tl;dr: I spent the last week helping organize the 2013 MIT Mystery Hunt; this year’s Hunt broke the record for the longest hunt ever, lasting 73 hours instead of the expected 48; congrats to the team whose name is the entire text of Atlas Shrugged (who we referred to as John Galt) for winning, condolences to them as well for having to do this next year; learning to appreciate bits of joy and hilarity in a sea of crushing fail. Also, check out this post for some sample Hunt puzzles and an as-yet-unseen puzzle written but not used for Hunt.
(warning: effin’ gigantic post)
(other warning: all of the below is my half-baked opinion based on incomplete information)
The Hunt We Wanted
If you aren’t familiar with the MIT Mystery Hunt, it’s a weekend-long puzzle-solving competition that’s been held at MIT for over thirty years. Each year the Hunt has a different theme or plot of some kind; last year’s Hunt was Broadway musical themed (particularly The Producers), and the year before that was video game themed, where you were Mario trying to rescue Princess Peach. Every year, to win the Hunt you must find the coin: a usually (but not always) coin-shaped object hidden on MIT’s campus that requires solving tons of puzzles and engaging in a scavenger hunt-esque “runaround” in order to find.
I’d heard of the Mystery Hunt years ago, as I had several East Coast friends and acquaintances who participated. Last year was my first time attending the Hunt, though, with a team called the Manic Sages. Since we won last year, we had the dubious privilege of organizing and writing this year’s Hunt. I volunteered in a few different capacities–writing and testsolving a few puzzles, hammering out the hunt theme, serving as the Art Director, and working closely with the tech team on this year’s Hunt website.
This year’s Hunt was heist themed. The premise:
The Mystery Hunt has run into some financial trouble. The Manic Sages took a loan from Enigma Valley Investment and Loan, and through some dodgy paperwork, the bank ended up owning the Mystery Hunt outright. Enigma Valley decided that the coin was a valuable financial instrument, so through various Coin Default Swaps they leveraged a trillion dollars on the bet that nobody would find the coin. To be sure of it, they locked the coin in a high-security vault in an undisclosed location, with security systems designed by some of the world’s finest minds.
MIT legend Alyssa P. Hacker is not happy about this state of affairs. She wants to save the Mystery Hunt, by legal means or no. Thus, she hacks into Enigma Valley’s servers to try and find information on the bank’s security systems. She also enlists IHTFP, a heist consultancy firm, to help us organize and pull off a heist to steal the coin, bring down Enigma Valley, and save the Hunt.
The Hunt was structured with the opening round on enigmavalley.com, which identified the security systems used by the bank and the experts who designed them: Danny Ocean, Richard Feynman, Maxwell Smart (later replaced by Agent 99), Indiana Jones, Marty Bishop, and Erno Rubik.
Then there were six normal rounds on coinheist.com, in which teams recruit the six experts to their heist team by finding reasons for them to turn against Enigma Valley.
When a team solved the round and recruited a new member, they would get to work through a practice obstacle, constructed by IHTFP, associated with that person. The six security system obstacles in the bank were guards, a safe, a laser maze, a puzzle box, an impenetrable door (so teams had to use ventilation shafts to get around it), and a keypad.
After solving the final round, Alyssa P. Hacker would congratulate teams, but before allowing the heist to go forward, insist they come up with a reason why they are motivated to steal the coin from the bank, so she can trust them. At the same time, teams would get an phone call with an automated recording from Enigma Valley, telling them about “some puzzling activity” on their account. The phone call would give teams the link to a final puzzle, styled like a list of security questions–sixty security questions! Ultimately, the form would never allow you to unlock your account, but you could extract the answer: ENIGMA VALLEY GIVES YOU THE RUNAROUND.
