2021 MIT Mystery Hunt, Part 3: Some Puzzles and Some Final Thoughts

(This is a recap/review of the 2021 MIT Mystery Hunt, which happened this month. Puzzles and solutions can currently be found here. This recap may contain spoilers.)

Some puzzles are not metapuzzles. In the puzzling community, we eloquently call these puzzlesthatarenotmetapuzzles. In alphabetical order, here are some of those I wanted to comment on:

Analog Circuitry

I joined this puzzle late, when it was one of three puzzles bottlenecking our progress in the Basketball round. Teammates had already figured out which color combos created the correct subwords, and they’d worked out about half of the gate actions, but I noticed that most of the identified actions weren’t clearly related to the type of gate (AND, OR, XOR, etc.). I found that a bunch of them were bitwise actions on the letters, and also hypothesized that the wire colors told us whether the outputs were valid words. That was enough for us to work out two of the three final outputs and infer the answer.

The main reason I wanted to bring this puzzle up is to highlight just how much the definition of a Hunt puzzle has changed. Back in 2007, I wrote Transmogrifiers, a puzzle in a somewhat similar spirit that asked solvers to analyze the results of applying sequences of functions to letter strings. Fourteen years later, this puzzle has better presentation, asks solvers to do way more, and is much more satisfying. Sometimes I think the entire 2000 Hunt would count as one puzzle in 2021.

Cafe Five

I definitely wandered into a Cafe Five session without initially knowing other people were working on it and that it was a shared instance. Unlike some other puzzles (like Divided Is You), it’s fortunately pretty hard to break; I guess I could have gotten really confused and spent all of our comps.

Anyway, once we actually connected with the rest of the group, this was tense but super fun! A lot of my personal puzzle strength lies in being able to solve a wide variety of puzzle types very quickly, and it’s fun to keep switching back and forth between those different sides of my brain. It was a little frustrating to repeatedly have to abandon logic puzzles because we broke them or their time was going to run out; I was involved in two runs, I don’t think we ever got President Reif to pay for his food.

Circular Reasoning

What a great puzzle! The concept is fairly simple, but the execution seems like it was not at all easy for the constructor(s), and given the constraints, the cycles (including the DELTA &lit) all seemed to be pulled off pretty well. And I feel like I’ve now had the same post-solve conversation with folks on half a dozen teams: “So, did you do a search at some point for words with an alphanumeric sum of 44?”

A Collection of Conundrums and Riddles

I joined this puzzle after some teammates had already done the initial steps, including locating the relevant Riddler column. My main contribution was letting everyone know that Bloomberg also has a weekly puzzle column, and that we needed to look at that too. I read both columns sporadically, but somehow hadn’t seen either in advance, and it boggles my mind that both of these (plus the Stata meta crossword) were blatantly signed Barbara Yew; that was a great “this is relevant” confirmation, but I’m surprised that they were out there in advance with a character name that had been pre-advertised.

And apparently we were not the only team that saw “Barbara S. Yew” all over the place and were absolutely convinced that the theme was going to be Baba Is You… to the point where we streamed some gameplay in a social Zoom on Thursday night to introduce it to Setec members who weren’t familiar with it. And yet we still didn’t solve Divided Is You during Hunt.

Filler Puzzle

A bunch of us descended on this all at once after the coin was found and our projection squad unlocked the unexpected bucket of kilo puzzles. At first the answers to the clues appeared to all be three letters (that’s a lot of work to clue EON), but other late ones were clearly longer. We spent a long time looking for flat-style pairings, until someone ID’d one of the names (Theda BARA, maybe?) and we all realized we were dealing with a pile of crosswordese in search of a crossword.

My initial theory was that unlike a normal crossword with mostly good entries and some filler, this grid would be filled with mostly filler and some good entries, which is where the extraction would arise. In fact, when we came to the answer, the fact that it was centrally placed made us confident enough to call it in, which means we never filled in the bottom half of the grid. So that was a lot of unnecessary image identification on the back end.

For Your Eyes Only

I mentioned above in my Cafe Five comments that I’m a sucker for a decathlon (or in this case, triskadeikathlon?) of straightforward mindbenders. I noticed this puzzle was available with nothing in our spreadsheet and asked why no one had done anything, and someone on our team said they’d tried it (and described the general idea) but that they needed to know a lot of codes that they didn’t know. Earlier, before I knew about this puzzle, someone on our team had wondered if the intention of the puzzle was for participants to cheat, so they wrote and asked, and we were definitively told not to. And thank god for that.

