by Will Eisenberg
Will is a French horn player, music teacher, and crossword constructor located in the Twin Cities. He is the current 3rd horn of Orchestra Iowa and a founding member of the Minnesota Horn Quartet. More of his puzzles can be found on the blog Half-Baked Puzzles.
Today, I’d like to share some thoughts about creating and solving puzzles. Hopefully you’ll find something of interest here whether you’re a fellow constructor or just enjoy solving puzzles: American crosswords, cryptics, and all the rest.
I first started thinking about writing this essay quite a while ago, after talking with a friend. They had just published their first cryptic, a huge achievement for anybody. It undoubtedly took lots of time and effort. Their cryptic was elegant, clever, funny, and played on the easy side. Another cryptic, also by a new constructor, had recently come out. This puzzle played significantly harder. Since these puzzles were solved on a few Twitch streams, it became easy to see the difference in difficulty.
My friend confided they felt as though their cryptic was a mere scribble in comparison to an elaborate, complex painting. I couldn’t have disagreed more. The difficulty in the other cryptic came from all the areas it had for improvement: obscure or loose definitions, inconsistent or improper wordplay, and so on. By the same token, the “easiness” of my friend’s cryptic came from the fact that it played fair and followed consistent rules. I told my friend “more difficulty isn’t always better,” but even at the time I knew I was paraphrasing a much deeper conversation. Of course, this is just my own take. Reasonable minds can and probably will disagree.
It boils down to this: The most important thing a constructor can cultivate is trust.
A puzzlemaker can strive to evoke many things: happiness, nostalgia, the feeling of being ‘seen,’ the joy of an “aha!” moment, and even frustration. But the ability to evoke these emotions from the solver is built upon a base of trust. This is because a puzzle offers the solver one fundamental proposition: It is going to take you some time to work out the solution, but it will be worth it.
This is not a small thing to ask. Life is short, time is limited, and options for entertainment are all but unlimited. A solver who does not feel their time is being well-used can and will find something else to do. And beyond all but the easiest of puzzles, the one thing a puzzle asks for from a solver is time.
“You will look at this grid, this level, this question” the puzzle says, “and you will not know the right thing to do. You won’t see the answer–you might not even see what an answer should look like. You will take some time. You will think it over. You will notice that I have given you hints along the way. You will start to make progress. You will form insights about the way things work–insights you’ll make without being told! You will feel good about this, and your progress will build on itself. You will come to the correct solution. You will know it is the correct solution, and you will know why it cannot be anything different.”
This, to my mind, is the contract between a puzzle constructor and a solver. It’s rarely spoken aloud, but it’s still there, and puzzles that violate this contract are likely to produce unhappy solvers. One key component of this contract is that the constructor must provide a way for the solver to figure out the puzzle. This doesn’t mean that the solver is going to have an easy time of it, or even that every solver will be able to complete the puzzle without help. But it does mean that a solution must be possible through hard work and cleverness.
Anyone who’s been puzzling for long enough can probably remember a really hard puzzle they solved that was also really good. But what’s key is that the second part of that equation doesn’t derive solely from the difficulty. Not all difficult puzzles are created equal. Rather, it comes from the trust the solver had in the constructor. That trust fueled the ability for the solver to keep working through a difficult problem, continuing to spend time on it, and come (eventually) to a correct answer. Without trust, that experience might be described as “a slog” or “a wild goose chase,” but with trust it can be “satisfyingly crunchy” or “a worthy challenge.”
So what’s the basis of this trust? One big component is consistency. In the context of American crosswords, there are lots of little rules that are hallmarks of consistency for the solver. The most common is probably the idea that the tense and part of speech of a clue and its answer will always agree. A plural clue will indicate a plural answer, and so on.
By strictly following rules about tense and part of speech, the constructor gains the ability to fairly make all kinds of misdirects. [Live] can be either EXIST or ONAIR, and so on. We’ve all been tricked by this type of misdirect before, but it always feels fair, because it still follows the rules consistently. In having a restricted basis for how clues operate, endless possibilities for wordplay and misdirection open up for the constructor. Restrictions often breed creativity.
A rule that might seem nitpicky is “duplicate entries should be avoided.” There’s justifiable push-back to this rule being ironclad. But when a constructor talks about avoiding dupes, it’s not because the idea of a duped entry is a huge stumbling block to the solver or inherently inelegant. It’s because the solver knows to expect every entry to be a different word or phrase.
Say a solver enters RUNSIN in the top of the grid, and then confirms it with crosses. Then later, they encounter the clue [Something makeup might do] and think “Well, it might be RUN… but we’ve had that already. Hmm, what else could it be? Let me think.” If the answer actually is RUN, this solver has experienced a blow to their trust: they’ve put faith in a consistent element of crosswords, and had that faith violated. The next time they encounter a potential dupe, they’ll have to stop and think. Maybe it is a dupe, after all.
