Glossary and Notation

Key terms and notation used throughout this series.

Glossary

Term	Definition
Noise	Random variation across repeated measurements
Uncertainty	Lack of confidence in an estimate due to noise, sampling, or bias
Variability	Genuine differences across users, contexts, or inputs
Bias	Systematic error that does not average out
Variance	Quantitative measure of spread

Scores are in percentage points ("points").
σ = standard deviation of one-run measurement noise for a single system (random error).
±h = approximate 95% confidence interval half-width (so CI width is 2h).
When comparing two systems with independent noise σ, the noise on the difference is √2 · σ.