Some background: Many of you know that I teach college math for a living. (Fun fact: As I write this, I’m watching an episode of Veep where a character declared that math teachers are terrorists! We’re not.) Exam cheating has always been a disheartening issue, and by disheartening, I mean the single worst aspect of my job. When I created and ran the alt-vocab memorization task Eggsam from the 2014 Hunt, someone showed off in a blog entry that they’d come in with a crib sheet on the inside of a water bottle, and it infuriated me. Since college courses moved online for the pandemic, academic dishonesty has unfortunately become a lot more common across the country; it was incredibly demoralizing over the course of the fall semester, and I’m pretty sure that if a Hunt puzzle had intentionally been about trying to cheat on an online test, it would have broken me and I would have had a much worse opinion of this Hunt as a whole.

So anyway, I was glad to hear that this was a legitimate mental challenge, and it sounded like a blast, so I jumped into the Zoom. On my first run, I was able to solve six mini-puzzles; I actually might have gotten more, but at six I was afraid I was starting to forget the previous answers (not only could you not write things down or look at references to solve puzzles, you also couldn’t write your answers down) so I decided to leave, record what I had, and come back for the rest in 30 minutes when we were allowed to send a representative again. I didn’t notice until I wrote down the answers that the vowels were all binary, and I almost got the right answer right there; I misremembered the total number of puzzles, thought that the answer was ??EINS????O?E?, and called in THE INSIDE COVER. So close…

Going into my second visit and knowing I just needed to know the binary bits, I memorized the Braille and Semaphore for I and O, but I came in at Puzzle 2 (right before a run of four answers I already knew). This meant I had some time to hang around, during which I saw the other solver present clearly looking back and forth at the screen and some kind of reference off-screen; they didn’t seem to be hiding it, so it seemed like they might not have understood the rules. The proctor clearly also saw this, and looked really uncomfortable about how to react. Ostensibly they were there to prevent this behavior, but they looked like the last thing they wanted to do was to confront them. Running Mystery Hunt is a lot of work, and seeing people solve your puzzles should be fun, not massively uncomfortable.

Anyway, my point of all this is that if you cheated at For Your Eyes Only, bite me.

Fun With Sudoku

Loved this puzzle! My eye was drawn to the skyscrapers first, which reminded me a lot of some of the circulant Latin squares I studied for the last math research project I worked on (a talk about minimal Futoshiki cluing for the 2012 Joint Meetings, which, the discrete math fanboy in me has to point out, was cited by Don Knuth). I was able to calculate the correct number of solutions for that puzzle, and I warned Matt, one of our tech gurus who was thinking of auto-solving the individual puzzles, that they would likely have more than one solution, and we might need to count them. He said he could do that by computer too, but I saw the thermo and warned him, no, some of these may have a LOT of solutions. (In that case, probably close to one-sixth of all sudoku solutions.)

While Jackie and I worked on the late puzzles, Matt wrote a program to successfully count the solutions to the first two, though for the third and fourth he just wrote something like “More than 5000.” Then we had two breakthroughs: figuring out that the parentheticals in the final puzzles represented digits, and realizing that if we could figure out what proportion of all sudoku solutions we needed, the total number of sudoku solutions is known and could be altered rather than building the number from the ground up. Jackie, TK, and I worked on the metasudoku while we generated givens and were making steady process, while Matt was working on a solver for the last puzzle; once we had two-thirds of the givens, we were down to under fifty options, and Matt was able to pull the one that generated a word.

While I’d like to think we could have done the math to solve this by hand if we used to, I really enjoyed the teamwork of hacking away at parts of this with programming and parts of it with math; we solved this at a time when we were stalled on a lot of puzzles and I wasn’t feeling very productive, and solving this was fun (as advertised) and revitalizing.


I hate to end on a sour note, but alphabetical order demands it. Jackie and I spent hours solving the slitherlinks on Mobius strips, and that part of the puzzle was great… but we were 100% unable to read the intended letters off of the correct hexagons, and even after seeing what they’re supposed to spell, I can barely read the letters. It doesn’t help that it’s not clear whether you’re meant to read all the fronts then all the backs or front-back-front-back, and how the resulting hexes should be oriented. Given how difficult it is to read the letters, you shouldn’t be trying to read them off of six different hypothetical hexagon sequences. Great puzzle idea, but the execution did not work for me.


