Chapter 19 Exercises
How to use these exercises: Work through the parts in order. Part A builds recognition skills, Part B develops analysis, Part C applies concepts to your own domain, Part D requires synthesis across multiple ideas, Part E stretches into advanced territory, and Part M provides interleaved practice that mixes skills from all levels.
For self-study, aim to complete at least Parts A and B. For a course, your instructor will assign specific sections. For the Deep Dive path, do everything.
Part A: Pattern Recognition
These exercises develop the fundamental skill of recognizing iatrogenesis across domains.
A1. For each of the following interventions, identify (i) the intended effect, (ii) the actual iatrogenic effect, and (iii) whether the iatrogenic effect was a predictable consequence of the intervention or a genuinely surprising outcome.
a) A city installs speed cameras on a dangerous stretch of road. Drivers slow down on that stretch but speed up on adjacent streets, which were not designed for high-speed traffic and have more pedestrians.
b) A company implements mandatory password changes every 30 days. Employees begin writing passwords on sticky notes attached to their monitors.
c) A school district implements zero-tolerance discipline policies to reduce violence. Suspension rates soar, disproportionately affecting minority students, and suspended students fall behind academically, increasing dropout rates.
d) A government subsidizes corn production to ensure food security. Corn becomes so cheap that it is converted to high-fructose corn syrup, which is added to processed foods, contributing to an obesity epidemic.
e) A hospital implements a checklist to reduce surgical errors. Surgeons, relying on the checklist, become less attentive to non-checklist concerns, and the rate of unusual complications increases.
f) A social media platform adds content moderation to reduce harmful content. Moderators develop PTSD from exposure to the content they must review.
g) A central bank raises interest rates to fight inflation. The higher rates cause a recession, unemployment rises, and the government must increase social spending, worsening the fiscal deficit.
h) A country builds a seawall to protect a coastal city from flooding. The false sense of security encourages denser development behind the wall, so that when a storm exceeds the wall's design capacity, the damage is far greater than it would have been without the wall.
A2. Classify each of the following as primarily an example of (a) the intervention bias, (b) the McNamara Fallacy, (c) moral hazard, (d) the intervention spiral, or (e) via negativa. Some may involve more than one pattern. Explain your reasoning.
a) After a building collapses, the government imposes strict new building codes, which increase construction costs so much that builders cut corners in areas the codes do not cover, leading to different types of structural failure.
b) A teacher decides to stop grading homework and instead focuses on removing distracting technology from the classroom.
c) A hospital measures "readmission rates" as a quality metric. To reduce readmissions, the hospital keeps patients longer than necessary, increasing hospital-acquired infection risk and costs.
d) Deposit insurance encourages banks to take greater risks, since depositors have no reason to scrutinize bank safety.
e) A military operation to eliminate a terrorist cell kills several civilians; the civilian casualties are used as a recruiting tool by the terrorist organization, leading to a larger cell that requires a larger operation.
f) A physician whose patient has mild hypertension decides to recommend diet and exercise rather than prescribing medication with known side effects.
A3. For each of the following pairs, identify which intervention is more likely to produce iatrogenic harm and explain why, using the Intervention Calculus framework.
a) Intervention A: Prescribing antibiotics for a confirmed bacterial infection. Intervention B: Prescribing antibiotics for a viral upper respiratory infection "just in case."
b) Intervention A: Deploying a well-tested security patch during a scheduled maintenance window. Intervention B: Deploying an emergency patch to production during peak hours with minimal testing.
c) Intervention A: Conducting a prescribed burn in a forest with carefully monitored conditions and firebreaks. Intervention B: Continuing to suppress all fires in a forest that has not burned in sixty years.
d) Intervention A: A central bank adjusting interest rates by 0.25 percent in response to clear economic indicators. Intervention B: A central bank cutting interest rates to near zero in response to a mild recession.
