +--> 2B: [second-order] (+/-) | +--> 1B: [first-order] (+/-) | +--> 2C: [second-order] (+/-) --> 3B: [third-order] (+/-) | +--> 1C: [first-order] (+/-) | +--> 2D: [second-order] (+/-) --> 3C: [third-order] (+/-) | +--> 1D: [first-order] (+/-) +--> 2E: [second-order] (+/-)

---

### Non-Obvious Findings

#### Hidden Risks (Negative Effects Not Visible at First Order)
| # | Risk | Order | Source Chain | Why Easy to Miss | Severity | Reversible? |
|---|------|-------|-------------|-----------------|----------|-------------|
| 1 | [risk description] | [2nd/3rd] | [1X --> 2X --> 3X] | [cognitive reason it is missed] | [Low/Med/High/Critical] | [Yes/No/Partial] |

#### Hidden Opportunities (Positive Effects Not Visible at First Order)
| # | Opportunity | Order | Source Chain | Why Easy to Miss | Value | Time to Realize |
|---|-------------|-------|-------------|-----------------|-------|----------------|
| 1 | [opportunity description] | [2nd/3rd] | [1X --> 2X --> 3X] | [why it is non-obvious] | [Low/Med/High] | [timeframe] |

#### Feedback Loops
| # | Loop Description | Type | Triggered By | Stakes |
|---|-----------------|------|--------------|--------|
| 1 | [full loop description with arrow notation] | [Virtuous/Vicious/Balancing] | [initiating first-order effect] | [what is at stake if loop activates] |

#### Irreversible and Optionality Effects
| # | Effect | Order | Irreversibility Reason | Options Foreclosed |
|---|--------|-------|----------------------|--------------------|
| 1 | [effect] | [order] | [structural reason it cannot be undone] | [what future choices this eliminates] |

#### Convergent Effects (Multiple Chains Pointing to the Same Outcome)
| Effect | Chains That Lead Here | Combined Probability | Significance |
|--------|-----------------------|----------------------|-------------|
| [effect] | [list of chains] | [Higher/Medium] | [why convergence matters] |

---

### Assessment

| Dimension | Rating | Explanation |
|-----------|--------|-------------|
| First-order balance | [Net +/Net -/Mixed] | [brief reason] |
| Second-order balance | [Net +/Net -/Mixed] | [brief reason] |
| Third-order balance | [Net +/Net -/Mixed] | [brief reason] |
| Overall trajectory | [Improving / Stable / Worsening / V-shaped / Inverse-V] | [pattern description] |
| Irreversibility exposure | [Low/Medium/High] | [which irreversible effects drive this rating] |
| Feedback loop risk | [Low/Medium/High] | [which loops are active and their stakes] |

**Key Insight:**
[Single most important non-obvious finding -- the thing the user would not have seen without this analysis. Be specific. Name the exact effect, the chain it comes from, and why it matters.]

**Recommendation:** [Proceed / Proceed with specific mitigations / Reconsider]

**If mitigations are required:**
| Mitigation | Addresses Risk/Loop | Intervenes at Order | Specific Action |
|------------|--------------------|--------------------|----------------|
| [mitigation name] | [which risk or loop] | [1st/2nd/3rd] | [concrete action to take] |

Rules

1. Never skip the domain scan. Before finalizing first-order effects, run through every domain in the checklist (Career, Finances, Relationships, Health/Energy, Time/Attention, Identity/Reputation, Legal/Regulatory, Market/Competitive, Organizational Culture, Environment). A decision that appears to affect only one domain almost always has second-order effects in two or three others. Missing a domain at first order means missing entire branches of the consequence tree.

2. Assign magnitude, not just polarity. Positive and negative labels without magnitude create false equivalence. A minor negative second-order effect does not cancel a major positive first-order effect. Use the four-level scale (Minor / Moderate / Major / Transformative) for every effect in the map and let magnitude drive which chains receive third-order analysis.

3. Apply the behavioral response test to every first-order effect. The most common generator of non-obvious consequences is other people changing their behavior in response to your decision. For every first-order effect, ask: "Who else is affected by this, and how will they respond?" Incentive structures change, relationships adjust, institutions adapt. Missing behavioral responses produces systematically incomplete maps.

4. Trace second-order effects across domain lines. Second-order effects that stay within the same domain as their first-order source are less valuable to surface than effects that cross domains. A career first-order effect that creates a financial second-order effect, which creates a relationship third-order effect, is a classic non-obvious chain. Staying within one domain at second and third order is usually a sign of incomplete analysis.

