In my last two posts (both titled Of Trits and Tangrams) I discussed the mechanics of a puzzle I created, but I didn’t talk a lot about the design process behind it. When I thought about why I didn’t feel the need to talk about the design process for that particular puzzle, I came to the following realization:

The quality of a puzzle is often unrelated to how interesting its design process is.

Naturally I’m biased, but I think that Of Trits and Tangrams is a fun, compelling puzzle which – despite being somewhat mechanical in nature – is nonetheless enjoyable. I’ve personally found that there’s something strangely satisfying in puzzles where one of the steps is a mundane task, as long as the designer doesn’t make these repetitive processes last too long. In fact, one could argue that Of Trits and Tangrams is one giant exercise in crank-turning; after all, the entire puzzle basically boils down to the following simple steps:

  • Find a bunch of tiles
  • Solve a few tangram puzzles
  • Google some Latin words
  • Look up the alphabetical equivalents for some ternary numbers

The puzzle works as a whole because the solver isn’t doing one sub-task for very long before moving onto the next one.

But wait – I’m getting sidetracked here. The point of this post wasn’t to argue the pros and cons of simple, repetitive tasks in a puzzle. I wanted to talk about the design process!

Well…the hard truth is that the design process for Of Trits and Tangrams wasn’t really all that interesting:

  • Pick a nine-letter word and get the ternary equivalents for its letters
  • Encode the ternary digits into something arbitrary (like colors)
  • Create the decoding mechanism

Since every letter has a ternary equivalent, I could pick literally any word I wanted as the solution. Colors were a somewhat obvious choice to signify different digits without giving away which is which. The idea to use tangram puzzles was clever (if I do say so myself), but let’s be honest here – hunting down a bunch of pieces and then having solvers assemble them is really just a mechanism to extend the amount of time required to solve the puzzle (again, this doesn’t necessarily make it any less fun!)

In a way, once I had come up with basic idea for the puzzle, I was really just “turning the crank” myself in terms of building. There were some design choices required – most notably the implementation of the decoding mechanism – but for the most part, Of Trits and Tangrams essentially wrote itself.

So instead, let’s talk about a puzzle I wrote where the design process is a lot more interesting. I love to play poker in my spare time, so naturally I had to try my hand at designing a poker puzzle. Texas Hold ’em has become so ubiquitous in the last few decades that almost everyone has a basic understanding of at least the hand values. Knowing this, it seemed reasonable to assume that a random team of solvers was highly likely to have at least one teammate with a solid grasp of the game. As such, I felt quite comfortable creating a puzzle where solvers would have to figure out how to beat an opponent at a series of poker hands, given advance knowledge of their cards and the board. In the real world, this would be blatant cheating of course; but when we are talking about fighting back the evil forces of the mythos, it seems trite to worry about something as cosmically insignificant as cheating at cards!

In designing this puzzle, I started with deciding upon a solution and essentially ended up working backwards from there. The solution to the puzzle had to be a nine-letter word to keep in line with the rest of the game. Unfortunately, nine is an awkward number to shoehorn into a typical game of Texas Hold ’em: Each player is dealt TWO cards; there are THREE cards on the flop, there are FOUR rounds of betting, and a total of FIVE cards on the board once the hand is dealt out. Does the number nine fit in anywhere?

Hmm…well, if you have two players in a hand and deal out the entire board, that’s a total of nine cards (two for each player, plus the five-card board). But if we use each card to represent a letter (seems reasonable), that means the puzzle would only consist of a single poker hand – not terribly interesting! Still, I did like the idea of having each card represent a single letter, and fortunately there’s a somewhat obvious way to do so.

Any seasoned solver is familiar with indexing into a word (using a number) to extract a letter. As luck would have it, a standard playing cards has a handy built-in indexing scheme that contains both a word and a number: its suit and its rank, respectively.

As an example, consider the Four of Hearts. Indexing into the suit of the card (HEARTS) by using the rank of the card (4) simply means that we would extract the 4th letter of the word ‘HEARTS‘ – in this case, the letter ‘R‘.

