Puzzle design
This is a particularly tricky topic, as philosophies about puzzle design can vary widely from one constructor to another. Aspiring constructors should take everything here with a grain of salt, and contributors should be sure to sign their opinions. If you disagree with someone else's advice, don't erase it, but instead offer your own viewpoint as counterpoint. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
The Basics
If you are absolutely new to writing puzzles, don't panic! Everyone has to start somewhere. An excellent general-purpose resource is Puzzlecraft, by Mike Selinker and Thomas Snyder. Puzzlecraft provides the technical details of creating a large number of standard puzzle types, and I recommend reading it through at least once if you are planning to write puzzles for a BAPHL or other puzzle hunt. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
Also, if you're a rookie constructor, consider collaborating with a more experienced constructor or editor your first time out. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
Things to keep in mind
I don't have my copy of Puzzlecraft on hand to give an exact quotation, but when defining what distinguishes a puzzle from a game, Mike Selinker says that in a puzzle, one side (the constructor) has all the advantages, but the other side (the solver) is expected to be able to win, every time. In other words, writing a puzzle is a battle of wits that the constructor should always plan to lose. I don't always agree with the "battle" metaphor, but the constructor's goal should never be to outwit the solver just to show off how clever they are. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
Of course, you also have to keep in mind the intended audience and the resources available to the solver. There is a difference in audience between a Monday crossword puzzle, a BAPHL puzzle, and an MIT Mystery Hunt puzzle, and a comparable difference in resources. You can expect that 2 or 3 people might work on a single BAPHL puzzle, and that they can draw on the combined knowledge of them, their teammates, and whatever they can find with a quick web search on a mobile device (assuming they can access the internet in the first place). Ideally, these three sources of knowledge should be used in decreasing order from left to right. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
Veteran constructor Foggy Brume (constructor for MH 2007 & 2009, P&A Magazine, Puzzle Boats 1 & 2, and many other hunts) posted a list of 12 guidelines he had developed over the years: part 1, part 2, and part 3. Some of these are dependent on the parameters of the particular puzzle hunt -- how many people are on a team, what resources they have available -- and none of them should be considered hard and fast rules. But if you're going to go outside Foggy's guidelines, you should have a good idea of what you're doing and why. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
Accommodating newbie solvers
BAPHL, with its tradition of 2 tiers of difficulty, professes to be friendly to newbie solvers. If you want newer solvers to have a good time, you have to take some steps to accommodate them:
While writing an entirely different set of puzzles for newbie solvers is impractical for constructors, (and prone to backfiring, when a "newbie" puzzle turns out to be harder than its "expert" equivalent) you should always consider what additional information you can give to newbie solvers. This can take several forms: enumerations/blanks for answers, clarifying particularly tricky clues, giving clues in a more helpful order, or simply making the presentation of the puzzle clearer. However, you should resist the temptation to give new solvers extra clues in flavortext, as newbie solvers will have difficulty figuring out what parts of the flavortext are relevant. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
In a GC Summit talk given by Ian Tullis a few years ago, he pointed out that certain puzzle conventions that experienced solvers take for granted, like indexing, can be difficult for new solvers to figure out on their own. Consider the difference between "Our first president (7/10)" and "Our first president: _ _ _ _ _ _ * _ _ _". These both convey the same information, but newer solvers will latch on to the latter form more easily. --Tortoise (talk) 18:53, 15 August 2014 (PDT)
In my own experience, judging from testsolving data from BAPHL 10 and other hunts I've helped run, one of the biggest stumbling blocks for less experienced solvers can be the ordering mechanism. I've seen many instances of teams getting the extraction but not figuring out the ordering mechanism, or even realizing that there is an ordering mechanism. How to address this issue will depend on the specifics of the puzzle, but think about ways to make the ordering mechanism (or even just the existence of an ordering mechanism) clearer. --Tortoise (talk) 18:53, 15 August 2014 (PDT)