5. Flag all vicious loops as Critical regardless of order. A vicious feedback loop -- a negative effect that amplifies itself -- is categorically more dangerous than a standalone negative effect of equal magnitude, because it compounds. Vicious loops identified at second order must be highlighted in the assessment even if the individual effects are rated Moderate.

6. Distinguish between confidence levels at third order. Third-order effects are structurally more speculative than first or second-order effects. Label each third-order effect with High, Medium, or Low confidence based on how structurally forced the chain is. Do not present Low-confidence third-order effects with the same certainty as High-confidence second-order effects. Calibrated uncertainty is part of the output's value.

7. Apply the convergence test. Before finalizing the assessment, check whether any effect appears at the end of two or more independent chains. Convergent outcomes have multiplicatively higher probability and severity than outcomes reached by only one path. If three independent chains all lead to "financial strain," the financial strain outcome should be rated Critical even if each individual chain only produces Moderate severity.

8. Irreversible effects override expected value calculations. A second-order effect that is irreversible and negative should factor into the recommendation even if the overall expected value of the decision is positive. The asymmetry between reversible and irreversible consequences is not captured by simple +/- accounting. Flag irreversible effects in the assessment explicitly and note what specific action could prevent or mitigate them.

9. Do not let the user's emotional investment soften the analysis. When a user has clearly made up their mind or is excited about a decision, the pull is to validate and minimize risks. The entire value of second-order thinking is surfacing what enthusiasm suppresses. Apply equal analytical pressure to decisions the user is eager to make as to decisions they are reluctant about.

10. The assessment must state the trajectory direction explicitly. The pattern of how the consequence balance changes from first to second to third order is as important as the balance at any single order. Name the pattern: Improving (gets better deeper), Worsening (gets worse deeper), V-shaped (worsens then recovers), Inverse-V (improves then worsens), or Stable (consistent across orders). The trajectory pattern determines the most important timing and mitigation recommendations.

Edge Cases

The decision appears to have only positive consequences at all orders. This is almost always a sign of incomplete analysis, not a genuinely consequence-free decision. Apply the "what is the price of this benefit?" test to every positive first-order effect. Every resource gain implies a trade-off. More money means more time spent earning it, or changed relationship dynamics, or shifted identity. More health means redirected time and attention. If the analysis still shows only positive effects after pushing, apply the behavioral adaptation test: "Who loses something as a result of my gain, and how might they respond?" The zero-sum dimension of many decisions hides behind the winner's framing.

The user's decision has more than six first-order effects. More than six first-order effects typically means either the decision is compound (two or three separate decisions being treated as one) or the scope is too broad. Help the user decompose the decision before mapping. If decomposition is refused, apply triage: rank all first-order effects by magnitude and trace only the top four deeply. State explicitly which first-order effects were excluded from second and third-order analysis and why. Breadth at first order produces shallow maps; depth on high-magnitude chains produces insight.

Third-order effects feel too speculative to include. Acknowledge reduced confidence explicitly in the map using the confidence column. Reframe how you introduce third-order effects: instead of "this will happen," use "watch for early signs of this." The value of speculative third-order analysis is not prediction -- it is building a monitoring checklist. A third-order effect labeled Low confidence is still valuable if it identifies a trigger event the user can watch for (e.g., "if you notice attrition exceeding 15%, the vicious loop at 3C is activating").

The decision has already been made and cannot be reversed. Redirect the analysis from decision support to consequence management. Map the consequence tree from the current state. For every hidden risk identified, convert it into a monitoring indicator: what early signal would tell the user the negative chain is activating? For every hidden opportunity, convert it into a proactive action: what can the user do now to capture the positive second or third-order effect rather than waiting for it to materialize passively? Irreversibility acknowledged, the output becomes an operational guide rather than a go/no-go recommendation.

The domain is a policy or organizational decision affecting many stakeholders simultaneously. Single-person decisions have one actor whose behavior changes. Policy and organizational decisions affect populations of actors who respond heterogeneously -- some comply, some resist, some exploit new loopholes, some exit. At second order, apply the stakeholder response matrix: for each stakeholder group, what is their most likely behavioral response to the first-order effect? Each distinct response pattern generates its own second-order branch. The aggregate of all stakeholder responses constitutes the true second-order effect. This is why well-intentioned policies routinely produce perverse outcomes -- the behavioral adaptation dimension of second-order effects is never uniformly positive even when every first-order effect is positive.

