MODELS
Menu

Quick Recipes

Try a focused change to a repeating pattern. Explore examples and alternative approaches.

04 /

How do I do it?

Try a focused change to a repeating pattern. Explore examples and alternative approaches.

Choose a named move below to open its instructions.

  • Loop in Three

    Draw 3 changing variables and see if the third genuinely feeds back to the first; if not, it's just a chain.

  • One More Turn

    Ask what this action makes more or less likely the next time the situation recurs.

  • Snowball or Thermostat?

    Test whether a deviation amplifies itself or recruits its own correction.

  • Bathtub Test

    Name what accumulates, what fills it, and what drains it.

  • Delay Detective

    Mark the lag before you double the dose or declare victory.

  • Fix That Feeds

    Ask whether your relief weakens the very capacity that would remove the need for relief.

  • Bottleneck Walk

    Follow the work until you find where waiting piles up and completions stop rising.

  • Ceiling Hunt

    Ask which resource, capacity, trust, or permission will cap the next gain before you chase it.

  • Yo-Yo Check

    Look for the pairing of overcorrection with delayed feedback that drives oscillation.

  • Winner Gets Fuel

    Check whether current success controls future resource allocation and starves the alternatives.

  • Arms-Race Mirror

    Write each side's defensive action as the other side's trigger.

  • Scoreboard Hijack

    Compare the rewarded proxy metric against an independent measure of the real outcome.

  • Goalpost Drift

    Freeze the original standard and check whether it has quietly slid without anyone choosing it.

  • Policy Boomerang

    Trace how a reasonable rule changes behavior in ways that later defeat the rule.

  • Break One Arrow

    Perturb one link while holding everything else steady so you learn which edge actually matters.

  • Friction Flip

    Add friction to a harmful loop or remove it from a helpful one.

  • Seed the Flywheel

    Subsidize the first few turns, then test whether natural returns can sustain the next turn on their own.

  • Shorten the Echo

    Move valid feedback close enough in time to affect the very next attempt.

  • Brake, Reserve, or Cap

    When things are unstable, decide whether you need less gain, more stored buffer, or a bounded response.

  • Four Steering Knobs

    Inspect the target, the signal, the rule, and the reward before blaming execution.