Did this little run-in (no pun intended) present an actual difficulty to the solver? Probably not. After all, that’s what crosses are for, and a duped entry is not such a stretch to imagine that it would cause a serious hold-up. And it’s pretty non-controversial these days to duplicate smaller elements like THE, IN, or ON. But the real problem with a serious dupe is that the solver’s expectations have been frustrated, and not in an exciting way. For a second, they weren’t thinking about solving the puzzle, but instead were taken out of the experience, wondering about the rules of the game. A small thing, but the type of issue that can compound over time, and leave a bad taste in the solver’s mouth.
Another element of consistency is copyediting. For an indie constructor, this is often something that happens mainly via test solving or just self-checking, and it’s not easy. Even the top publications have editing errors in their puzzles; for most indie constructors it’s a matter of “when” and not “if” some kind of mistake will slip by. But I still think it’s worth putting in the effort to be as clean as possible, and this includes spelling, consistency of punctuation, formatting, fact-checking, and so on.
This has nothing to do with grammar pedantry, or trying to avoid looking silly for making a typo, or anything like that. It’s about creating consistency. When you’re scrupulous about the details, the solver knows that what they see on the page is exactly what’s meant to be there. And though I don’t think the odd mistake here and there is anything to lose sleep over, when they start to proliferate, the solver can begin to feel like a tester or a guinea pig. The longer you can keep the solver in a perfect “flow” state, the better, and editing errors tend to interrupt this flow.
“Fairness” is a tricky issue in crosswords. There’s a wide range of opinions as far as the amount of outside knowledge that should be necessary to solve a puzzle. One extreme sees crosswords as largely a logic puzzle, where the solver should be able to suss out what letters are needed without any outside knowledge. The flip side emphasizes teaching the solver new concepts, showing off the constructor’s areas of expertise, and prioritizing fresh and less familiar entries.
Context is key when determining fairness. I generally side with the idea that puzzles are best when they teach the solver something new. But this impulse is balanced by the understanding that the average person knows much less about any given topic that you are deeply interested in than you do. One way to ensure that difficult crosses feel fair is to make sure their clues draw on different areas of knowledge. At the end of the day, fairness in regard to outside knowledge has a lot to do with knowing your audience, and hopefully in time, your audience coming to know your voice as well.
What does it look like when trust is cultivated properly? I want to give an example from my own solving. If you haven’t done Brooke Husic’s puzzle from Lollapuzzoola 2021, click that link and you can solve it on your own before we discuss it.
When Brooke’s tournament puzzle, the hardest of the day, came around with the warning “What you see is what you are supposed to be seeing,” I felt both fear and excitement. In starting the puzzle, it was immediately apparent that something was up: most of the down answers were gone, replaced with a “loading” symbol. This was, of course, terrifying.
Delving into a hard puzzle can feel, at times, like drowning. When you’ve read a dozen clues and don’t have a single letter in the grid, things can start to feel hopeless. It’s in this moment of darkness that trust becomes critical. Without trust, a solver may well simply give up. Even if they don’t give up, it’s likely the entire experience will begin to feel sour. But when a constructor has cultivated trust in their solvers, the floundering is tempered by the knowledge that a solution can, and will come with time. This bit of faith from a solver is what enables them to soldier on when there’s no answer in sight.
Brooke’s puzzle was terrifying. But it was terrifying in the same way a good horror film is, and for the same reason: because I was never in any real danger. I knew that, if I could keep calm and engage my brain, I’d be rewarded with the answer. And it wasn’t fast, and it wasn’t easy, but I did figure it out, and it was immensely satisfying.
Brooke is a constructor in whom I have tremendous trust. I’ve solved a lot of her puzzles, and so I’ve come to know what to expect: some of Brooke’s blog puzzles are extremely hard, and Brooke’s voice is definitely unique. But Brooke is consistent, and fair, and even though I’ve been stumped by her puzzles in the past (including recently at this year’s Lollapuzzoola), I’ve never had my time wasted.
I’m not here to advocate for only easy puzzles. I appreciate the constructor’s desire to be devilish, to provide a real challenge for solvers and to see them tear their hair out a little bit. But I think some constructors, especially newer constructors, jump right to the idea of trying to make their puzzles as hard as possible without thinking about what makes a difficult puzzle good.
I’d encourage every constructor, whether they make American-style puzzles, cryptics, or any other kind of puzzle, to start by making sure their puzzles are consistent, fair, and above all, solvable. This may, at first, result in some easier puzzles. But it allows for trust to be built between the constructor and their audience. And it allows difficulty to emerge, over time, as a function of the constructor’s experience: not from setting out to be as hard as possible right away, but from realizing the most fair way to mislead the solver, the best way to put the answer right in front of their nose while keeping it seemingly out of reach. To me, that’s what solvers really want in a challenging puzzle, and ultimately I think it’s a more rewarding way to provide difficulty as a constructor.
I believe deeply in constructing crosswords with care. It’s worth it to spend the extra time sanding off rough edges and cleaning up little mistakes. It’s worth it to center the solver’s experience when making a puzzle, even when that means discarding ideas that seem fun. It’s worth it because in the long run, the solver will feel the effects of this work. Whether they can even put a name to it, there will be a bond of trust that grows with each positive solve. This trust is what keeps solvers around. And without solvers, a puzzle is nothing at all.