Collaborative Creativity: Idea Recombination in Concept Space

Team

Contributors 4

Novelty vs. usefulness 50% novel

Mutation (idea variation) 30%

Selection pressure 8 kept

Session

Generation

Avg. score

Diversity

Best fit so far

How it works

Every idea is a point in an abstract 3-axis concept space. A glowing target marks the current "goal" region — the kind of solution the team is actually looking for. Each round:

child = lerp(parentA, parentB, t) + mutation·noise
usefulness(idea) = 1 − distance(idea, goal) / maxDist
novelty(idea)    = mean distance(idea, current pool) / maxDist
score(idea)      = w·novelty + (1−w)·usefulness

Every contributor proposes two children per round by recombining two existing accepted ideas — a simple model of building on a teammate's idea — then adding random mutation to explore nearby concept space. All proposals are scored and only the top-scoring ones (set by selection pressure) survive into the next round's parent pool, exactly as a team's converging discussion keeps the strongest contributions and drops the rest.

Novelty vs. usefulness — the weight w above. High novelty rewards ideas far from anything already accepted (divergent thinking); high usefulness rewards ideas close to the goal (convergent thinking). This is the classic novelty-and-usefulness definition of creativity used in creativity research (Amabile; Runco & Jaeger).
Mutation — how far a recombined idea can drift from its two parents; too low and the pool stagnates, too high and useful traits get lost.
Selection pressure — how many of each round's proposals survive. A small kept-pool converges fast but can lose diversity; a large one explores more but converges slower.

Gen 0

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.