Second Order Effects
Second Order Effects: Every action produces a first-order effect (the intended outcome) and second-order effects (what happens because of the first-order effect). Most policy failures, product missteps, and investment mistakes are second-order failures. The discipline: before any consequential action, ask "and then what?"
What Is a Second Order Effect?β
Second order effects are the consequences of consequences. When you take an action, the immediate, direct result is the first-order effect. What happens next β as a result of that first-order effect β is the second-order effect. And what follows from that is the third-order effect.
Most human intuition operates at the first order: "If I do X, then Y will happen." This is often correct for the immediate consequence. Where intuition systematically fails is in anticipating what happens next β what Y then causes.
The concept is embedded in multiple disciplines under different names: second-order thinking (investing), second-best theory (economics), unintended consequences (sociology), blowback (international relations), and iatrogenics (medicine β harm caused by the treatment). All describe the same phenomenon.
The key insight from complexity theory: in interconnected systems, the second-order effects are often larger than the first-order effects, because the system reacts to and adapts to the first-order change. Ignoring this is the source of most well-intentioned failures.
How It Worksβ
Order of Effects Framework:
First Order: Direct, intended effect
Action: Lower drug prices through price controls
First order: Drugs become less expensive for consumers
Second Order: Effect of the first-order effect
Pharmaceutical companies reduce R&D investment in the controlled markets
(Less profitable β less incentive to innovate)
Third Order: Effect of the second-order effect
Fewer new drugs developed β Worse health outcomes over 20-year horizon
Practice: The "And then what?" chain
Action β Effect 1 β "And then what?" β Effect 2 β "And then what?" β Effect 3
Apply specifically to:
1. How will affected parties adapt their behavior?
2. What feedback loops will the first-order effect trigger?
3. What unintended incentives does this create?
4. What happens if many people do this simultaneously?
Three Real-World Examplesβ
Cobra Effect (Colonial India)β
The British colonial government offered a bounty for dead cobras to reduce the snake population in Delhi. First-order effect: people killed cobras for the bounty β supply of dead cobras increased. Second-order effect: entrepreneurs began breeding cobras specifically for the bounty. Third-order effect: when the government cancelled the program, breeders released their now-worthless cobras β the wild population was larger than before the intervention.
The program was designed by people who thought only at the first order: "Bounty β more dead snakes β fewer snakes." The second-order effect (incentive to breed snakes) was predictable β it follows directly from the incentive structure β but wasn't anticipated.
Uber and Taxi Supply in Citiesβ
Uber's entry into cities had the first-order effect of providing more car-ride supply and lower prices. The second-order effect: reduced usage of public transit (some bus and subway riders switched to Uber for cost and convenience). The third-order effect: reduced transit funding and ridership density, which degraded service quality, which reduced ridership further, creating a vicious cycle. Studies of cities where Uber and Lyft entered found significant increases in vehicle traffic β the opposite of the "ride-sharing reduces congestion" argument made at launch.
This doesn't mean Uber was net negative. The point is that second-order effects of sufficient magnitude can partially or completely reverse the first-order benefit.
The Invention of Air Conditioningβ
Air conditioning was invented to enable worker comfort and productivity in hot climates. First-order effect: buildings in hot climates became comfortable to work in year-round. Second-order effects: (1) enabled the economic development of the US Sun Belt (Phoenix, Houston, Miami would be far smaller without AC); (2) enabled the construction of sealed glass skyscrapers that require constant mechanical cooling; (3) created a massive energy consumption feedback loop that contributes to the heat that makes outdoor temperatures worse. These second and third-order effects on urban development, energy consumption, and climate dwarfed the original first-order effect in magnitude.
When to Use Itβ
β Use Second Order thinking when:
- Designing policies with significant human behavioral responses
- Making product decisions that will change user incentives or behavior
- Evaluating investments (how will the market react to the investment thesis if it's right?)
- Anticipating competitive responses to your strategic moves
- Any consequential decision in a complex, adaptive system
β Limit when:
- The system is genuinely simple and non-adaptive (physical processes with no feedback)
- Time constraints require a quick decision β in these cases, note the unresolved second-order risks and plan to revisit
| Pairs well with | Why |
|---|---|
| Feedback Loops | Second-order effects often flow through feedback mechanisms |
| Unintended Consequences | Second-order failures are one major category of unintended consequences |
| Incentive Theory | Most second-order failures involve behavioral responses to changed incentives |
| Pre-mortem | Pre-mortems surface second-order failure modes before commitment |
Common Misusesβ
Stopping at the second order. Truly complex decisions require at least third-order analysis. The second-order effect is often anticipated; the third often isn't.
Using second-order thinking to justify inaction. Every action has uncertain second-order effects; so does every inaction. The goal is not to avoid all action but to anticipate and mitigate the most significant second-order risks.
Treating second-order effects as unforeseeable. Many second-order effects are predictable before the fact β they follow directly from the incentives and feedback structures of the system. "We didn't know" is rarely accurate; "we didn't think" is more honest.
Related Modelsβ
- Unintended Consequences β the broader category; second-order failures are the most common type
- Cobra Effect β a specific pattern of second-order failure where intervention worsens the problem
- Feedback Loops β the structural mechanism behind most second-order effects
FAQβ
How far should second-order analysis extend?
As far as is practically useful β typically 2β3 orders for most decisions. Beyond that, uncertainty compounds quickly and the analysis becomes speculative. Focus on: (1) what are the most likely behavioral responses to the first-order effect, and (2) what feedback loops will the first-order effect trigger? These two questions capture most practically important second-order effects.
What makes second-order effects so reliably underestimated?
Two cognitive factors: first, temporal discounting (we weight near-term effects more than future ones, so first-order effects feel more real); second, WYSIATI (What You See Is All There Is) β we build our mental model from available information, and second-order effects require imagining a future state that doesn't exist yet. The combination makes first-order thinking the cognitive default.
Is second-order thinking the same as unintended consequences?
Related but not identical. All second-order effects are potential unintended consequences, but not all unintended consequences are second-order effects. Some consequences are simply unpredictable; some are first-order effects on parties not considered in the original analysis. Second-order thinking specifically traces the consequence chain: Effect A β causes Effect B β causes Effect C. Unintended consequences is the broader category of outcomes the decision-maker didn't plan for.
Apply with AIβ
π Trace second-order effects in MindMax β
Further Readingβ
- Howard Marks, The Most Important Thing (2011) β Chapter 4 on second-level thinking in investing.
- Robert Merton, "The Unanticipated Consequences of Purposive Social Action" (American Sociological Review, 1936) β The foundational sociological treatment.
- Nassim Nicholas Taleb, Antifragile (2012) β Chapter 5 on iatrogenics and second-order harm from interventions.
This page is part of the MindMax Mental Models Knowledge Base.