The Sandbox Notebook
Open-ended games, written down
A game without an ending cannot be reviewed in the usual way. This page explains what is recorded instead, and what that record is worth.
The Sandbox Notebook
Open-ended games, written down
A game without an ending cannot be reviewed in the usual way. This page explains what is recorded instead, and what that record is worth.
Fourteen games, 856 hours, and a set of rules about writing things down while they happen rather than afterwards. Everything on this site depends on those rules, so they are set out here in full, including where they fail.
Every session is logged at the desk, in a physical notebook, with the clock time at the left margin. Four things go in: when play started and stopped, what I was trying to do, anything that happened worth remembering, and any goal I set myself along with the moment I set it.
Sketches are used freely, because a settlement layout takes thirty seconds to draw and five minutes to describe. Several of the diagrams that produced the figures in Ledger B are drawings rather than sentences.
The headline figure for every entry, because in a game with no ending nothing else is comparable. It is wall-clock time at the desk, taken from the log rather than from any in-game or platform counter, and it excludes time spent reading about the game elsewhere.
An entry requires a minimum of four hours before it is written at all. Two notebooks in the file sit close to that floor and both say so explicitly, because a four-hour log is a record of a first impression and should not be read as anything more.
The timestamp at which I first wrote down an objective the game had not requested. To qualify it must be specific, unprompted, and pursued for at least twenty minutes afterwards.
General curiosity does not count, and neither does learning the controls. This is a deliberately strict definition and it pushes every figure later than a looser one would; Ledger A depends on it being applied identically across all fourteen games.
Every memorable event is classified at the moment it is written as one of two things. Supplied means it happened because the game did something — a season turned, a structure failed, a settler left. Invented means it happened because I decided something the game had not proposed.
Ambiguous cases are recorded as supplied. That biases every invented percentage downward by a few points, which is stated here and on the ledger so that the bias is visible rather than hidden. It does not affect the ordering between games, which is what the findings rest on.
Every entry names one reason, chosen from what the log actually shows rather than from how I felt afterwards. The six categories in use are boredom, satisfaction, exhaustion of a system, a permanent consequence, a patch, and a lost save.
The distinction between boredom and exhaustion matters and is applied carefully. Boredom means the activity stopped being interesting; exhaustion means I understood everything the system could produce. Quarry Gap is the clearest case of the second, and its last log line — built the gap crane again, better, felt nothing — is why.
Every game is purchased at retail with the site's own money. No review copies, no early access, nothing played while it is still in beta, and no game entered into the notebook on the strength of a free weekend.
Games are chosen because they are open-ended, not because they are new. Several notebooks in this file concern games that were released years before they were logged, and the file will never be current with anything.
A wrong figure is struck through with the replacement beside it and the date of the change. A wrong finding is handled differently: the previous position stays on the page and the reasoning that overturned it is written underneath.
Ledger B has already done this once. It was started to demonstrate that a high invented figure indicates a generous sandbox and now argues close to the opposite, and both positions are printed in order.
First, one player. Every figure here is filtered through one set of habits, and a player who arrives at a sandbox with a project already in mind would produce different numbers from the same games. Both ledgers say where this matters most.
Second, small samples. Fourteen games, and the most interesting band in Ledger B contains four of them. Nothing here should be read as a general claim about design.
Third, memorability is self-assessed. An event enters the count because it seemed worth writing down, and quiet supplied events — a season passing without incident — are almost certainly under-recorded.
Fourth, no multiplayer and no mods. Every notebook is a single-player game in its shipped state, which excludes an enormous amount of what people actually do with sandboxes.
Score anything. There is no number out of ten anywhere on this site, because a sandbox has no quality independent of what somebody does with it. Hours logged is a record of how long a set of materials held one pair of hands, and that is the largest claim this method supports.