I have been thinking about puzzle instructions and example puzzles lately. It is just a coincidence (a happy one though) that there was a discussion in the UK forum touching the matter of instruction texts recently, since it means such topics are really of relevance, so I will address the first subject today. Presumably I will follow up on example puzzles in another article soon.
Before I explain what I think puzzle instructions should look like, let us ask the question what the purpose of these instructions is. That may sound like a silly question, but I assure you it is not. Why do we have written puzzle rules in every contest? More to the point, who are they for?
Basically, there are two groups of puzzlers: those who are already familiar with the puzzle type, and those who are not. Last week I spoke about an “elite” in the puzzle community, and I guess it is reasonable to assume that elite solvers are familiar with most of the puzzle types that occur in competition. But we should keep in mind that there are many others out there with a more limited solving experience.
Let me say straight away that I consider myself part of said elite. When I go over an instruction booklet (especially for an event where I do not plan a more extensive preparation), I typically check the puzzle names first and then skip the parts which I already know. For example, when I read Masyu, Fillomino or Star Battle, I am confident that the rules I would find there are the ones I know by heart. This may sound reckless to the point of arrogance, but I trust the authors/hosts to change the puzzle names – another interesting topic for another article – if the rules deviate from the “standard” ones somewhere.
Sometimes the puzzle requires an additional piece of information which I know will be provided with the puzzle, such as the range of numbers that must be entered in Japanese Sums. In Latin Squares the information comes with the puzzle size, hence once again there is need to check the rules in Skyscrapers or Kropki. Sometimes however, the extra information is qualitative in nature and cannot be expected to be given with the contest puzzle.
Typical examples would be if cells in Arukone can remain empty, or whether the available shapes (in Placement or some Dissection puzzles) may be reflected – rotations one can often take for granted, but the matter of reflection is unclear on occasion. When such cases occur, I definitely study the relevant part of instructions.
There are puzzles for which no standards have been established, for instance in rare Loop puzzle types (when it is not clear if the loop must pass through all grid cells) or in some unfamiliar Placement puzzles when one cannot assume if the objects in question are allowed to teach each other or not. Also, there may be a standard which I frequently forget about, such as the question whether the shaded area in a Coral may have interior islands.
So you see, the less familiar I am with a puzzle type, the more thoroughly I go over the instructions. However, and this is crucial, we must accept that unexperienced solvers are not familiar with any puzzle type to the degree that they could afford to skip this process. And it is my belief that the puzzle instructions are mainly for those solvers.
Basically, a set of instructions should be such that someone who has not seen the puzzle type at all has a good idea what he is supposed to do. There are virtually always example puzzles for illustration purposes, but most of the time one can demand that the rules themselves are sufficiently clear even without an example.
The first part of the instructions should describe the main task, e.g. “Draw a closed loop …” or “Divide the grid into the given shapes …” – a short and yet powerful sentence which makes it clear from the start in which direction the puzzle will go. When instructions for new puzzle types are to be written, I suggest that they always lead with this kind of objective.
The second part of the instructions should contain details for the structure of that which must be drawn or entered. For example, in Latin Square puzzles the next clause that follows is something like “each number must occur exactly once in every row and every column”. In Shading puzzles like Nurikabe or LITS, this is the part where it is explained that the shaded cells must form one connected area (potentially without 2×2 squares) – you get the idea.
This is not about language issues. When writing the rules for contests with (I hope) international participation, I do my best to use good English. Sometimes I spend ages on annoying (and perhaps superfluous, to a native speaker) questions, for example what the difference between “each” and “every” is in the context of a puzzle constraint. But my thoughts are meant to apply to instruction texts in all languages, since they are about the structure of the contents.
The rest of the instructions should, obviously, cover the role of any kind of clues. It is impossible to give a more precise guidance here, since every puzzle type is different regarding the scope and the nature of these rules. As said before, though, one thing to always keep in mind is that they must be comprehensible for someone who does not know the puzzle type yet.
In particular, the final part should cover relevant situations which the rules are not entirely clear about, such as clues of 0 in a Double Block puzzle or the directive which kind of circle must be given between adjacent numbers 1 and 2 in a Kropki. Such situations might or might not occur in the actual contest puzzle, and the instructions should not be suggestive in either direction.
Several months ago I wrote an article about a potential taxonomy of logical puzzles, and in my head I still see Ulrich snicker – he knows this is a pet topic of mine. But in a context such as this, it can be quite useful. You see, the structure I suggested above implies that puzzle types which are related in the taxonomic tree (perhaps at the level or Class or Order) will start with the same sentence or paragraph, and there is no need to vary the first part of the instructions. Only later, when it comes to the finer points of the Genus or Species, the instructions diverge.
I am not suggesting a complete and utter standardization of puzzle rules. (A “library” for puzzle instructions might be valuable, but I guess such a project would be totally impracticable for various reasons.) Instead I am trying to point out where the priority in the phrasing of instructions should lie.
One closing remark. There are some purely abstract puzzle types such as Sudoku, Japanese Sums, Maysu or Tapa, and then there are some which contain references to real-life objects, such as Skyscrapers, Anglers or Tents. One should be very careful about at which point terminology from real life should enter the rules.
The point is, references can create wrong associations. As I see it, the first part should always remain as abstract as possible. In Skyscraper puzzles, for instance, I always lead with a sentence like “Enter number from 1 to 5 …” and explain only afterwards that the numbers represent buildings and the clues are based of a concept of “visibility”. In my opinion this is the way to go.
In the past I have sometimes seen Skyscraper instructions starting with “Enter a building …”, at which point I think the goose is cooked. How do you enter buildings on a piece of paper? Once again, imagine someone who does not know the puzzle type. I have seen several people treating the grid as some kind of lateral view instead of the ground layout it is supposed to represent, and it took me some time – basically providing the detailed version of the instructions – to sort things out.
The attempt to save space in this regard is misguided in my humble opinion. If it takes an extra sentence to makes the rules comprehensible, so be it. In the course of the WSPC preparations, Ulrich convinced me of the above structure for puzzle rules, and I have found it to be a good idea ever since.