The runaround then consists of solving each of the six real obstacles, each of which is slightly harder than they were in the practice versions, due to the bank beefing up security. A corporate newsletter smuggled out by a sympathetic member of the bank gives teams information to figure out what has changed about each obstacle and how to get around the changes. Finally, teams get into the vault–and are faced with an unexpected seventh obstacle, where the coin is hidden in a bunch of “counterfeit” coins, and teams have to use a scale to determine which is the real coin before the guards come back to the safe.
The Hunt We Had
Exhaustion–physical, mental, emotional.
Not just that this year set a record for the longest hunt (73 hours long), even with us handing out free answers like candy and cutting a metapuzzle towards the end. Heck, technically we even cut the runaround, since we declared a winner before the winning team completed it (less than an hour before the wrap-up ceremony). But for me (and most Sages), Hunt started long before Friday. Hunt had been distracting me from work for months. The week before the Hunt, all I did was sleep, eat, and work on the Mystery Hunt. I barely saw the sun.
Honestly, the predominant emotion I feel right now about this year’s hunt is despair. The Hunt starting an hour behind schedule, cutting so much of the hunt material in desperate bids to end the Hunt on time, having the Hunt go until Monday afternoon–I felt like we had failed. We abused teams’ time and sanity, letting the Hunt go into Monday. There was so much cool stuff that people worked their asses off for that no one–or only John Galt (the winning team) and Palindrome (the only other team to get to the runaround)–ever got to see: the security questions puzzle, the runaround traps, the coin design, even the finished hunt home page, all effectively for naught! On Monday I laid on the floor crying in the tech room for a while, not really able to productively handle it. I skipped all of the post-hunt events except wrap-up, mainly because I couldn’t deal with being around hunt people anymore. It was all just a reminder of how badly we–I–had fucked up.
Objectively speaking, I was the art director, and art and front-end development actually went basically fine, despite miscommunications and last minute feature requests. My puzzles worked well and I got some positive feedback on them. I was a new Sage, and didn’t have authority either in terms of seniority or in terms of being a puzzle expert, so it’s not like I had grounds to have opinions about the length and structure of the Hunt. But I didn’t–don’t–feel responsible for just my little circles of contribution. I was emotionally invested in the entire Hunt, and have had difficulty recalibrating that.
When my warnings from months ago to our tech lead that writing our own brand new hunt website software was a mistake he was too intelligent to make came to fruition, I wasn’t at all happy to be right. I stayed up until 4 AM Thursday night hoping that I could somehow help fix the backend tech, and feeling completely fucking helpless that I couldn’t usefully contribute. The two main website tech volunteers hacked something like 40 hours straight, and even with that herculean effort we still had to delay the start of Hunt an hour.
After the rush up to hunt, and the tech embarrassment, I had been hopeful that the rest of Hunt would be relatively relaxing. Maybe I’d be called upon to fix minor front-end tweaks, but fundamentally once the Hunt site design is done, it’s done, and it’s unlikely to explode in sudden unexpected ways. But then we noticed that the teams were progressing much more slowly than we (and our automated projections) ever expected. Round after round unlocked based on the time-based side of our puzzle release mechanism, but nobody was completing the rounds. We’d intended the opening round to be so easy that every active team would solve it, yet even by Saturday many teams hadn’t gotten it.
By the time we realized just how badly things were going, there was little we could do about it. From Saturday afternoon through Sunday, HQ was full of the thousand-yard stares of sleep-deprived, overworked people, all wondering what on earth we could possibly do to save the Hunt.
We couldn’t cut a full round for many reasons–one, the plot already had indicated you needed to recruit six people, and two, the last round’s leader (and the people whose puzzles were in that round) would kill us if we didn’t include it. We would’ve needed to cut it before the Hunt even started, if such a decision were politically feasible even then. If we’d had better projections of how long our hunt would take to solve, and had known late Friday/early Saturday that the damn thing would go into Tuesday without intervention, there was only one thing we could have done at that point: hand out a utterly absurd number of options (free answers). Explicitly restructure the Hunt so teams could choose their favorite 90 or so puzzles out of the hunt, solve them, and use options for the rest. As it happens, that’s what we ended up doing, but we didn’t realize we were doing it at the time, and we hesitated too long for it to make a radical difference.
