Individual reviews describe one world. These accumulate across all six volumes, which is where the figures start to mean something and where they can be argued with properly.
48 games logged · 184 landmarks recorded · both notes revised at least once
Survey
Twenty-nine games with a compass ribbon and nineteen without. The compass games are consistently easier to travel and consistently harder to remember, and after six volumes the gap is wide enough that the burden of proof has moved.
Catalogue
Every structure I actually used as a reference point across six volumes, sorted by the property that made it usable. Height helps less than expected, asymmetry helps far more, and the single most useful quality is one nobody designs for deliberately.
A new note is only started when there is enough logged data to say something that a single review cannot. Two have reached that bar in six volumes. Three others were started and abandoned because the pattern dissolved once the sample grew past a dozen games.
Both surviving notes have been revised, and both keep the earlier version visible with the reasoning that overturned it written underneath. The revisions are the most useful part of either page.
The hypothesis was that fast travel destroys recall. It does not, at least not measurably — games where fast travel is restricted to landmarks score the same as games without it. The claim did not survive twelve games.
No relationship found at all. The largest world I have logged fully and the smallest both produced full recall, and the second-largest produced none. Size predicts nothing.
Started on the theory that vertical worlds are harder to learn. Nine games in, the split was even and the interesting variable turned out to be whether elevation is visible from below, which is now part of the landmarks note instead.
