Thirty-one games have been measured; six have been reviewed. Everything learned from the other twenty-five lives here, in three logs that grow as new figures arrive.
The central log. Frequent saving turns out to predict almost nothing about whether a game can be left. What predicts it is whether a unit of progress ends where a person would naturally stand up.
Intended session length recorded against actual, twelve games, twenty evenings each. The overrun correlates with one variable and it is not difficulty, length or quality.
A brief note on legibility after midnight: daylight-calibrated brightness curves, no low-light preset, and interface text under thirteen pixels at 1440p.
A review needs a game that does something distinctive at the moment a session ends. A log needs only a pattern that holds across enough games to be worth stating.
Rule
No session figure enters a log on fewer than twenty real evenings. Session length varies enormously with tiredness, and a single measurement is worthless.
Rule
The clock is stopped and written down when the session actually ends, not reconstructed the next morning. Tired estimates run short by roughly a quarter.
Rule
When an update changes a save system, the old figures are struck out rather than replaced, and the game is measured again from scratch.