Thus concludes my ramblings on Mystery Hunt, and in a sense concludes an incredible year of puzzlehunt content (yes, I know it’s January, but there’s a sizable gap coming up on the puzzlehunt calendar). It’s been frustrating not to have live puzzling events like BAPHL and the NPL convention and of course the Mystery Hunt, as well as not being able to have parties to solve any of the online events with friends in person. But from Penny Park to now, we’ve had two excellent (if overstuffed) Mystery Hunts, as well as great smaller events ranging from Puzzle Boat to Puzzles are Magic to Teammate Hunt to Puzzle Potluck to CRUMS to REDDOT to Labor Day to CMU to UMD. This has been a terrible period for the world, but puzzle folks have worked hard to create excellent content, and I want to thank everybody who’s provided content in the COVID era. Here’s hoping life gets better and puzzles stay awesome.

9 thoughts on “2021 MIT Mystery Hunt, Part 3: Some Puzzles and Some Final Thoughts

  1. The people running Palindrome’s Discord server downloaded the entire Baba Is You icon set and turned them into Discord icons we could use over the course of what was obviously going to be the Baba Is You Hunt.

    Liked by 1 person

  2. I also really enjoyed our Fun with Sudoku teamwork, and it was satisfying to have different members of the team collaboratively applying their respective skills to the same puzzle. A very minor additional bit of detail for anyone who’s curious: by assuming that we’d extract numbers in the range 1 to 26, we were able to produce 227 possible solutions with the counts we’d found so far, and we noticed our final answer in a list of extractions from those 227 solutions.


  3. Seconding that Fun with Sudoku was indeed fun (and probably my favorite solve of the hunt, since rob99 and I started it cold and stuck with it straight through to the end, picking up Jonny on the way). It worked exactly as I suspected at the outset, and yet, it still took us well over an hour to do all the partial solving and the combinatorics [by hand, thank you very much!], which is kind of the perfect puzzle. (We didn’t calculate *all* of the sub-puzzles, however. We did the arrow, the classic, the extra region, the frame, and skyscraper and then the possible answers on the final puzzle were reduced down to a small enough set that nutrimatic spit out the answer word for us.)

    Also seconding that the extraction on Slitheɹlᴉuʞs was… unfortunate. Solving the individual mobius strips was super fun, but we never actually got the answer to the puzzle (well, we backsolved it), even though we were trying to do exactly what the solution intended. Feedback for the authors/future puzzle authors: if you’re going to have ambiguous ways of reading letters, the phrase needs to be a common English phrase. If you start getting MUFU… that’s not confirming much of anything in terms of right track, especially when it’s not clear whether the M or the U comes first or if the M is possibly also an E or a W. Not to mention ending with a digit that could just as well be an E, M, or W! Really disappointing ending to what could have been my favorite puzzle of the hunt.

    Dan, did you get to see ★ in the students round? I feel like you and Jackie would enjoy that one. (Because of the order of my solving, it served as a warm up to Fun With Sudoku.)


    • We did solve ★, and I was fortunate that I’ve heard of the “Miracle Sudoku” video. Scott and Jackie and I solved the sudoku step, and it was very satisfying. (I didn’t have much more to say about it than that, which is why I didn’t include it in the blog entry. This also saved me the decision of where to alphabetize it.)


    • Yeah, we ended up with NA FUNC OR LENGTH ADD [EMW3], which I interpreted to mean LN (natural function, or an abbreviation for ‘length’) plus some character, so I tried a bunch of variations on LNE, LNW, etc, along with some anagrams thereof. Then we discovered the MU function, so we tried a bunch of things with that, but we never got ADD to become MOD, so never got to a (forward-solved) solution there.

      The slitherlink puzzles themselves were a ton of fun, though.


  4. fyeo was one of the favorite puzzles i testsolved. it felt so good for my memorization of semaphore, braille, etc. to be vindicated. i managed to solve the encodings and stuff in the first run and memorized some of the other things and worked on them afterward, and then i went a second time and managed to solve it afterward. i was very proud of myself for solving it in only two visits, only to later find out that some members of our team solved it without even leaving the room!


  5. A couple yelps:

    – In A Park was a really clever idea and a tremendous team effort, and I would have totally done it if it wasn’t pouring rain outside when we got it. Instead, a few of us reconstructed most of what we needed from memory and internet, and sent the person who lives very close (I only live pretty close) to finish it off.

    – Thanks for including Debate Quotes, and to whatever divine force ensured that was the one puzzle I managed to be the one to unlock in the Projection Device.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s