As I have done every year since 2004, I spent the second weekend in January playing (or solving, to use the field-specific lingo) in the MIT Mystery Hunt. I always feel quite privileged to play; each hunt iteration represents a one-time-only interactive artwork that a team of passionate amateurs spends the better part of a year planning and constructing, culminating in a single weekend where a thousand puzzle-hungry solvers trample through it.
Like an informational World’s Fair, it leaves its husk behind for the late-but-curious to tour: you may browse all this year’s puzzles online, and note that they seem to be arranged around a theme of ill-advised Broadway mashups. Without the context of the hunt alive around them, though, the puzzles lose a certain amount of motive force. When presented all at once like this, they lack the light but necessary hunt-specific narrative that organizes paths for the solver to follow. (This year, it featured a storyline based on the further adventures of the swindling showmen from The Producers.)
I would also argue that, even though each puzzle now links to its own solution page, these puzzles must still seem impossibly obscure to curious layfolk who stumble upon them. So in this article, rather than examine the hunt’s overall form where carefully paced groups of puzzle-sets slowly reveal the twisty superstructures of meta-puzzles, I’d like to highlight a few of the several dozen individual challenges which defined the weekend for the hunt’s players.
Let’s start with the puzzle titled Slash Fiction, designed by (and starring) Seth Schoen and Vera Yin. It makes a nice blog-post headliner because it happens to take the form of a six-minute video, one as fun to watch as to solve.
Have you watched it? All right, then: your challenge, as with every hunt puzzle, is to somehow definitively produce an English word or phrase based on this input.
A novice solver might start searching through the video to see if the answer word flashes by at some point, or might forgo analysis entirely to simply guess answers like
PARIS. A more experienced solver knows that well-constructed puzzles avoid expressing any information that don’t contribute to their own answers. This player will take the more fruitful route of examining the video shot-by-shot, probably creating their own table of contents for the video on paper or in a spreadsheet. They would then proceed to look for interesting patterns, and move forward from there.
In this case, the first level of pattern that emerges is that the video is broken neatly into groups of three shots: first the gentleman-mime does something in close-up, followed by an oddly minimal intertitle of “/” (or sometimes “/x-“), after which the lady-mime does something else in a sunny urban setting. After a flash of static, the pattern repeats. The video contains
twelve twenty of these static-separated scene-triplets, and nothing else (beyond a delightful accordion-and-piano soundtrack). [Thanks to Seth Schoen for the correction.]
Having built a neat three-by-twenty table describing the scenes, a seasoned solver now has reason to suspect that each such triplet corresponds to a single alphabetical letter, and that these letters should, when ordered properly, spell out an answer. This may seem like quite a logical leap to the uninitiated! However, encoding letters into distinctly non-alphanumeric stimuli (such as video snippets of capering mimes) represents a very common design pattern in hunt-style puzzles, and solvers who’ve played through a hunt or two learn to recognize its signs. The solver’s ability to organize the puzzle’s surface-data into a regular table is a strong indicator that this puzzle involves a string of encoded letters, one letter per table-row.
So now, there is the simple matter of working out those letters, as well as the order in which to arrange them. This particular puzzle calls for further pattern recognition regarding the landmarks which the lady-mime cavorts around, as well the significance of the mimes’ antics paired with those odd intertitles. The willingness to perfom a bit of web-based research helps in both cases. I invite the reader to try working out the answer, perhaps with a friend or two. You can find the answer, along with a full explanation of its derivation, on the puzzle’s solution page.
It happens that I didn’t get to work on Slash Fiction during the hunt, sadly. I, a single player from a team of around 40 solvers, see only a few puzzles on any given hunt weekend. A typical puzzle can take hours to work through, while the hunt only lasts two or three days — and unlike some of my harder-core teammates, I like to leave campus at night to sleep in my own bed. Before calling a taxi home on Friday night, however, I had the pleasure of solving most of Andrew Lin, Elan Pavlov, and Jit Hin Tan’s The Undiscovered Underground, a puzzle which, while suffering from broken design, worked well enough to lead two teammates and I to build a very satisfying logical bridge, and follow it up with an unexpected and rewarding bout of real-world exploration.
