Systems Theory
Policy Resistance: Why Systems Resist Change
1970s. Forrester models Boston's housing policy: subsidies for cheap housing destroy the neighborhood over 15 years. The US spends $1 trillion on the War on Drugs over 50 years. The result? More drugs, stronger cartels, overflowing prisons. Why do good intentions lead to the opposite results? Because systems **resist** attempts to change them. And the harder the push, the stronger the resistance.
- **War on Drugs**: Bans → prices ↑ → cartels grow rich → production ↑
- **Prohibition (1920s)**: Alcohol ban → mafia, bootlegging, violence
- **Soviet quota**: Tonnage plan → enormous nails, shortage of small ones
- **Speed bumps**: Speed reduction → detours → more routes → more accidents
- **IT metrics**: Lines of code → bloated code, commits → micro-commits
What Is Policy Resistance?
1970s. Jay Forrester models Boston's housing policy. The authorities subsidize cheap housing - and 15 years later the neighborhood has deteriorated worse than without subsidies. The US declares a War on Drugs. Prisons overflow, enforcement budgets keep growing. The result? Drug use went **up**, and prices rose. Why do good intentions lead to the opposite results?
**Policy Resistance** - the phenomenon where a system actively counteracts attempts to change it. The harder the push, the harder it pushes back.
**Key idea**: A system has multiple actors with different goals. When one pushes in their direction, the others push back. The system gets stuck in a **frozen equilibrium**.
Policy resistance is not malicious intent. It is a natural reaction of a system where each actor protects their own interests. The problem is not in the people, but in the **structure of conflicting goals**.
Why might the War on Drugs worsen the problem instead of solving it?
Mechanisms of Resistance
Policy resistance works through three main mechanisms: **balancing loops**, **delays**, and **bounded rationality**.
**Balancing Loop** - feedback that returns the system toward a goal. When each actor has their own balancing loop, they counteract each other.
**Mechanism 2: Delays hide consequences**
An intervention produces a quick effect, but the side effects appear later. By then, the intervention has already been intensified.
**Mechanism 3: Bounded Rationality**
Each actor sees only their part of the system and optimizes locally, generating global chaos.
| Actor | What they see | Local optimization | Global effect |
|---|---|---|---|
| Police | Arrest statistics | More arrests = success | Overcrowded prisons, criminalization |
| Doctors | Patient pain | Prescribe opioids | Addiction epidemic |
| Factory (USSR) | Tonnage quota | Make heavy chandeliers | Shortage of lightweight goods |
| Drivers | Travel time | Detour around traffic | Traffic jam on the detour road |
Delays make resistance invisible. By the time consequences appear, the cause is forgotten and the intervention has been intensified. This creates **escalation of intervention**.
A city installs speed bumps to reduce speed. Six months later, accident rates are up. Why?
Cobra Effect: Perverse Incentives
British India, 19th century. Delhi is overrun with venomous cobras. The authorities offer a bounty for every dead snake. What happens? People start **breeding cobras** for profit. When the authorities cancel the program, the breeders release the snakes. There are now **more** cobras than before the program.
**Cobra Effect** - when an attempt to solve a problem creates a perverse incentive that makes it worse. Named after the cobra story.
**Classic examples of perverse incentives:**
| Policy | Goal | Perverse incentive | Result |
|---|---|---|---|
| Bounty for dead cobras | ↓ Snakes | Breed cobras to sell | ↑ Snakes |
| Nail quota by weight (USSR) | ↑ Production | Make huge nails | Shortage of small nails |
| Bounty for rat tails (Vietnam) | ↓ Rats | Cut tails off, release rats | ↑ Tailless rats |
| Fine for being late to daycare | ↑ Punctuality | "I pay, so I can be late" | ↑ Late pickups |
| Code metrics (lines/commits) | ↑ Productivity | Write useless code | ↓ Quality |
**Why does this happen?**
- **Metric ≠ Goal** - we measure the wrong thing
- **Goodhart's Law** - "When a measure becomes a target, it ceases to be a good measure"
- **Rational adaptation** - people optimize for incentives, not intentions
- **Invisible consequences** - complex systems always find workarounds
The Cobra Effect is especially powerful when **the incentive is stronger than the problem**. The cobra bounty exceeds the harm they cause → breeding becomes a business.
**Modern examples:**
A company pays developers per line of code. What happens?
How to Overcome Resistance
If pushing harder doesn't work, what should be done? Three strategies: **find a consensus point**, **change the structure**, or **work with the system, not against it**.