A4. The chapter identifies five features that make iatrogenesis likely: (1) system complexity, (2) incomplete models, (3) asymmetric visibility of benefits and costs, (4) institutional bias toward action, and (5) delayed feedback. For each of the following systems, rate each feature on a scale of 1 (low) to 5 (high) and assess the system's overall vulnerability to iatrogenic harm.
a) A local bakery adjusting its bread recipe b) The global financial system c) A household adjusting its thermostat d) A national education reform program e) An individual choosing to take a daily vitamin supplement
A5. Identify three interventions in your daily life that might be iatrogenic -- actions you take that may be causing more harm than the problem they address. For each, describe the original problem, the intervention, the possible iatrogenic effect, and what a via negativa alternative might look like.
Part B: Analysis
These exercises require deeper analysis of iatrogenic patterns.
B1. The Iatrogenic Audit. Choose one of the following domains and conduct a systematic audit of iatrogenic interventions:
- Criminal justice
- Public education
- Urban transportation planning
- Environmental regulation
- Cybersecurity
- Healthcare administration
For your chosen domain:
a) Identify at least three interventions that are widely practiced in this domain.
b) For each intervention, trace both the intended effects and the unintended iatrogenic effects.
c) Assess whether the iatrogenic effects were predictable at the time the intervention was introduced.
d) Evaluate whether the intervention, on net, produces more benefit or more harm.
e) Propose a via negativa alternative to at least one of the interventions.
f) Explain why the iatrogenic intervention persists despite its harmful effects. What institutional, political, or psychological forces sustain it?
B2. The Intervention Calculus in Practice. A mid-sized city has a problem: traffic congestion is increasing commute times by an average of twenty minutes per day for 100,000 commuters. The city council is considering three interventions:
- Option A: Build a new highway bypass (cost: $2 billion, construction time: 5 years)
- Option B: Implement congestion pricing (charge drivers $5 to enter the city center during peak hours)
- Option C: Invest in public transit (new bus routes and a light rail line, cost: $1.5 billion, construction time: 7 years)
For each option, apply the Intervention Calculus:
a) What are the intended first-order effects?
b) What are the plausible second-order effects? (Hint for Option A: research "induced demand.")
c) What are the plausible third-order effects?
d) What is the visibility asymmetry? Which effects will be immediately visible, and which will be delayed or diffuse?
e) Does the intervention create moral hazard or perverse incentives?
f) Is there a via negativa alternative -- something that could be removed or stopped rather than added?
g) Based on your analysis, which option is least likely to produce iatrogenic harm? Defend your answer.
B3. Antibiotic Resistance: A Full Iatrogenic Analysis. Using the framework from Section 19.3 of the main chapter:
a) Map the complete causal chain from antibiotic development to antibiotic resistance, identifying each intervention point and its iatrogenic consequences.
b) Identify the actors whose individual rational behavior contributes to the collective iatrogenic outcome (patients who demand antibiotics, doctors who prescribe them, farmers who use them in livestock, pharmaceutical companies that market them).
c) Explain why the problem is a tragedy of the commons. Who bears the cost of each individual prescription? Who receives the benefit?
d) Evaluate three proposed solutions: (1) restricting antibiotic prescriptions, (2) banning antibiotics in animal agriculture, (3) investing in new antibiotic development. For each, assess whether the solution might itself be iatrogenic.
e) Is there a via negativa approach to antibiotic resistance? What harmful practice, if removed, would do the most to slow resistance?
B4. The Fire Paradox Quantified. Research the following data points for the western United States:
a) Average annual area burned by wildfire in the 1960s vs. the 2010s.
b) Average cost of wildfire suppression in the 1990s vs. the 2020s (adjusted for inflation).
c) Number of structures destroyed by wildfire per decade from the 1970s to the 2020s.
Using this data, assess the following claim from the chapter: "A century of fire suppression did not prevent catastrophic wildfire; it caused it." Does the data support this claim? What alternative explanations (climate change, population growth in the WUI, changes in forest management) might account for the trends?
Part C: Application to Your Own Domain
These exercises connect iatrogenesis to your area of expertise.
C1. Identify the most common intervention in your professional domain -- the "standard treatment" that practitioners apply to the most common problem. Then:
a) Describe the problem the intervention addresses.
b) List the known iatrogenic effects of the intervention.
c) Estimate what fraction of practitioners are aware of these iatrogenic effects.
d) Describe the institutional pressures that favor this intervention over alternatives (including the alternative of non-intervention).
e) Propose a via negativa approach that might address the underlying problem with less iatrogenic risk.
