02 / UNDERSTAND
A small vocabulary.
A different way to see.
Start with what accumulates, what connects, and what changes over time. Each concept is a tool for asking a more precise question.
Boundaries
Every view includes something. Every view leaves something out.
A boundary defines what belongs inside an analysis and what is treated as its environment. It may follow a physical edge, an organization, a period of time, or a question. Change the boundary and an apparent improvement can become a transfer of costs.
A team clears its own queue by sending unfinished work to another team. Its local metric improves; the end-to-end waiting time may not.
Who benefits, who carries the cost, and what happens just outside the frame?
Stocks & flows
The level remembers the history of the rates.
A stock accumulates. A flow changes that accumulation per unit of time. Water in a reservoir is measured in volume; water entering it is measured in volume per time. The stock rises whenever the total inflow exceeds the total outflow—even if the inflow is slowing.
An inbox grows when 50 messages arrive and 40 are resolved each day. Cutting arrivals to 45 slows the growth; it does not shrink the inbox.
Are you measuring how much is there, or how quickly it is changing?
Explore it in the lab →Sources
Feedback
An outcome becomes part of the next cause.
A feedback loop connects the effects of an action to future action. Reinforcing loops amplify a change. Balancing loops oppose a departure from a reference. “Positive” and “negative” describe direction, not whether the effect is good or bad.
Interest added to a balance generates more interest. A thermostat, by contrast, reduces heating as the room approaches its target.
Can you trace a closed chain from a change back to itself?
Explore it in the lab →Delays
The response may already be on its way.
Information, decisions, construction and transport all take time. If you respond to a gap before earlier actions have taken effect, the accumulated response can overshoot. Delay alone does not guarantee instability; the decision rule and response strength matter.
You turn up a slow shower again before the warmer water arrives, then suddenly find it too hot.
Which effects are still traveling through the system?
Explore it in the lab →Buffers
A reserve buys time, at a cost.
A buffer is an accumulation that separates fluctuations in supply from fluctuations in use. Its value depends on its size relative to the flows and the disturbances it must absorb. Storage can be costly, and a larger buffer can make changes in the overall level slower.
Stored water can sustain use through a dry spell. A larger empty reservoir, by itself, supplies no extra water.
How long would the reserve last under the disturbance that matters?
Explore it in the lab →Sources
Emergence
Local interactions can produce a collective pattern.
An aggregate pattern may require an explanation in terms of relationships among components. Calling something emergent is the beginning of that explanation: identify the local rules, connections and observable group behavior.
Drivers responding to nearby cars can collectively produce a backward-moving traffic wave, without a bottleneck in the road.
What interaction rule could generate what you see at the larger scale?
Nonlinearity
Twice the push need not mean twice the response.
A nonlinear relationship does not preserve simple proportionality. Saturation, thresholds and interactions can change the response as conditions change. Chaos is a more specific phenomenon; a complicated-looking curve is not sufficient evidence for it.
A simplified congestion model may make travel cost increase disproportionately as road use approaches capacity. Whether that model fits a road needs evidence.
Where do thresholds, ceilings or feedback change the relationship?
Sources
Resilience
Staying useful through disturbance is a particular kind of success.
Resilience and rapid return to an equilibrium are different questions. Specify the disturbance, the timescale and the functions that must persist. A resilient arrangement can also be unfair or harmful; persistence does not establish desirability.
A service can maintain essential functions while operating in a degraded mode. Returning every variable to its previous value is a different objective.
Resilient to what, for whom, and with which functions preserved?
Sources
Networks
A connection can carry help. It can also carry trouble.
A network describes entities and their links. A link might represent communication, a debt, a road or a dependency. The topology and the rule governing what travels along each link jointly determine the behavior.
Sharing a small loss among many counterparties may cushion it. Under other conditions, those obligations provide routes for a much larger failure to spread.
What exactly travels along an edge, in which direction, and under what rule?
Adaptation
Participants can change the rules of their own behavior.
People, organizations and algorithms may respond strategically to a policy or learn from its effects. A relationship measured before an intervention can change after the intervention alters behavior or who is observed.
A recommendation system changes what people see, affecting the interactions later used to train it. The next dataset is partly a consequence of the previous model.
How will people or models respond once this rule becomes part of their environment?
Modularity & scale
Some connections are tighter than others.
In a nearly decomposable system, interactions within a subsystem are stronger or faster than those across subsystems. This can make it useful to study parts over short periods while retaining slower connections at larger scales.
A software component may handle its own internal operations while depending on other components for slower coordination. Whether the separation works depends on the actual dependencies.
Which connections dominate at the timescale you are examining?
Sources
Rules & governance
The arrangements around a resource shape how it is used.
Rules define who can act, what information is visible, how disputes are resolved and how consequences are shared. Common resources need not have a single inevitable fate: institutions can alter the incentives and possibilities for cooperation.
Resource users may monitor use and develop rules fitted to local conditions. Those arrangements require attention to participation, enforcement and change.
Who can make, enforce and revise the rules?