(The other change we could have made at that point would be to make it obvious in the as-yet-unlocked rounds which answers contributed to which metapuzzles, which would help teams be more strategic. But that would have made much less significant a difference than the above.)
It seems fairly clear why we failed.
- First, last year’s hunt was around 100 puzzles long, and it ended early. We heard lots of whining about the early end. So we thought that this hunt needed more puzzles in order to last until Sunday noon. In the early Hunt structure conversations, some individuals vociferously argued that the Hunt needed to be around 180-200 puzzles long, or else it wouldn’t be long enough. Arguably, this shifted the Sages HQ Overton Window in terms of what we thought the Hunt required–people retorted that 135 was enough, not that we needed to *lower* the number back to 100.
- Second, the team was too self-deprecating. We thought that the average Manic Sage was a below-average Mystery Hunt participant, and that the only reason we won last year was because of sheer manpower. So if a typical testsolving group only barely managed to make it through a puzzle, after a few attempts, it went in, with the thought that it would be easier for a “real” team. If a group of ordinary humans couldn’t solve the puzzle, but one of our best solvers could, it went in, hamstrung also by the belief that since most of our best solvers were spoiled on many puzzles due to being writers and editors, the sampling wasn’t realistic. If a meta was solvable by our best puzzler with only half the answers, or no answers at all, we dialed up the difficulty. For the most part, we didn’t ship broken puzzles–certainly not more than other Hunts. Every puzzle we used was solved in testsolving. We just thought they were easy puzzles when they really, really weren’t. We now know that the average Manic Sage is a slightly-above-average puzzler, and our best puzzle-solvers are as good as anyone else’s. Had we believed this before, it would have improved our estimates quite a bit.
- Third, as stated earlier, the tech team succumbed to Not-Invented-Here syndrome. The new system was designed to handle lots of connections without a lot of load, and it did succeed at that. Where last year’s load average was around 900 at the start of Hunt, ours was 0.1. We could have run the Hunt on a Raspberry Pi if we’d wanted. That part went great. But the brilliant new architecture was arcane, so when bugs appeared hardly anyone could do anything about it.
This was a far less impactful failure, in the end, though; it “only” cost us an hour in starting Hunt, various broken links that got fixed over time, some other useful but not crucial site features for showing teams their previous answer submissions and other status information, and the tech team’s health and sanity. On Friday I remember team members worrying that we looked unprofessional, that this hunt would be remembered as sloppy. If we’d only known how right we were…
It is difficult for me to focus on the positive. It feels like everybody hated this Hunt, and I seem to be thoroughly enveloped in post-Hunt depression. But there were some successes and beautiful, hilarious moments.
Codex’s mock, Occupy-style protest at the opening ceremony filled me with glee. The only thing Codex knew (like every other team) was that there was a bank involved with this year’s hunt theme. Yet their protest turned out to be super thematic! It was actually a little *too* realistic–I had to run interference between the security guard at the opening ceremony and Codex to assure the former that it was just a prank, not a real protest, and plead with the latter to avoid alarming any cops when they left! Even with that, afterward Lobby 7 had a heavy police presence.
Scribbling on the opening ceremony bank slides as Alyssa P. Hacker: lots of fun. :D
Half the hunt website frontend–including the entirety of the Rubik round–got designed and built during the week before hunt. And it looked fucking awesome, if I say so myself–Diana’s crazy-detailed casino art, Alice’s world map calligraphy, Rob’s revisions to the Sneakers round design, DD’s shoephones and the rest of the Get Smart design, and Sean’s manipulable, browser-compatible 3D Rubik’s cube… Plus the recruit cartoons on the hunt home page and the evil bank’s logo and brand identity. Even if we had to take away a few features due to the techpocalypse, I am seriously proud of this hunt’s art and front-end design and glad I had the chance to take the lead on it.
