MIT Mystery Hunt 2020 Recap

Posted on January 27, 2020

I participated in the MIT Mystery Hunt last week with ✈︎✈︎✈︎ Galactic Trendsetters ✈︎✈︎✈︎. This was my fourth time hunting with them, and my last time as an undergraduate (because I'm graduating, not because I'm quitting). I had considered making a post like this in previous years, but never bothered to do follow through; since we actually won Mystery Hunt this time, I figured that this year was as good as any.
I'm not going to add links to any of the puzzles here because they're all currently available on, but will eventually be moved to the Mystery Hunt Archive. Also, keep in mind that I'm a novice puzzlehunter, in addition to being a bad writer, so take my commentary with a grain of salt. Lastly, and most obviously, lots of puzzle spoilers below. Read at your own risk.

First off, thanks to Left Out for running such a smooth hunt; it looked like things were going to get rocky after we got the first couple of emails restricting overnight hunting, and I'm very content about how the policy ended up. The actual wedding during kickoff was completely unexpected for me; I felt pretty weird attending a wedding for people I didn't know, but props to them for actually having the courage to get up there. I appreciated the references to past hunts, but most of them flew over my head.

Also, thanks to whoever on Left Out that included a json API endpoint /js/puzzles for retrieving the team's currently unlocked puzzles. As the person on ✈︎✈︎✈︎ Galactic Trendsetters ✈︎✈︎✈︎ in charge of the puzzle scraping script, I was pretty alarmed when selenium refused to work properly when scraping the normal puzzle page at /puzzles (for some reason, it thought that all the links on the page had no link text, and I really needed those texts to populate the puzzle name field).

Now with all that out of the way, onto the individual puzzles:

Stress Test—Cute little puzzle about stressed syllables. Someone else made the aha and found the first pair, and then I found the COMbat (warfare) and comBAT (resist) pairing. The puzzle progressed quickly from there.
Checkerboard—I spent way too long trying to investigate the Wizard's Checkers red herring mentioned in the flavortext, even though it only appeared in a couple seconds of Fantastic Beasts and Where to Find Them in a tiny newspaper column. Should have instead thought more about the "don't be a dick" part, which was cluing Richard Nixon.
Monty Minotaur's Magical Menagerie—I'm glad that word search solvers exist, and I'm glad that other people know how to use them. Learned about what a ushi-oni was, correctly guessing the ox half (from ushi) and being pleasantly (?) horrified at the spider half.
Birds—Seemed like a straightforward puzzle, but we had a ton of trouble with extraction, eventually putting it on hold until it was backsolved. We found the bird song page and knew we were supposed to index into the bird songs, but we assumed that the index was for either letter or word (or hyphen-separated syllable). We also were confused about if the first number referred to song number or line number.
The Scrambler—I just really like this type of parallelizable puzzle. Also the fact that it mentioned McDonald's. Good stuff.
Food Court—I didn't even work on this puzzle; I just found it hilarious that over 40 people were looking at this puzzle's sheet a couple minutes after it was unlocked and tagged as a math puzzle.
Story—We actually found the numbers in the page source before we finished extracting the first cluephrase from the printer's devilry-ish part. While we were in the middle of extracting the second cluephrase, and figured out that the animal was pika, I actually joked that perhaps this puzzle had to do with the MIT living group. Then we found out we actually had to do that after extracting the second sentence. I imagine a puzzler who was more familiar with pika could have skipped the first two parts entirely, especially since there are some very specific rooms mentioned (e.g. electric bluegaloo).
The Aerialist—Obviously this was a semaphore puzzle. Got through half the video, noting that it was hard to figure out when I was supposed to transcribe the letter, before someone told me I needed to have audio on. Oops. I then rewatched the video properly, transcribed a bunch of seemingly-gibberish letters, and left the actual extraction for someone else.
Hall of Mystery—I only started looking at this puzzle when it was stuck at extraction. I thought the flavortext here was more confusing than helpful; we spent a lot of time trying to overlap the grids so that the pairs were directly touching, which obviously didn't work.
Dragon—I only worked on level 3 of this puzzle. It was pretty comfortable to control the MMO holy trinity of tank/healer/DPS, but I feel like the tank and healer skillsets were a bit lacking. Namely, the tank's rotation was incredibly simple (taunt, attack or refresh DoTs 4x, repeat), and the healer had practically infinite mana and no offensive abilities, and so just spammed heals all the time. Dodging lasers also became quite easy after I realized I could just hug the boss or run far away at the start and avoid every vertical laser. The three textboxes as input was quite frustrating to deal with, especially when I wanted to adjust just a move that happened on a specific turn. This was exacerbated by the fact that the time dragon could cause the move order to jump around non—linearly. Nevertheless, I had a lot of fun working on this puzzle. That being said, I wouldn't play this as a standalone game, so I guess my standards for "game in a puzzlehunt" are much lower than my standards are for games in general. Does that mean that putting any okay video game with remotely puzzle-like elements into a hunt would increase my enjoyment of said game? ... Maybe, but that raises more uncomfortable questions, so I'll just cut the train of thought off there.
Horse—Incredible puzzle... how was this constructed?! We initially got stuck on this puzzle because we made the assumption that duplicated words would show up twice in the enumations in the bottom, so then Akroan Horse was the only valid horse to use since the other horses (Bronze Horse and Zodiac Horse) did not have their enumerations show up twice. Using this assumption resulted in contradictions that would only become apparent at the very end of the puzzle, so by the time we realized our approach didn't work, we were too burnt out to try anything else. This is the first puzzle where backsolving it helped its forward solve. In particular, we got GRAVECALLER as one of its answers from a backsolve, and then started working from right to left. This was relatively simple with only one Gravecaller in Magic (Rakshasa Gravecaller). This helped reinforce the fact that we needed to get rid of our original assumption, and solving for the remaining two answers was straightforward (I didn't really help with the logic puzzle stuff from this point on, and only made sure that everyone else was searching Scryfall efficiently).
Jungle Cruise—This was a silly puzzle. I feel like the Cluekeeper portion of this was shoehorned in, since the puzzle would have worked just as well with a series of videos. I didn't realize that this puzzle was a reference to an actual amusement park attraction (namely, the Jungle Cruise).
Ladies First—Again, similar to The Scrambler, I enjoy these types of parallelizable puzzles. Unfortunately, this one didn't include McDonald's.
Water Guns—Didn't manage to forward solve this during the hunt. I even made the connection between red and Fire, blue and Water, and green and Grass!
Goliath's Goof—Felt like playing Minecraft with finnicky controls. I really should have used my mouse instead of a touchpad for this. The biggest problem with this puzzle was the fact that it would run out of memory and crash after playing for too long; this came up twice on two different computers!
Old West Revue—I saw the hiragana script, called out "Weeb Puzzle!," and then closed the tab. I didn't contribute to the solve.
The Horse Whisperer—Another simple and easy to parallelize puzzle. Was pretty funny hearing the whole room try to pronounce and devoice the silly phrases.
Sheriff's Stars—As far as typeface-related puzzles go, this was a fairly enjoyable one. I still have nightmares about Helvetica Is Only an Okay Font, the Bacon's cipher puzzle from last year's Mystery Hunt that used Arial and Helvetica.
Wanted: Gangs of Six—I opened the sheet, saw that the My Little Pony and Jojo's Bizarre Adventure characters were already identified, and realized that I had nothing to contribute.

