Tensions in Practice: Uniform output leaves selection nothing to sort¶
Two candidate populations with the same initial mean
Compare two invented four-member populations. One has scores 2, 2, 2, 2; the other 0, 2, 2, 4. Both start with mean 2. Apply the same selection rule: choose a highest-score existing member and make four faithful copies, with no mutation or score change during copying. The uniform population stays at 2; the varied population can become all 4. But before selection, the varied population contains a score-0 member and has less consistent performance.
Deliver a consistent current level
Keep all present members at the already-established score 2.
Retain support for a higher selected level
Include existing variants on which selection can act, while accepting their initial spread.
Why these aims pull against each other
Selecting more sharply cannot move a uniform population beyond the sole score it contains. A spread with the same mean supplies a higher candidate, but also includes a lower one and loses its spread when copied from the maximum.
Choose an arrangement to see what changes and what remains difficult.
Before lists the starting scores; After lists four new copies of a selected maximum. The initial mean is fixed at 2 in both arrangements. Variance measures score spread, not uncertainty in an estimate.
What this choice protects
What it costs
When it fits
Compare the arrangements
Start uniform
All four members score 2. The maximum is 2; copying any highest member gives four new members at 2. A second identical selection step would still give 2.
| Before | After | |
|---|---|---|
| Slot 1 | 2 | 2 |
| Slot 2 | 2 | 2 |
| Slot 3 | 2 | 2 |
| Slot 4 | 2 | 2 |
| Mean | 2 | 2 |
| Variance | 0 | 0 |
- What it protects
- Current performance is consistent at 2, with no below-2 member in the initial population.
- What it costs
- Selection among these existing values has no way to raise the mean; it would require a new value from some process outside this rule.
- When it fits
- Plausible when reliable delivery at 2 is the current aim and generating or retaining off-level variants is not worth its burden.
Illustration note: This is an editorial, deliberately bounded illustration. Its stated rules and any numbers are invented, not observations, recommended settings, or predictions.
Start with a spread
The initial scores are 0, 2, 2 and 4. Copy the score-4 member into all four new slots. The mean rises to 4 and the variance falls from 2 to 0.
| Before | After | |
|---|---|---|
| Slot 1 | 0 | 4 |
| Slot 2 | 2 | 4 |
| Slot 3 | 2 | 4 |
| Slot 4 | 4 | 4 |
| Mean | 2 | 4 |
| Variance | 2 | 0 |
- What it protects
- An existing high-score variant gives this selection rule a route to a higher mean without inventing a value during copying.
- What it costs
- Initial performance includes score 0. After selecting all 4, this rule has no remaining spread and cannot move beyond 4 without a new source of variation.
- When it fits
- Plausible during a bounded improvement phase when the initial low-score member is tolerable and faithful selection/copying is feasible.
Illustration note: This is an editorial, deliberately bounded illustration. Its stated rules and any numbers are invented, not observations, recommended settings, or predictions.
What this illustration does—and does not—establish
The source establishes the structural tension; the concrete alternatives and their conditional costs are editorial synthesis. No arrangement is a universal recommendation.
- Scores and populations are invented. This finite maximum-selection rule is not a calibrated biological, genetic or organizational response model.
- Variance is the population average of squared deviations from the mean: (4 + 0 + 0 + 4) / 4 = 2 for the varied initial population. No continuous linear response law is claimed for this hard selection rule.
- After columns are newly copied slots, not unexplained improvements inside the original members. Copying preserves the selected score and never creates a new value.
- The starting populations are a restricted comparison with equal mean. This does not claim that a producer already able to supply all 4 should prefer all 2, or that more variance in an irrelevant trait improves the actual goal.
Source entries
Variance Bounds Selection Response
Variance bounds selection response Scalar: High Variance Versus a Stable Mean supplies the local tension. The setting, alternative arrangements, and stipulated consequences are editorial applications.
Scalar: High Variance Versus a Stable Mean
The prime optimizes for rate of mean change, but many systems need the mean to *hold*, not move — a converged policy to exploit, a bred line to propagate true, a strategy to run at scale. Variance that fuels adaptation simultaneously prevents the consolidation that exploitation requires.