The bank security system obstacles were a ton of work to construct, but totally fucking cool. I only wish everyone had gotten to see both versions! I was especially giddy about the “twist” in the Feynman bank obstacle, which was a sentence Blake came up with months and months ago: “DIP IT IN ICE WATER”. So we constructed a safe puzzle where the demo safe’s combo was HOT LAVA, and the real, runaround safe’s combination was OBSIDIAN.
Also, we had not one, but TWO real, working laser mazes. The second maze’s solution? CLAP OFF. Yeah.
No apologies necessary for the obstacles portion of the Hunt; they were pure fucking awesome. Darby, Sparky, and all the Sages who helped tape cardboard: you guys rule.
The coin. Good God the coin was sexy this year, even if we forgot to show it to teams at the closing ceremony. DD drew the beaver and suitcase; the rest of the design was mine. Our production lead Sparky, and the company that 3D printed the coins, did a fantastic job–the company even added fur texture to the beaver!
One bright spot midway through the hunt was Heist Test Kitchen, a puzzle wherein teams had to cook a dish containing certain ingredients (the ingredients varying depending on their team composition) and bring it to HQ. Most teams’ dishes were quite delicious–even the one that used rice cakes and Doritos as ingredients, shockingly–and between the various items it meant that we didn’t have to cook much of a separate team dinner for Saturday night! A+ puzzle, would include again. :)
Events: The decorations for Casino Lobdell looked amazing, and the pictures for the Thomas Crowne event also came out fantastic. And, as in that one psychology test, at Thomas Crowne we had a guy in a gorilla suit that only half the participants there even noticed (their focus was on watching the bag handoffs)! Even the non-hunting journalist videotaping the event thought we were joking or lying about the gorilla–”Really! Play back your video!”
We got quite a few amusing submissions to the hunt’s answer queue:
- Here is a list of people you don’t want on a heist (aka wrong answers we got for the opening round):
- MARTY MCFLY / SEAMUS MCFLY / CALVIN KLEIN (lot of guesses along these lines for Marty *Bishop*’s puzzle)
- FANTASTIC MR. FOX
- RUBE GOLDBERG
- FELIX BAUMGARTNER
- ZEFRAM COCHRANE
- KING OF FRANCE
- GLORY HOLE (what?)
- For Sages Style, Providence Transplantations submitted HOMEOMORPHICALLY IRREDUCIBLE TREES (the real answer was GRAPHS)
- For Ex Post Facto, Luck submitted HALF SOLVED (which is an instruction you get partway through the puzzle). After we told them that was incorrect, they tried SOL. *facepalm*
- Multiple teams submitted WONDERPIG for a puzzle whose correct answer is WONDERING. I get you don’t have all the letters, but why would you guess that *first*?
- Strange New Universe tried BLESSED IS THE DETERMINANT for the Feynman metapuzzle. Would have been a great answer!
- For my email puzzle, Random Thymes tried QUIT AND RENAME THE DIR. Nope.
- For Tuva or Bust, Super Team Awesome tried BUST. They also tried RUSSIA, RUSSIAN FEDERATION, and about half the dictionary in a fit of frustration. The real (unfortunate) answer, that they had all along? RENAMES.
- Luck submitted “best liquor for puzzle solving”. They immediately followed up with “Sorry, thought this was Google”. Neither of these was an answer to any puzzles in this year’s Hunt.
- There was a puzzle called “Time Conundrum”, as part of which we had to call teams and tell them that their answer to the puzzle, RED HERRING, was incorrect two minutes *before* they unlocked the puzzle. (Most teams found this merely confusing; one actually tried to argue with the caller that they hadn’t called any such thing in!) After the puzzle was unlocked, a few teams called in RED HERRING. We responded, “Don’t call in repeat answers.” They were insistent, though: “We know it’s wrong; we just wanted to prevent paradoxes.”
