Artificial Life evo-edu.org

Weasel-style selection demonstration for variation, selection, and cumulative change.

Application

Cumulative Selection Explorer

Use a direct responsive workbench to compare score improvement under cumulative selection, while keeping the older Weasel-style route available as a legacy reference.

Learner guide Teacher guide About and provenance Open legacy Weasel
First challengeRun the default target once, then raise mutation rate and compare whether score improvement stays smooth.
Key habitTreat this as a model for discussing cumulative selection and model limits, not as a miniature substitute for full biological evolution.
Legacy memorializedThe older Weasel-style route remains available as a legacy reference while this direct workbench replaces the iframe workflow.

Run Actions

Controls you will use first

Keep run, save, export, and import actions separate from target and mutation settings so the first useful comparison is clear.

Current Run Status

Final score 0
Target length 0
Final generation 0
Mutation rate 0

Try This First

Step 1 Run the defaults and watch how score changes across generations.
Step 2 Raise mutation rate and compare whether progress becomes noisier or faster.
Step 3 Ask what this model does show about cumulative selection and what it does not show about real evolution.

Main Display

Score progression across generations

The line shows how many target characters match at each generation. Use this to compare different mutation rates, targets, and starting strings.

Matching-character score Target length ceiling
What to notice Compare visible score improvement with the explainer panel below, which shows why population-level preservation can make rollback of the best candidate rare.
Compare mechanisms Change one factor at a time so you can compare mutation-rate effects without confusing them with target changes.
Notebook links Population Thinking Natural Selection Adaptation
Candidate and target
Target
Candidate
Interpretation prompt

Run the baseline case first, then compare one changed parameter.

Settings

Selection-demo parameters

These values control the target phrase, starting candidate, mutation rate, number of generations, and random seed.

Probability Explainer

Why best-candidate rollback can become rare without locking

This panel visualizes the population-level quantity that matters in the classic Weasel debate: the probability that a daughter population contains at least one candidate preserving all currently correct positions from the previous best candidate.

One daughter preserves all current correct bases 0
Population preservation probability at max N 0
P(population preserves all current correct bases) = 1 - (1 - (1 - u(K - 1)/K)C)N
Interpretation A larger population sharply increases the chance that at least one daughter preserves everything the current best candidate already got right.
Why this matters This is the population-thinking point behind the claim that visible rollback of the sampled best candidate can be rare without any locking mechanism.
Austringer companion About and provenance Population Thinking