As this puzzle is literally unsolvable by people who don’t happen to have immediate access to the MIT campus, I shall speak less coyly about its construction. The puzzle presents you with a faux text adventure. While its title and opening paragraph both directly quote a real game, the rest of the text is only Infocom-esque pastiche, with no real next parser; the player may only key in compass directions to move around the game’s map, unable to interact with anything its text mentions.
Clearly, the first step to cracking this one involves drawing that map on paper. A wise solver will follow up by marking the map with the locations of all the conspicuously repeating props and phrases that appear throughout the text, such as mentions of carrier pigeons or the message “you can rest here”. Steps three through ten, for my two co-solvers and I, involved turning this information around every which way. My friends experimented with folding the map into some sort of three-dimensional object — maybe you were traversing the inside of an person’s body, somehow? — while I played through the actual Undiscovered Underground interactive Fiction game, looking for similarities with the puzzle text.
After an hour or two of this, we took a break to compare notes and bat ideas around out loud. This is when I hit on a breakthrough: one room’s description of three portals leading to vertical shafts, one strangely giant-sized, reminded me of the elevator doors found on every floor of the very building we all sat in, which included an extra-large freight elevator. So if mentions of “vertical shafts” referred to real-life elevators, then the “carrier pigeons” might be… mailboxes? On a hunch, we called over another teammate, an MIT alumnus intimately familiar with the university’s famous underground tunnel system. “Oh yeah,” he said, as soon as he saw our drawn-out game map. “This part here’s under Lobby 7, and over here would be underneath Building 26…”
After getting directions to the nearest mapped intersection our teammate recognized, the three of us bolted down the stairs and into the tunnels. Very quickly, and to our profound delight, we saw we were right: the “pigeons” were indeed FedEx drop boxes, while “you can rest” statements pointed to nearby bathrooms. Other notable features of various tunnel locations all appeared in the game text, though similarly obscured into the language of fantasy role-playing quests. Most rewardingly, every place the text mentioned a pile of coins, there existed in reality a plaque with a you-are-here map of the whole tunnel system. Each of the 20 plaques we found prominently featured a single, large, unique letter. Score! Surely, we were steps away from the answer now. Once we completed our unexpected guided tour through MIT’s bowels, we rushed back to the our team’s headquarters to wrap the puzzle up.
Unfortunately, our team progressed no further with this puzzle. Unlike an elegantly constructed puzzle like Slash Fiction, The Undiscovered Underground presents far too much information. We had connections between these screenfuls of bogus text-adventure text and letters, but no idea how to shake an answer out of them, because there were so many things we could do. And the connection with the actual Infocom game continued to nag at us — why would the fake game’s text extensively quote the real game, if it wasn’t important somehow? Before we gave up, we pursued the possibility that the game’s exact wording about whether you picked up the coins you found clued whether you should “pick up” the associated letter and add it to the answer, but that went nowhere.
As the puzzle’s solution page states, the real answer involves taking note of features in the map that weren’t reflected in the actual tunnels, and using only those locations’ letters to spell out the answer. Sadly, this wouldn’t have worked for us: in at least one case, I was able to connect one of these extraneous features (“a temple to the goddess of wisdom” — a reference to Athena, MIT’s campus computer network) to an object at that location the designers apparently hadn’t noticed (a locked but clearly labeled network-utility closet). The following afternoon, during the hosting team’s wrap-up presentation, The Undiscovered Underground received some catcalls and boos, so I suspect we were not the only players who fell victim to a puzzle that asked its solvers to not pattern-match as much as they could.
In closing, I can link with pride to Andrew Lin’s Any Old Puzzle, the only puzzle I managed to definitively co-solve all weekend, right up to confirming the correct answer. I suspect, alas, that it was also one of the easier ones (though not the most minimal). If you can wring the right word out of this one, you’re at least as good a solver as I.