- John Galt tried EWE for the puzzle I Can’t Spell You.
- John Galt also called in THEREARETWOPEOPLEIN26-100HAVINGSEXRIGHTNOW. This turned out to be correct, but not for the Highlight Reel puzzle. Folks, this is not what the answer queue is for…
There were also a variety of amusing HQ moments not related to answers…
- One of the oldest teams in the Hunt, Setec Astronomy, used to win rather frequently. As a result of being forced to write so many hunts, the team now deliberately does their best to not win, usually by applying liberal amounts of alcohol if signs indicate that they are near the top of the pack. Early in the hunt–I think late Friday–Setec Astronomy solved the opening round meta. This put them in the lead for a while, both by total puzzles and total rounds solved. Upon yelling “SETEC is in the lead!” in HQ, we resolved to never ever inform them of this. :)
- When Catherine overheard the tech team talking about how the automatic time unlock feature wasn’t working (but we could still manually trigger it), she exclaimed, “What do you mean we have to hit a button every two hours or the hunt explodes? That has to get fixed. This isn’t Lost!”
- HQ ran out of plastic forks on Sunday, which worried the kitchen crew:
Veronica: This is the last fork. Soon we’ll run out of food and we’ll have to start eating each other.
Tom: Worse, we’ll have to eat each other with our hands!
- At one point in the Hunt, Random Thymes had only gotten one of the three wheels to their Enigma machine for the Sneakers metapuzzle, but thought that they could crack it anyway. One of their members had access to a supercomputer in Ohio, so they plugged the puzzle into it. As they told us, “If the British can do it with a Turing machine, we can do it with 8000 cores.” Sadly (?) the team completed the other parts of the round and had all three wheels by the time they solved the puzzle, so we will never know if this meta could have been brute-forced!
- The second place team, Palindrome, was unhappy that their number of options was uneven (they had 32 options, which at that point in the Hunt would never amount to anything, since free answers cost multiples of 100). So, they contacted HQ and offered to do something entertainingly humiliating for 68 options. This resulted. We gave ‘em the points. :)
- Late at night on Sunday, a notice popped up in the contact HQ queue from the team Immoral, Illegal, and Fattening. Its text? “um, we set our metapuzzle on fire…”.
While basically all of HQ was incapacitated with laughter for several minutes, the team also sent an unrelated request along with “(also we are sorry about the flaming metapuzzle still)”. We finally managed to call them back, and they answered the phone: “Hello, this is Immoral, Illegal, and Flammable!”
As the team explained, they thought the piece of paper the Get Smart metapuzzle was on was treated with the kind of invisible ink that appears if you apply heat to the paper (rather than the kind that appears if you apply vinegar, which was in fact the case). To test this hypothesis, instead of using an iron set to low, like a sane person, or a candle flame, like a less sane but more understandable person, they decided to put the paper inside a toaster.
This was pretty much the “As God is my witness, I thought turkeys could fly” moment of the Hunt.
Fortunately, we had plenty of duplicates — we had made 100 copies of the metapuzzle, but ended up using only four of them, including the burnt one.
A number of hunt participants tweeted great things using the hashtag #mysteryhunt. I especially enjoyed the retweets from the (fictional) @manicsages account. Most of them are compiled in the Tech’s Storify. I’m glad I didn’t see these during Hunt, though, as a couple of them are rather disheartening, but I can appreciate them now. I liked Eric Berlin’s hunt wrap-up post, as well. There are other blog posts reviewing our hunt in much more disgruntled terms… but, right, this is the positive section.
Perhaps most of all…
I don’t know yet if I’ll be back for Hunt next year. I was never a very good puzzler, and maybe we’re now the Depressed Sages? Got a whole year to recover, I suppose.
I am eagerly looking forward to once again having a life outside of the Hunt, in any case.