Hey, wait – is employee performance Gaussian distributed?
Many commenters challenge the common corporate assumption that employee performance follows a neat bell curve, arguing instead that output and value often resemble power-law or Pareto distributions with a few people driving disproportionate results. They emphasize that “performance” is hard to define and measure, heavily confounded by luck, role assignment, team dynamics, politics, and management quality, which makes stack ranking and forced bottom‑10% firings statistically dubious and often counterproductive. The thread also raises broader concerns about incentives: companies optimize for cost and risk management rather than fairness, and formal review systems can obscure real contributors while rewarding those best at navigating internal politics.
Shape of Performance Distribution
- Many argue employee performance is not Gaussian: real-world outputs (sales, sports salaries, national wage data, some big-tech data) often look Pareto/power-law with a few “superstars.”
- Others counter that inside a single company or role, samples are small, hiring is selective, and multiple Gaussian-like distributions across roles could aggregate into a Pareto at population level.
- Several insist the article leans too much on national salary data; firm‑internal “hard” performance data is rare and mostly unshared, so claims remain under-evidenced.
What Is “Performance” and Can We Measure It?
- Repeated complaint: “employee performance” is undefined, multi-dimensional, and highly context- and team-dependent.
- Simple metrics (tickets closed, LOC, features shipped) are seen as invalid or heavily distorted by luck, task difficulty, and other people’s bottlenecks.
- Reviews often reward visible heroics and shiny features over prevention, maintenance, and “firefighting” that keeps systems running.
- Some say the only thing consistently measured is “doing what the review system rewards,” not true value creation.
Stack Ranking, Layoffs, and HR Practices
- Stack ranking / “rank and yank” is widely criticized as:
- A political tool to justify soft layoffs and cost-cutting.
- Statistically unsound, often firing people almost at random given measurement error.
- Damaging to collaboration, pushing people into gaming metrics.
- A minority see stack ranking as occasionally useful diagnostic signal, but not a good basis for firing decisions.
Time, Luck, and System Effects
- Single-year performance is seen as a poor predictor of long-term contribution; examples from sports contracts and injury risk are cited.
- Performance is framed as a function of individual ability, manager quality, team composition, and organizational design; many argue the “system” dominates.
- Distinction drawn between “low performers” and “toxic” high-output individuals who damage teams; the latter are seen as especially dangerous.
IQ, Distributions, and Meta-Methodology
- Long side debate on IQ tests: constructed Gaussian outputs vs underlying traits; central limit theorem misuse; polygenic vs environmental effects.
- Used as a caution against naively assuming normality and building entire HR systems around that assumption.
Compensation, Value, and Power
- Debate over whether wages reflect marginal productivity versus bargaining power and information asymmetry.
- Several note the decoupling of productivity growth and wages, and argue performance systems often serve shareholder interests and legal/PR needs more than fairness or accuracy.