Teamwork Time—I only got to work on 3/5 of these, but I really liked their concept, and I hope to see more of these in the future.
Hat Venn-dor—It was possible for a team member to lock out the rest of the team from progressing if they put their letters in the wrong place in the Venn diagram and forgot about the tab. I ended up going to our other room to make an announcement to close the puzzle page if they weren't actively working on the puzzle; I'm not sure if this actually helped, but I certainly felt better after doing it. The offending letters were removed soon after too.
Tug of War—Fun, but felt more like a bandwagon than a game of tug of war.
Town Hall Meeting—The speaker gimmicks were pretty cute, but we ended up doing something super gimmicky for speaker 3. It turns out the pyenchant dictionary considers every English letter by itself to be a valid word, so we had sentences consisting of a large amount of Zs followed by an additional letter.

Spaceopolis meta—We were stuck with the partial cluephrase **GUCRATER for a really long time. I was disappointed that it didn't turn out to be POGU CRATER. I didn't contribute to this solve.
Balloon Vendor meta—Incredible flavor. The room erupted into WAAAAAs as we realized the transformation.
YesterdayLand meta—Not a fan of how solving puzzles progressively unlocks content on the meta page. It's weird that Tthis meta becomes harder if you don't record how its page changes after every feeder puzzle solve. Perhaps a better way to present this puzzle would be to have the meta page contain only the shell for a jigsaw-ish puzzle, and then have each feeder puzzle solve return a jigsaw piece associated with that answer? I'm not sure what the cleanest presentation for this would be, but I think it's very important to preserve the pairing of blanks and answers.
Workshop metameta—This puzzle decimated my sleep schedule. We unlocked it at about 5:30am. Since ✈︎✈︎✈︎ Galactic Trendsetters ✈︎✈︎✈︎ is a big team, and we only had one set of the physical pennies, we could only have a small fraction of the team work with the physical pennies at any given time. I wasn't particularly excited by this puzzle, so I assumed that I could just stare at spreadsheets until someone else looking at the actual pennies would solve the metameta. This didn't really work out, and soon later someone created a webapp so that everyone could manipulate digital pennies on their laptops. I was able to use this webapp to figure out the correct layout for the pennies about an hour later we unlocked the metameta, and showed some other people my findings. We all thought it was cool but weren't too impressed, since there was no obvious extraction method from this diagram. I then tried to arrange the physical pennies in the correct layout, but got stuck because I didn't realize that each penny had a proper orientation ("Wait, how come there are two pennies with West/South/East icons?"). I then figured out my mistake another hour later, and properly arranged the physical pennies this time. A lot of other people then saw the correct penny configuration laid out on the table and were very excited about it, but at this point I was having doubts about it since I had already been staring at this arrangement for an hour to no avail. We didn't get the final aha until about 4 or 5 hours after that. In the meantime, I knew that I lacked the brainpower to contribute to this puzzle anymore, so I was mostly just sticking around so I would be there when the runaround started. This was unfortunate in retrospect, since there was a good five hours where I didn't contribute to any puzzles, and I really needed that sleep. Of course, I couldn't just leave and go to sleep, since we could have figured out the metameta at any minute.
Final runarounds—I was part of the emoji runaround group. We were able to figure out the 14S clue pretty quickly, but we started midway through the runaround since the clue about the two jungle murals was quite obvious, and we ran into it on our way to 14S. By the time we got back, most of the runarounds were progressing smoothly and didn't need additional help, except for the Wizard's Hollow runaround, which had essentially no progress made on it. A non-MIT remote solver (I think?) figured out the Alchemist sculpture aha, which made those of us who are current or former MIT students really ashamed.

That's all I got. I'm looking forward to running Mystery Hunt next year!