Measurement principles

Methodology

A useful score needs a defined rule, a comparable group, and an honest account of the noise. This is how Play by Offline Observer approaches all three.

01

Measure the whole browser path

Reaction results include the visual presentation, display refresh, human response, input hardware, and browser event delivery. We never describe the number as a direct laboratory measure of the nervous system.

02

Keep unlike conditions apart

Game, scoring version, mode, duration, daily seed, and broad input class form a ranking group. A 10-second touch run does not compete with a 60-second mouse run.

03

Prefer distributions to labels

A percentile is a comparison with a specific sample. We show the sample size, median, and scoring version when shared data becomes available.

04

Version every rule change

If timing, difficulty, or scoring changes materially, a new scoring version starts. Historical scores never silently move into a new ruleset.

Still curious?

Read how scores are grouped, stored, and compared across games.