Image of the MIT Mystery Hunt Closing Ceremonies with permission from photographer Chris Ball
“A dim witted love god.”
I was gazing at the dense, tall pine trees around us, a refreshing change from the dry brown and yellow landscape we had already driven past. My wife and I, both Boston natives, were driving south from San Francisco for a wedding, and entertaining ourselves with one of our regular puzzle games. The first person provides a simple description, and the other must answer in the form of a rhyming adjective and noun pairing.
“Stupid Cupid,” I stated rather than asking, confident in my answer. It’s not a tough game, especially when you’ve played it together before as much as we have. That was in September of last year, and that drive inspired us to evolve our casual game into a much more challenging form: a puzzle for the 2012 MIT Mystery Hunt.
Last year our team Codex won the 2011 Hunt, which is held in January over Martin Luther King, Jr. Day weekend. It’s a team-based puzzle solving competition that draws over a thousand diverse fans every year. The victors’ prizes are well-earned respect, and the responsibility of writing and organizing the following year’s Hunt. Each Hunt has a theme, ostensibly to provide a reason for solving all the puzzles. 2011’s Hunt led by the team Metaphysical Plant, had a theme centered around video games. For 2012, Codex chose to focus on musical theater, specifically The Producers.
For the past eight years I competed in the Hunt and even wrote a handful of puzzles for friends, but none had the level of complexity and polish usually found during the Hunt. Every long-time Hunter has a list of puzzle ideas they would like to write someday if they given the opportunity. Translating those ideas into over a hundred working, solvable puzzles takes many thousands of man hours. As our team quickly recognized, years of solving puzzles doesn’t immediately translate to creating puzzles and organizing a live event for hundreds of people. Thankfully, Codex’s team of leaders and editors provided a framework for both novice and experienced writers to participate in the process.
Using software that is handed down from one winning team to the next (with additions and refinements), writers on Codex started by submitting a puzzle idea. If the idea had potential, the head editors assigned a specific editor to oversee the writing process. To be considered, a Mystery Hunt puzzle idea needed to be more than a concept, rising above “I want to write a Doctor Who puzzle.” Instead, the writer needed to describe the process solvers need to follow in said puzzle. The process could involve a single step or multiple parts before ending with a single answer. Once the editor approved of a working concept, the writer(s) developed a full draft.
When my wife and I submitted our initial idea for a puzzle based on our rhyming game, our editor let us know that it was too simple, and left too many possible solutions for each clue phrase. For several weeks we discussed how to make a cleaner and more complex puzzle without abandoning the rhyming template. Eventually, we thought of using eye rhymes–words that look like they should rhyme but when spoken aloud don’t rhyme audibly, often the result of changes in pronunciation over decades or centuries. Just for fun, we’d set the rhymes in bawdy poems and limericks. For an additional layer of complexity, we added references to films that had remakes, like Ocean’s Eleven. The year of the original film and the year of the remade film would provide an ordering mechanism for the eye rhymes. Reading the first letter of each rhyme word in the chronological order would produce our final clue phrase “FOR NERD CASTE / A MATTER OF BAD ____” (clueing an eye-rhyme of caste, the true puzzle answer: TASTE).
Codexians worked on their own time and in regular online meetings throughout the year. But on the first weekend of November, Codex held its first all-team work weekend that we called a “Puzzle Panic.” Those who could make the trip gathered at MIT, while others worked online, remotely. The Puzzle Panic was not just all-hands work time, but a chance to collaborate on writing and testing. Collaborative testing better reflected how teams would approach puzzles during the Hunt, and new puzzle ideas were born and developed rapidly. The Puzzle Panic was such a success that we held two more.
Even at the end of the first Puzzle Panic, I didn’t know that the theme of our Hunt was The Producers. I was one of a dozen people on a reserve team called “blind solvers.” We were intentionally kept in the dark about the structure of the Hunt, so that we could do final test solving of the round puzzles, called meta puzzles. The concern was that knowing some piece of inside information would allow testers to solve these crucial puzzles, where Hunters would be stuck. After the Puzzle Panic, the blind solvers focused their efforts on solving the metas, revealed one after another as they would be in the actual Hunt. We finished testing the meta-puzzles a month later, although we had originally planned to be done in October.
From the first day after the 2011 Mystery Hunt, Codex was in a race against time: we had a hard launch deadline of noon on January 13th, 2012. Every puzzle needed to be written, checked, tested, and formatted for the final website. Additionally, the team needed artwork, software updates, pacing tests, and dozens of other administrative steps. Our progress was visually measured by the “Graph of Doom.” During the last two months, every member of Team Codex received a daily email with the graph showing our progress—visual proof of how far behind schedule we were.
By December, the rhyming puzzle had gone through multiple drafts and multiple rejections. What had started as “too simple” was by then too complex, with too many false leads. The editors didn’t like the use of bawdy poems, because the bawdiness wasn’t a clue and might be confusing. Our poems contained light hints at movies, which only serious movie buffs would have a chance at recognizing. So we scrapped all of the poems and started again. We re-wrote the poems to be about the movies instead of containing hints, and changed our output phrase as well. We also gave the puzzle its final name: “Bad Poetry.”
Every puzzle in the Hunt had to be tested to eliminate mistakes, poor designs, and required leaps of logic that a writer assumes but a solver wouldn’t guess. “Bad Poetry” went to test solving, but returned unsolved—it had more problems. We edited most of the poems to create a consistent rhyme scheme, so solvers knew they needed rhymes in the blanks. We rewrote some to consistently describe the remake versions, instead of sometimes describing the original movie. We ditched one poem because the movie was too generic. We rejected another because a test solver found a second remake. A fact checker (someone who validates the steps of a puzzle while working with the solution) found an alternate pronunciation of an eye-rhyme word that made it rhyme audibly, and we needed to replace that pair. More changes when testers found a different eye rhyme pair than the one we intended.
By January and many additional drafts, we were worried that “Bad Poetry” was going to be dropped from the Hunt in the face of an immovable deadline. We sacrificed what little sleep we were getting for late night revisions, and produced a version that our editors agreed was ready for a final round of test solving. The testers identified the movies, looked up the years . . . but were stuck on the final answer extraction. The editors suggested one final hint to guide solvers in the right direction: a series of blanks at the end of the puzzle to indicate extracting single letters from each rhyme. It passed testing with only days before the Hunt.
Musical Theater provides a fitting analogy for the Hunt: from the opening performance to the closing awards ceremony, the teams saw the show we put on for them, but none of the behind-the-scenes frenzy. As the clock ticked closer to noon on opening day, our coders found a significant bug in the Hunt website; after the opening ceremonies, puzzles were released at 1pm, just as we announced they would be. Backstage, we hustled, swapped costumes, fixed broken props and dealt with new problems on the fly. We styled our closing ceremonies on Sunday as a Tony Awards show. We gave out awards in categories such as “Best Musical” and “Best Puzzle” (as voted by solvers), and performed a final musical number for the crowd. Of course we congratulated the winning team, Manic Sages, on winning the Hunt itself.
I am incredibly proud of what Codex accomplished in putting on the 2012 MIT Mystery Hunt. Personally the experience of writing puzzles for the Mystery Hunt has made me a far better puzzle solver, as I now understand many of the assumptions that a writer can and cannot make. We are looking forward to try our hands at the puzzles that Manic Sages write, perhaps almost as much as we are enjoying having free time again. I don’t know if Codex will win next year, or if we are even ready to write another Hunt yet. But my Mystery Hunt ideas list still resides on my desktop, and somehow it has even more unused puzzle ideas than it did a year ago.