Deciding to use the four suits gave me access to all the letters in those words as my pool from which to craft my solution word:

  • SPADES
  • HEARTS
  • DIAMONDS
  • CLUBS

Luckily, these four words cover a large swath of the alphabet – most of the off-limits letters are used much less frequently than those available to me. Still, this meant that my solution word could not contain any of the following letters: F, G, J, K, Q, V, W, X, Y, Z

As a fun challenge, I decided to impose an additional design restriction on myself by creating the entire puzzle using only a single deck of 52 cards, rather than multiple decks (which would emulate a single deck being shuffled and dealt over several hands). A consequence of this is that I would only be able to use each letter from my card suit letter pool exactly once (or not at all). In other words, my solution word could not have any repeat letters unless those letters appeared multiple times across the four suits.

I ultimately decided on the word PENUMBRAS. You know? Those lighter parts of shadows that appear at their edges (from the Latin paene + umbra, literally: almost shadow). I wanted to design the puzzle such that the word would be spelled out from left to right via indexing into the cards (as described above) once they had been placed in the correct order. With that in mind, I picked out some cards that gave me the letters I needed:

On a broader level, I knew that I wanted the mechanism for solving the puzzle to be assigning pairs cards to predetermined hand scenarios such that the solver would “win” every hand. But with an odd number of letters in my solution word, I was one card short of a pair. The fix was simple – I just added a card to the end, for a total of ten:

This final card – the Seven of Hearts – does not translate to a single letter if we use the familiar indexing scheme. Since the word HEARTS only contains six letters, indexing seven letters has the effect of running us off the end of the word. Put another way, there is no 7th letter in the word HEARTS.

These ten cards represent the solver’s “hole cards” – two cards for each of five separate poker hands. The only thing left to do was to build the hand scenarios using the remaining 42 cards in the deck. I thought this would have been a tough exercise, but it really wasn’t too bad. Admittedly, I may have just gotten lucky:

Is it cheating to know how all the cards will be dealt? (Yes. Yes it is.)

Each column represents a single hand of Texas Hold ’em. The pairs of cards in the top row are the opponent’s hole cards for each hand and the vertically aligned cards represent the board for that hand: the flop, turn, and river – all dealt out. On the bottom row are the spaces for our solver to place their own pairs of hole cards such that they win each hand according to the rules of poker.

See the small black dots in the “Your Cards Here” spaces? Those are to help the solver place the pairs of hole cards in the correct orientation. Remember, upside-down is undefined for playing cards!

Even though you already know the solution to the puzzle, it might be fun to take a moment to work out how each of the five pairs of cards “beats” the corresponding hand. Otherwise, here is the complete explanation:

With the design complete, I had to decide how to present the puzzle to the players. One thing I love to do is to give solvers a tiny piece of the puzzle before giving them the puzzle itself. When I do this, I always try to frame it in a way that evokes curiosity, rather than leading them down a rabbit hole. This works…most of the time.

I decided to grant players access to A High Stakes Game in the midgame of The Madness. I set it up in the second room of the game behind a locked gate and a translucent curtain. By craning their necks and squinting their eyes, players could just barely see an object in the far corner of the room that looked to be…some sort of casino-themed gaming table? Maybe?

The first room of The Madness contained two locked chests. Most teams were generally able to unlock both of these chests within the first 30 minutes of the game; each of them contained a variety of different objects including…a pair of tiny playing cards! So before even gaining access to the room containing the related puzzle, the solvers were given two pairs of playing cards – the purposes of which they could only speculate wildly.

Of course, any puzzle creator worth their salt knows the value of evoking catharsis in their solvers, and I aimed to deliver. Once they unlocked the first gate and peeked behind the curtain, the solvers discovered a bright green felt tabletop, with the other three pairs of cards all sitting nearby. “Ahhh…I get it now!”

That’s the phrase we all want to hear.