Four studies so far. Each one takes a single interface element, measures it identically everywhere, and publishes the table rather than the conclusion alone.
Slot counts, sort options and action cost across nineteen games, with the full table. The grid survives because it is inexpensive to build and easy to balance, and the two games here that dropped it both improved.
The deepest options menu we have recorded, described screen by screen, and an argument that menu depth is the most reliable available proxy for whether a game's settings were used by anybody outside its own team.
Four investigation games measured for how much of the case file is legible without zooming. Three require a zoom to read a card the game expects you to cross-reference from memory, which converts deduction into note-taking.
We record full rebinding support separately from every other option, because a game that cannot be remapped cannot be operated at all by a substantial number of the people who bought it. Fifty-two games checked; the results are worse than expected.
