Thinking in Systems

Thinking in Systems backs five of Tom's lessons. Under each of his lessons, what the book says about it, and on which pages.

• Signals show before numbers, and numbers come at month four

  Stocks change slowly and delays hide results, so early signals matter and patience before month four fits the book's warning against giving up too soon. pp. 37, 38, 125 and 126. In the book: Is it a system or a bunch of stuff, Questions for testing the value of a model, The eight system traps, Three ways to avoid the tragedy of the commons, Helping a system help itself.

• Macro and micro at the same time

  Meadows says a hierarchy needs enough central control for coordination and enough autonomy for the parts, and that you must watch the short and long term together. pp. 105, 206 and 207. In the book: The eight system traps, Three ways to avoid the tragedy of the commons, Helping a system help itself, Places to intervene in a system, Guidelines for living in a world of systems.

• Move the people you have before you hire

  The book shows a workforce grows by cutting outflow as well as adding inflow, and urges backing the capacities a system already has before bringing in outsiders. pp. 36, 37 and 202. In the book: Is it a system or a bunch of stuff, Questions for testing the value of a model, Places to intervene in a system, Guidelines for living in a world of systems.

• Week one is an audit, and two formats are never up for testing

  The first guideline is to get the beat of the system by studying its behaviour and history before disturbing it, which is what a week-one audit does. pp. 194 and 195. In the book: Places to intervene in a system, Guidelines for living in a world of systems.

• Assumptions made too fast, and risk avoided

  Meadows urges small experimental steps, exposed mental models and error-embracing, and leaning on both intuition and analysis, which matches data plus gut and backing yourself. pp. 195, 196 and 204. In the book: Places to intervene in a system, Guidelines for living in a world of systems.

Meadows argues that a system produces its own behaviour through its stocks, flows, feedback loops and delays, so outside events and individual players are rarely the real cause of a problem. Because our mental models are partial and we attend to events instead of structure, systems keep surprising us: delays cause oscillation, growth meets limits, and bounded rationality turns reasonable local choices into results nobody wants. Recurring structures, the eight traps, explain why fixes fail, and each has a way out through changed goals, rules or information. Her ranked list puts parameters at the bottom of the leverage scale and information flows, rules, goals and paradigms near the top. She closes by saying that systems cannot be controlled or predicted but can be designed and danced with, which takes humility, visible models and constant learning.

• Look to structure for the cause: a system largely causes its own behaviour, and replacing the person in a role rarely changes the result (p. 16, p. 130).
• A purpose is read from what a system does, not from what it says (p. 28).
• You can fill a stock by cutting the outflow as well as raising the inflow (p. 36).
• A delay in a balancing loop makes oscillation likely, and reacting faster can make it worse (p. 72, p. 75).
• Systems deliver exactly what you measure, so do not confuse effort with result (p. 162, p. 217).
• Escape traps by changing goals, feedback or rules; pushing the same lever harder fails (p. 134 to 165).
• Numbers and parameters are the weakest leverage points; information flows, rules and goals are far stronger (p. 170, p. 179, p. 181).
• Missing, late or biased information causes most system malfunction (p. 196).
• Manage for resilience, not only for productivity or stability (p. 98).
• Watch the system's behaviour and history before you disturb it, and put your mental model on paper where others can challenge it (p. 194, p. 196).

• What makes a format repeatable, as opposed to a hit that happened once?

  Events are one-offs, while repeatable behaviour comes from underlying structure, so a repeatable format is a structure with feedback and not a lucky output (p. 108 to 110).

• What does AI actually change about the shape of a content team?

  New information flows are a high leverage point, so AI matters where it adds or restores feedback to the people making decisions (p. 179).

• What is the smallest team that can produce at the volume I am used to?

  The watchmaker fable shows stable subassemblies of about ten elements survive interruption, which suggests small autonomous units under light coordination (p. 102 to 103).

Tom works with creators and founders one to one, from a single session to a fractional role. Work with Tom