C2. Identify a case in your professional experience where an intervention made things worse. Analyze the case using the Intervention Calculus:
a) What was the original problem?
b) What was the intervention?
c) What were the iatrogenic effects?
d) Were the iatrogenic effects predictable? Were they predicted by anyone? Were those predictions heeded?
e) What corrective interventions were applied? Did they produce their own iatrogenic effects?
f) What would via negativa have looked like in this case?
C3. Design an "iatrogenic early warning system" for your domain. What signals would indicate that an intervention is producing iatrogenic harm? What metrics would you track? How would you distinguish between the normal costs of intervention and genuinely iatrogenic effects? Who would have the authority to halt or reverse an iatrogenic intervention?
Part D: Synthesis
These exercises require integrating ideas across multiple chapters.
D1. Iatrogenesis and Feedback Loops. Chapter 2 introduced positive (reinforcing) and negative (balancing) feedback loops.
a) Draw the feedback loop of the intervention spiral. Label each step: problem -> intervention -> iatrogenic effect -> new problem -> new intervention. Is this a positive or negative feedback loop? Why does it amplify rather than correct?
b) Design a negative feedback loop that would counteract the intervention spiral. What information would need to flow to which decision-makers to create a self-correcting dynamic?
c) The chapter argues that the intervention spiral continues until the intervener exhausts its resources. Is this the only way the spiral can end? Can you identify a case where the spiral was broken by a deliberate change in strategy rather than by exhaustion?
D2. Iatrogenesis and Goodhart's Law. Chapter 15 showed that metrics used as targets are corrupted by optimization pressure.
a) For three of the domains discussed in Chapter 19 (medicine, economics, foreign policy, software, ecology), identify the metric that functions as a Goodhart target, incentivizing intervention regardless of whether it helps.
b) For each domain, propose an alternative metric that would capture both the benefits and the iatrogenic costs of intervention. Why is this metric harder to implement than the existing one?
c) Is it possible to design a metric that is resistant to both Goodhart corruption and iatrogenic bias? Or do these two failure modes interact in ways that make them mutually reinforcing?
D3. Iatrogenesis and Redundancy. Chapter 17 argued that redundancy is not waste.
a) Identify three examples from Chapter 19 where the iatrogenic intervention involved stripping redundancy from a system (removing buffers, eliminating backup systems, consolidating suppliers, etc.).
b) For each example, explain how the removal of redundancy constituted iatrogenic harm.
c) The chapter argues that the drive to "fix" perceived inefficiency by cutting redundancy is itself a form of iatrogenesis. Construct a one-paragraph argument for this claim, drawing on examples from both chapters.
D4. Iatrogenesis and Annealing. Chapter 13 argued that controlled randomness helps systems escape local optima.
a) The fire paradox illustrates what happens when natural disturbance (fire) is suppressed. Identify two other examples from this chapter where the suppression of natural disturbance produced iatrogenic harm.
b) Explain how the annealing framework predicts the iatrogenic consequences of fire suppression. What is the "temperature" in the fire analogy? What happens when the temperature is reduced to zero?
c) Is there a general principle here? Does suppressing natural disturbance always produce iatrogenic consequences? Can you identify a counterexample -- a case where suppressing disturbance genuinely improved the system?
Part E: Advanced Challenges
These exercises push beyond the chapter's material into deeper or more speculative territory.
E1. Research Ivan Illich's concept of "social iatrogenesis" and "cultural iatrogenesis" from Medical Nemesis (1976). Illich argued that beyond direct medical harm (clinical iatrogenesis), the medical system harms society by medicalizing normal life experiences (social iatrogenesis) and by destroying people's ability to cope with pain, aging, and death without professional intervention (cultural iatrogenesis). Evaluate Illich's argument. Is the concept of cultural iatrogenesis applicable to other domains? Consider: has the professionalization of education produced "educational iatrogenesis"? Has the professionalization of food production produced "nutritional iatrogenesis"?
E2. The Intervention Calculus places the burden of proof on the intervener. But in practice, asymmetric urgency often makes this principle difficult to apply: the costs of inaction are immediate and visible (a patient is in pain, a building is on fire, an economy is contracting), while the costs of intervention are delayed and invisible. Design an institutional framework -- a set of rules, roles, and processes -- that would apply the Intervention Calculus in a real-world context of your choice. How would you balance the urgency of action against the risk of iatrogenic harm?
