The rules you don't write down
Every published game is surrounded by a second, unpublished game: the local agreements, the house rules, the things everyone at the table knows and nobody printed.
Ask anyone about a board game they played as a child and within a minute you will hear a rule that is not in the box. Money under the corner of the board. A free pass on the first turn. A ban on a move everyone agreed was unpleasant. These are usually told as anecdotes, but they are design evidence, and they are the most widely distributed design evidence there is.
The interesting question is not whether people play games wrongly. It is what the corrections are for.
Three kinds of unwritten rule
The first kind repairs a pacing failure. Almost every house rule that adds a bonus at the start of a game is fixing a slow opening, and almost every one that accelerates the end is fixing a decided-but-unfinished middle. Players will not usually say “the pacing is wrong”; they will invent a patch and carry on.
The second kind protects the social occasion. Rules against targeting the same person twice, against certain kinds of bluffing, against playing the strongest available line when it would end someone's evening — these exist because the game is not the only thing happening at the table. Designers who dismiss them as bad play are describing an idealised table that does not exist.
The third kind is pure local invention: a new phase, a new piece, a variant scoring rule, kept because a group liked it. This is folk design, and it is often better than it has any right to be, because it has been tested against exactly one group's tastes with a sample size measured in years.
Playground rules are the clearest case
Children's outdoor games have no publisher at all, and they show the process without any of the artefacts. The rules are agreed at the start, renegotiated in the middle, enforced by argument and forgotten by the following summer. A game is whatever survives being explained by a seven-year-old to another seven-year-old, which is a compression test no design document has to pass.
What survives is instructive: a simple win condition, a small number of very physical verbs, and an enormous amount of local variation in everything else. That is roughly the profile of the board games that stay in print for decades.
Why this matters for anyone making things
If your rules will be amended, you can choose whether to make amendment easy. A game with clearly separated modules is straightforward to modify without breaking; a game whose systems interlock at every point is not, and its players will either play it as written or abandon it. Neither is wrong, but the choice is worth making deliberately rather than discovering it in a forum thread.
It also suggests where to look during testing. When a group invents a house rule in front of you, they have located a problem and proposed a solution in one motion. The solution is frequently wrong — it is a patch, made quickly, by people who cannot change the underlying structure. The location is nearly always right.
Writing about the unwritten
For anyone writing criticism, the unwritten layer is a standing correction to a common error: treating the rulebook as the game. The rulebook is the part the designer controls. The game is what a specific group of people did on a specific evening, using the rulebook as their main but not only source.
This does not dissolve the object into pure subjectivity. It means the honest unit of description is a play, not a product, and that a review which never mentions who was at the table has left out one of its variables.