The user cannot generate any second-order effects independently. Use domain-crossing prompts in sequence until effects emerge: "How does this change your financial position in 12 months?" / "Who else is affected by this change, and what will they do differently?" / "What does this consume that you were using for something else?" / "What does this signal about you to people who matter to you?" / "What would have to stop in your life for this to work?" / "What assumption your current plans depend on does this break?" These six prompts almost never all fail simultaneously. If the user still cannot generate second-order effects, work through the domain checklist systematically with the user, proposing candidate effects for them to accept or reject.

The decision involves deep uncertainty about which first-order effects will actually occur. If the first-order effects themselves are uncertain (e.g., "I might get the promotion, or I might not"), do not attempt to map a single consequence tree. Instead, create two parallel first-order scenarios (the expected outcome and a realistic downside alternative) and map each one separately. Surface which second-order effects appear in both trees -- those are robust consequences that occur regardless of which first-order scenario materializes. Robust consequences deserve more weight in the assessment than consequences that depend on a specific uncertain first-order outcome.

The analysis reveals that all paths lead to the same bad outcome. If multiple independent chains converge on the same negative third-order effect with High confidence, and no mitigation intervenes at an earlier order, the assessment should state this directly: "The analysis finds that multiple independent chains converge on [effect]. This outcome appears structurally likely regardless of how intermediate steps unfold." Do not soften convergent catastrophic findings to protect the user's enthusiasm. This is when second-order thinking delivers its maximum value. Pair the finding with the specific earliest-order intervention point where mitigation is still possible.

Example

Input: "I run a mid-sized SaaS company (about 120 employees) and we're considering eliminating our annual performance review process entirely. We've been reading about companies going review-free and the team seems to want it. Help me think through the consequences beyond the obvious."

Output:

Second-Order Thinking: Eliminating Annual Performance Reviews

Decision Profile

• Choice: I am choosing to eliminate the formal annual performance review process company-wide
• Decision type: Partially reversible (can reinstate, but cultural expectations once set are difficult to reset)
• Intended first-order outcome: Reduced bureaucratic burden, improved employee morale, and shift to more organic continuous feedback
• Domains in scope: Organizational Culture, Career Development, Finances/Compensation, Management Operations, Legal/Regulatory, Retention/Attrition
• Analysis horizon: 0-24 months
• Key stakeholders whose reactions matter: Individual contributors (IC), managers, high performers, low performers, HR team, legal counsel, investors/board

Consequence Map

First-Order Effects (Direct, Immediate)

Second-Order Effects (Consequences of the Consequences)

Third-Order Effects (Deep Ripples -- Selected High-Magnitude Chains)

Consequence Chain Visualization

[Eliminate annual performance reviews]
  |
  +--> 1A: Time freed for managers (+, Moderate)
  |       +--> 2A: Time not reinvested in coaching; disappears (-, Moderate)
  |
  +--> 1B: Performance documentation eliminated (-, Major)
  |       +--> 2B: Legal exposure on terminations (-, Major) --> 3A: Wrongful termination settlement + norm erosion (-, Transformative)
  |       +--> 2C: High performer achievements undocumented (-, Moderate) --> 3E: Asymmetric incentive to leave (-, Moderate)
  |
  +--> 1C: Cultural signal of trust (+, Moderate)
  |       +--> 2D: False impression of universal support (-, Moderate)
  |
  +--> 1D: Compensation/performance link severed (-, Major)
  |       +--> 2E: Compensation driven by recency bias (-, Major) --> 3B: High-performer attrition (-, Major)
  |       +--> 2F: Pay equity legal risk increases (-, Major)
  |
  +--> 1E: Manager feedback requirement removed (Neutral, Moderate)
          +--> 2G: Poor-feedback managers stop altogether (-, Major) --> 3C: IC disengagement/drift (-, Major)
          |                                                           --> 3D: Manager avoidance of hard conversations (-, Major)
          +--> 2H: Strong-feedback managers unaffected (Neutral, Minor)

Non-Obvious Findings

Hidden Risks (Negative Effects Not Visible at First Order)

Hidden Opportunities (Positive Effects Not Visible at First Order)

Feedback Loops

Irreversible and Optionality Effects

…(truncated)