The Limits of Control
Why do plans fail in complex systems, and what can people who run organisations and products do about it?
Every plan is a model of the future, and complex systems punish models that are too confident. This path collects books that explain why.
It starts with how systems behave: feedback loops, delays and constraints that make organisations and markets surprising. It then examines how planners fail, from the grand schemes of states to the overruns of large projects, and the special danger of rare events that nobody’s model includes. It ends with more hopeful questions: how to build systems that benefit from disorder, and whether great outcomes are better pursued indirectly.
Several Roads Not Taken, notably Iridium and Itanium, are case studies in the limits of control. The Ideas on legibility, the outside view and irreversibility connect the books to decisions.
-
The clearest introduction to stocks, flows, feedback loops and delays, the vocabulary needed to see why systems surprise the people who run them.
- Read it for
- Feedback loops, leverage points and why obvious interventions often backfire.
- Carry forward
- Which feedback loop is producing the behaviour I want to change?
-
Applies systems thinking to organisations and introduces the idea of the learning organisation.
- Read it for
- Mental models, shared vision and the archetypes, such as "fixes that fail" and "shifting the burden".
- Carry forward
- Which of our recurring problems are caused by our own past solutions?
-
A novel about a factory that makes one idea vivid: a system's output is set by its constraint, and improving anything else is waste.
- Read it for
- The theory of constraints and the difference between local and global efficiency.
- Carry forward
- What is our bottleneck, and are we improving anything other than it?
-
Shows on a grand scale how simplifying a complex world in order to control it can destroy what made it work.
- Read it for
- Legibility, high modernism and the practical knowledge Scott calls mētis.
- Carry forward
- What does the messy existing system know that our clean plan does not?
-
Brings the argument to projects, with evidence on why large projects overrun and how the successful ones are planned.
- Read it for
- Reference class forecasting, planning slowly and acting fast, and modular design.
- Carry forward
- What is the reference class for our project, and how did those projects go?
-
Turns to the rare, high-impact events that models miss and stories explain afterwards.
- Read it for
- Silent evidence, the narrative fallacy and the difference between domains where averages work and domains where extremes rule.
- Carry forward
- What would a single extreme event do to our plan?
-
Asks what kinds of systems gain from disorder rather than merely surviving it.
- Read it for
- Optionality, redundancy and the asymmetry between small known losses and large unknown gains.
- Carry forward
- Where could we arrange to benefit from volatility instead of fearing it?
-
Why Greatness Cannot Be Planned
Ends with an argument from AI research that ambitious objectives can be deceptive, and that open-ended exploration of interesting stepping stones often reaches further than direct pursuit.
- Read it for
- The idea of deceptive objectives and the value of novelty as a guide.
- Carry forward
- Which interesting stepping stones are we ignoring because they do not serve our stated goal?
Follow the thread