E3. The chapter argues that the intervention bias is partly driven by narrative bias: interveners tell stories in which they are heroes, and "I chose not to act" is not a heroic narrative. Research the concept of "negative expertise" -- the expertise of knowing when not to act. Is this concept recognized in any professional domain? How could educational and professional institutions cultivate negative expertise? Write a 500-word proposal.
E4. Taleb's concept of via negativa is fundamentally conservative: when in doubt, subtract rather than add. Critics argue that this principle, if applied too broadly, would prevent beneficial innovations (new medicines, new technologies, new policies) that carry uncertain risks but also enormous potential benefits. Evaluate this criticism. Is via negativa a general principle or a domain-specific one? Under what conditions should the principle be overridden?
E5. The chapter discusses antibiotic resistance as a form of iatrogenesis on an evolutionary timescale. But the evolution of resistance is not limited to antibiotics. Research one of the following and analyze it as iatrogenesis: (a) pesticide resistance in agricultural pests, (b) herbicide resistance in weeds, (c) the evolution of drug resistance in cancer cells during chemotherapy, (d) the evolution of resistance to genetic engineering in target organisms. In each case, identify the intervention, the iatrogenic evolutionary response, and the implications for the long-term viability of the intervention.
Part M: Mixed Practice (Interleaved Review)
These exercises mix concepts from Chapters 15-19 to build integrated understanding.
M1. A hospital system measures physician performance by the number of patients seen per hour (Goodhart's Law, Ch. 15). Physicians, under time pressure, prescribe medications for conditions that might resolve on their own (iatrogenesis, Ch. 19). The medications have side effects that require additional visits, increasing the physicians' patient counts and appearing to justify the metric (positive feedback loop, Ch. 2). Analyze this system using concepts from all three chapters. How would you redesign the metrics, the incentives, and the intervention protocols to break this cycle?
M2. A country's agricultural policy subsidizes monoculture farming for maximum efficiency (redundancy vs. efficiency, Ch. 17). The monoculture requires heavy pesticide use (iatrogenesis, Ch. 19). Pesticide use kills pollinators (second-order effect), requiring further intervention in the form of managed pollination services (intervention spiral). The subsidy is measured by total crop output per acre (Goodhart target, Ch. 15). Trace the complete iatrogenic cascade, identify the Goodhart target at each stage, and propose a via negativa alternative that addresses the root cause.
M3. A software company's security team measures its effectiveness by the number of patches deployed (Goodhart's Law, Ch. 15). Each patch is a potential source of regression (iatrogenesis, Ch. 19). The company's architecture has no redundancy -- a single codebase, a single deployment pipeline, a single production environment (redundancy deficit, Ch. 17). When a patch causes a regression, there is no fallback, and the regression must be patched with another patch (patch cascade). Design an alternative approach that addresses all three problems simultaneously.
M4. A financial regulator, responding to a banking crisis, imposes strict new capital requirements (intervention). The requirements reduce bank profitability, leading banks to shift risky activities to unregulated "shadow banking" entities (iatrogenic effect). The shadow banking sector grows until it poses a systemic risk that the regulator cannot monitor or control (unintended consequence). The regulator responds with new regulations on shadow banking (intervention spiral). Analyze this dynamic using the iatrogenesis framework and the Goodhart's Law framework. Is there a via negativa approach that would produce less iatrogenic harm?
M5. A national education system, concerned about poor student outcomes, implements standardized testing to measure school performance (Goodhart's Law, Ch. 15). Schools "teach to the test," reducing instruction in non-tested subjects (metric gaming). The narrowed curriculum reduces students' ability to think creatively and solve novel problems (iatrogenic effect, Ch. 19). The system responds by adding more tests for more subjects (intervention spiral). The proliferation of tests consumes more instruction time, further reducing actual learning (patch cascade). Meanwhile, the system eliminates "redundant" programs like art, music, and physical education to focus resources on tested subjects (redundancy stripping, Ch. 17). Design a comprehensive reform that addresses all three failure modes.