NEED MODEL
TO MAKE PLAN

Rocky names the whole method.

Three sentences. One operating model for complex work.

Most teams are comfortable with sentence two. The work is protecting sentence one and sentence three.

Most project work lives in sentence two.

We rush from symptom to plan before anyone has built a shared explanation of the system.

The missing middle is the modeling conversation: the work that turns a symptom into a shared explanation.

"We need a platform adoption plan." Maybe. But first: why is adoption stalled? Lack of awareness? Bad fit? Broken incentives? The wrong platform model altogether?

The heroes do not find the answer.

They improve the model until the right action becomes visible.

The drama is learning. The team iterates through explanations until the right action becomes visible.

It takes fifty people to fix what five understand.

The size of the incident call is a symptom, not a response strategy.

Everyone else is helping reconstruct it in real time.

A box is a noun. A loop is a verb. Outages are verbs we never wrote down.

Stocks. Flows. Feedback loops.

Three primitives turn tacit expertise into a model the team can inspect.

A shared vocabulary for making system behavior something a team can argue about and change.

Six shapes describe how systems break.

The stack changes. The failure structure repeats.

Failure shapes name the behavior. The model tells you where to intervene.

Need model to know where to act.

A model selects the depth of intervention, not just explains the behavior.

Same symptom. Different model. Different leverage.

Not every intervention is equal.

Meadows' leverage points move from visible adjustments to deeper changes in system behavior.

Most teams iterate at the surface. The model tells you when the behavior lives deeper.

Organizations overuse the weakest levers.

The visible levers are manageable. The powerful levers require questioning the model.

Shallow levers are easy to reach. The model tells you when the behavior lives somewhere they cannot touch.

The loop that drives your growth can also drive your worst collapse.

"The model of the system became the system. And the system outgrew the model."

The six don't fire at random.

They generate each other in a fixed order. Learn the topology once and you stop meeting these failures for the first time at 2am.

Paradigm lock starves an information channel: the missing signal lets a reinforcing loop compound, the loop drains a buffer until it collapses, and the balancing loop fires into a system already past the point of easy correction.

Every mode in the chain is invisible from inside the paradigm that created it.

Stand at any node. Name what's upstream. That's the next incident before it surfaces.

"Platform adoption has stalled."

The model changes what the team can imagine.

Same symptom. Different model. Different leverage entirely.

"AI quality is drifting."

A deeper intervention improves the system. Raising the eval threshold improves the gauge.

When instruments co-evolve with the thing they measure, the dashboard can stay green while reality gets worse.

"The incident keeps coming back."

A postmortem that stops at the surfacing layer buys the recurrence.

The fix changes from "restart the consumer" to "change the loop that keeps recreating the same failure."

A repeatable practice for learning where the system can actually be changed.

Not a workshop. A way of working.

We stopped asking "Who owns the box?"

We started asking what behavior the system was producing, and what leverage point governed it.

Good teams execute.
Amazing teams model together.

The modeling conversation is the work. Coordinated action follows from it.

You cannot change a system until you find where the leverage lives.