STRATA
Skip to study
XLVII

III.11 / INTELLIGENCE

Collective intelligence and distributed cognition

When does pooling individual estimates improve a group’s prediction, and how much does the result depend on shared errors?

A source-led introduction1 source references1 related experiments

Start with the question.

Collective intelligence concerns how a group combines partial observations into a judgment or action. Galton's historical crowd-estimation report offers an early case, while the deeper research question concerns independence, diversity of error, and aggregation. A large group can still repeat one source's mistake. The structure of communication therefore belongs beside the final average when explaining a collective result.

Compare an average with a median.

For illustrative estimates of 8, 9, 10, 11, and 100, the mean is 27.6 while the median is 10. Ask which error process could produce the extreme value and which summary fits the decision. Then allow participants to copy the first estimate; the group size stays the same while the independence of its information changes.

What would you inspect?

  • The distribution of individual estimates and dependence between contributors.
  • Aggregation rule, incentives, and whether revisions occur before or after seeing others' answers.

Take the next step.

Construct a small public-data example comparing independent aggregation with copied or correlated estimates, then explain when each summary fails.

Put part of the question to work.

These models isolate a mechanism you can vary. Their assumptions define the part of this topic they can illuminate.

Try connectivity and removal.

Change one parameter, compare the numerical output, and export the result. The experiment loads when you open it.

Open its full page

Watch the quantities change.

12 seconds · Original model visualization · Silent

Connectivity and removal

Compare local connections with the paths that hold a network together.

Read the explanation and method

Forty-eight points are drawn from one fixed random seed. A link joins two points whenever their distance is within the displayed radius.

The radius grows from 0.05 to 0.4 of the unit square. Existing links remain while new links merge previously separate connected groups.

Orange marks the largest connected group. Connectivity here describes paths through this constructed graph; it does not measure information quality, influence, or causality.

Method. Two nodes share an undirected edge when their Euclidean distance is at most the connection radius. After removal, a depth-first traversal assigns every active node to a connected component. Groups are ranked by size; ties use their lowest node identifier.

Read the supporting material.

  1. Vox Populi

    Francis Galton / Nature · A historical crowd-estimation report; does not imply independent or unbiased crowds in all settings.

Follow the comparisons.

Read the neighboring questions.