Each game played to the far end without guides, with every journey logged as guided or unguided. The score is not a verdict on the game; it is a verdict on how much of its world survived in my head.
412 hours logged · 4 of 6 recalled from memory · median marker reliance 68%
Exploration
9.2
Forty-one hours of coastline with no map, no compass and no markers. Navigation happens entirely through the shape of the shore and the position of three lighthouses, and after a week away I could sketch the whole thing accurately enough to be useful.
Urban
8.7
Nine districts, no minimap, and a hanging-sign system that teaches the layout without ever explaining itself. The one game this volume where I learned a city the way you learn a real one — by getting the same wrong turn wrong four times.
Underground
8.4
The map is a grid page in the pause menu that you fill in yourself, badly. Forty hours in caves where being lost is the intended condition rather than a failure state, and the only game here that made me keep paper notes.
Open world
6.1
Ninety-one square kilometres, sixty-two hours, and a waypoint on screen for ninety-four per cent of travel. Handsome, enormous, and after a week I could not draw a single junction of it. The clearest case study this volume.
Survival
8.1
One idea, executed without mercy: navigate white fog by a line of marker poles, and if you lose the line you are in real trouble. Twenty-eight hours, almost no landmarks, and the highest recovery cost I have ever logged. Notice only.
Station
7.8
The deck plans painted on the walls are accurate. The ones in your inventory are three refits out of date and contradict them. Thirty hours of a joke that never stops being funny and never stops being useful. Notice only.
Hours logged are timed rather than estimated, and include time spent standing still working out where I was. Marker reliance is the share of travelled distance with a waypoint visible on screen, logged journey by journey.
Recall is the drawing test: a sketch made a week after finishing, scored against the real layout. Full, partly, or not at all — and the method page explains why that scale is deliberately coarse.
