ANALYSIS TOOL BOX

Standing Ovation Model

Will a group follow a few early movers? Define the actors, their private inclinations and conformity thresholds, and who can observe whom — then run the model many times to see how close the group is to a tipping point, which actors' moves matter most, and what to watch.

How this technique works
Key question (required)
Analyst
In this analysis, “standing” means…
Teaching Note

Each actor first decides on its own — it stands with probability equal to its private inclination. Then actors look at those they can observe and switch to conform when at least their threshold share is doing the opposite. Running the model many times shows how close the group is to a tipping point, which actors matter because they are highly visible (the front-row effect), and whether a consensus would be independent judgment or a fragile cascade.

Start from a setup

All setups are fictional and their numbers illustrative. Loading one replaces the current scenarios and comments; saved versions are kept.

Scenarios

Model

Default conformity threshold

Used for every actor without its own threshold.

Actors
ActorGroupPrivate inclinationThresholdStatusRationale & source
Inclination: the chance the actor stands on its own judgment (L 0.1 · M 0.3 · H 0.6). Threshold: share of observed actors who must be doing the opposite before it switches — blank uses the default (0.6).
Who observes whom
Observer ↓ watches →Sees
4
4
1
1
4
4
3
4
5
4
4
3
Seen by22111122122321

Rows are observers, columns are the actors they watch; asymmetry is allowed. Click a column name to make everyone watch that actor (e.g. a shared broadcaster); click a row name to let that actor watch everyone. Actors seen by many are the “front row.”

Dynamics
Dynamics
Updating rule

Each round, actors update one at a time in a fresh random order, each seeing the moves made before theirs.

Can actors reverse?

Two-way: standing actors sit back down when enough of those they see are seated (Miller & Page). One-way suits irreversible moves like withdrawals or resignations.

Rounds per run

Runs stop early once nothing changes (no noise).

Noise per round

Chance a free actor flips for reasons outside the model.

Full cascade at

7 or more of 12 standing (≥ 58%)

Fizzle at or below

3 or fewer of 12 standing (≤ 25%)

Assumptions and rationale

Every parameter is an analyst judgment. Record why you chose it, where it came from, and — where one exists — the Key Assumptions Check row it rests on.

Cascade and fizzle levels
Cascade ≥ 58%; fizzle ≤ 25%

No Key Assumptions Check in this browser. Start one to link rows here.

Private inclinations
Arvenia 0.6; Dalmar 0.6; Belmora 0.3; Corvia 0.3; Estrada 0.6; Feronia 0.3; Galvane 0.3; Hestia 0.1; Ilvar 0.3; Jorund 0.3; Kestrel 0.1; Lunaria 0.1

No Key Assumptions Check in this browser. Start one to link rows here.

Conformity thresholds
Default 0.6; overrides: Belmora 0.7, Galvane 0.3, Hestia 0.3

No Key Assumptions Check in this browser. Start one to link rows here.

Visibility (who observes whom)
41 observation links among 12 actors

No Key Assumptions Check in this browser. Start one to link rows here.

Updating rule and rounds
Asynchronous — random order, 12 rounds, two-way

No Key Assumptions Check in this browser. Start one to link rows here.

Noise
2%

No Key Assumptions Check in this browser. Start one to link rows here.

Seeded early movers
Arvenia (stands), Dalmar (stands)

No Key Assumptions Check in this browser. Start one to link rows here.

Run

Run the model to see the distribution of outcomes, the time-to-tipping curve and the influence ranking. Everything runs in your browser.

Replay one run

Not standingStanding on own judgmentStanding because others didForced standing (early mover)Forced seated
Round 3 · 3/3
ArveniaDalmarBelmoraCorviaEstradaFeroniaGalvaneHestiaIlvarJorundKestrelLunaria

2 of 12 standing (17%) · own judgment 0 · followed others 0 · early movers 2

Run #1 of the current model. Hover an actor to see who it observes and why it moved.

Sensitivity sweep

Parameter
From
To
Points
Runs per point

Varies the default threshold; actors with their own threshold keep it.

Compare scenarios

Run two or more scenarios (or use Run all) to compare their outcome distributions here.

Bottom line

Drafts use the ICD 203 scale (almost no chance · very unlikely · unlikely · roughly even chance · likely · very likely · almost certain). Edit freely — the saved text is what exports use.

Indicators to watch

Comments and review

    Analyst notes

    Versions

    Your work autosaves in this browser. Saving a version records a restorable snapshot of every scenario’s parameters with the time, author (set your name above) and note. Results aren’t stored in versions — each scenario’s seed is, so re-running a restored version reproduces the same numbers.

    No saved versions yet.

    Runs entirely in your browser. A Python function with the classic model’s parameters is SimulateStandingOvationModel (its random draws differ, so seeds won’t match exactly).

    About this technique

    What is the Standing Ovation Model?

    The Standing Ovation Model asks a simple question: when a performance ends and a few people stand, will the whole audience follow, or will the ovation fizzle? Each audience member forms a private judgment of the show, then adjusts based on what the people they can see are doing. John Miller and Scott Page used it in 2004 as a teaching example for agent-based modeling. The same audience, with the same information, can produce opposite outcomes depending on who sees whom and who moves first.

    When to use it

    Use the model when actors combine their own judgment with what they see others doing. Examples include states joining a sanctions regime, depositors withdrawing funds, accounts amplifying a narrative, and banks exiting a jurisdiction. It won't give you a point prediction. What it gives you is a structured answer to the question of how fragile or self-reinforcing a cascade is, and which actors' early moves matter most.

    How to use this tool

    1. State the key question and define what "standing" means, such as "publicly announces suspension of participation."
    2. Choose a mode: the classic auditorium of rows and seats, or custom actors, where you specify who observes whom.
    3. Set private judgment: how good the "show" is, the bar each actor needs to act alone, and how much perception varies.
    4. Set conformity and early movers: how many visible neighbors it takes to switch, and who moves first.
    5. Run many simulations to see the spread of outcomes, the time to tipping and an influence ranking of actors. Replay any single run step by step.
    6. Run sensitivity sweeps and compare scenarios, then write the bottom line and the indicators to watch.

    Common pitfalls

    • Reading one run as the answer. Outcomes vary from run to run, so look at the spread.
    • Guessing visibility. Who watches whom drives the results, so base it on evidence.
    • Calling it a forecast. It explores how a cascade could unfold; it doesn't predict that it will.

    Related