DESIGN-STAGE PLANNING TOOL

ParEvo iteration planner

Compare how three design settings—constant participants, iterations and a chosen number of surviving lineages—affect the chance that final storylines descend from only a few original contributions, if parents are chosen independently at random.

1. Set up an exercise

Assumed constant throughout
Examples:
Iteration 1 establishes the original lineages
For example, at least 3

Used to calculate the probability of ending with fewer surviving lineages than the number selected. It is not a required target.

2. Estimated outcomes at the final iteration

Convergence on one surviving lineage
—
Chance that only one original lineage remains
Fewer than chosen number
—
Fewer than 3 lineages survive
Expected surviving lineages
—
90% prediction interval: —

3. How risk changes with more iterations

Probability of fewer than 3 surviving lineagesProbability of one surviving lineage

Each point on a curve shows the probability of that outcome after the number of iterations on the horizontal axis, not the probability that exactly that many lineages survive. The selected iteration is marked by a vertical line. These are neutral-model probabilities, not forecasts of participant choices.

How to interpret these estimates

Model: There are the same number of contributions in every iteration. Each contribution after iteration 1 independently selects one parent from the immediately previous iteration, uniformly at random, and several contributions may select the same parent. All iteration-1 contributions are treated as separate original lineages.

What is calculated: Exact probability distributions are obtained with a finite-population ancestral sampling model, equivalent to the neutral Wright–Fisher ancestry process. No Monte Carlo variability is involved.

Design trade-off: More iterations increase opportunities to develop and diversify the later portions of narratives, while also increasing the risk of losing earlier alternatives through extinction. The model does not estimate how different the later narratives will actually become.

Genetic drift: This random ancestry model is analogous to neutral genetic drift in a finite biological population, where lineages disappear by chance rather than through differences in fitness.

What this does not measure: Storyline quality, novelty or narrative richness; participants' real selection preferences; the usefulness of longer storylines; time required; or actual future participation. Structural convergence does not necessarily imply diminishing substantive returns. The results are calculations, not recommendations about which settings to choose.

The first iteration creates the initial set of lineages. Thus, an 8-iteration exercise has seven rounds of lineage sampling after the initial iteration. Version 1.3.