**Consensus point** - a goal that all actors support. Instead of a conflict of goals, create a shared goal.
**Strategy 1: Find the consensus point**
| Problem | Goal conflict | Consensus point | New policy |
|---|---|---|---|
| Drugs | Prohibition vs Freedom | Public health | Legalization + treatment + education |
| Speed bumps | Speed vs Time | Children's safety | Redesign streets: narrow, winding (calming design) |
| Nail production | Quota vs Quality | Happy customers | Pay for nails sold, not produced |
| Daycare late pickups | Fine vs Freedom | Children's comfort | Fine + explanation of harm to the child |
**Strategy 2: Change the structure of the system**
Instead of fighting symptoms - redesign the system so that the desired behavior becomes beneficial.
**Strategy 3: Small experiments instead of large interventions**
Large interventions generate large resistance. Small pilots allow learning and adaptation.
- **Pilot zone** - test in one district/department
- **Reversibility** - can be undone if it doesn't work
- **Iterability** - adjust based on feedback
- **Actor participation** - involve everyone in the design of the solution
Overcoming resistance **does not mean defeating other actors**. It means finding a solution where everyone wins, or at least doesn't lose critically.
A team resists a new process. A manager can: A) Order it implemented, B) Find the problem that the process solves for the team. What will work?
Working With the System, Not Against It
The main lesson of policy resistance: **systems are smarter than us**. They find ways to circumvent any intervention. Instead of fighting - understand the system and work with its nature.
**Aikido principle**: Use the force of the system rather than opposing it. If the system resists - the push is in the wrong direction.
**How to work with the system:**
**Examples of working with the system:**
| Problem | Against the system | With the system | Result |
|---|---|---|---|
| Overeating | Diet, willpower | Smaller plates, no junk food at home | Environment helps, not resists |
| Procrastination | "Just do it!" | Pomodoro technique, remove distractions | Structure supports the habit |
| Traffic jams | More lanes | Subway, bike lanes, remote work | Alternatives instead of fighting |
| Staff turnover | Raise salaries | Understand why they leave → change culture | Addressing cause, not symptom |
**Checklist before an intervention:**
- **Who are all the actors?** Whose interests are affected?
- **Which balancing loops?** Who will push back?
- **What delays?** When will consequences appear?
- **What perverse incentives?** How can the metric be gamed?
- **Is there a shared goal?** Or only conflict?
- **Can we start small?** Is the solution reversible?
If an intervention requires constant escalation (more money, more control, more pressure) - the system has captured the effort. **Stop and find a consensus point**.
**Donella Meadows**: "You can't make a system do what it doesn't want to do. You can understand what it wants and work with that."
If a policy isn't working, you just need to push harder
The harder the push, the stronger the resistance - this is the Escalation archetype
Policy resistance is a structural problem, not a lack of effort. More pressure intensifies the conflict of balancing loops, generating escalation. The right solution is to change the structure or find a consensus point where all actors support the change.
A sign that the work is going **against the system**, not with it:
Key Ideas
- **Policy Resistance** - a system resists change through conflicting goals among actors
- **Balancing loops** - each actor pulls toward their own goal → frozen equilibrium
- **Delays** - consequences appear later → intervention is intensified → things get worse
- **Cobra Effect** - perverse incentives when metric ≠ goal
- **Consensus point** - a shared goal for all actors instead of conflict
- **Work with the system** - don't push, find the leverage point and change the structure
Connections to Other Concepts
Policy Resistance explains why the Fixes that Fail and Escalation archetypes are so common. Leverage Points show where the system can be changed without resistance.
- System Archetypes — Fixes that Fail and Escalation are manifestations of resistance
- Leverage Points — Find the point of application of force instead of pushing
- Delays — Delays hide the consequences of intervention
Вопросы для размышления
- Where is the pressure currently increasing, but the resistance only grows?
- Which metrics in life or work can be gamed? What does that say about the goals?
- Recall a situation where a solution met resistance. Who were the actors? What were their goals?
- Is there a conflict of balancing loops in the system? Can a shared goal be found for everyone?
- Where is it possible to work **with the system** instead of against it? What leverage point has been overlooked?
Связанные уроки
- st-01-feedback-loops — understand feedback loops before analyzing policy
- st-04-leverage — leverage points as the answer to resistance
- st-14-delays — delays hide and amplify resistance
- st-03-archetypes — Fixes that Fail archetype as a form of resistance
- cc-01-dags — causal diagrams for analyzing policy conflicts
- de-01