By now I should expect every MIT Mystery Hunt to blow my mind. Tall order, especially intimidating for me since I’ll be involved with writing the next one, and doubly especially intimidating because this year’s was run by teammate, whose smaller hunts have noticeably higher production values.

As usual, spoilers follow, so if you don’t want to be spoiled on hunt content, consider averting your eyes.

This blog post is not a puzzle. I’m not that crazy.

Prelude: à l’hiver d’un aigle

It would be an understatement to say I was super excited for this hunt. Of course, there’s all the usual reasons Mystery Hunt is the most-anticipated event of the year—fun puzzles, hard puzzles, puzzles that make you have an aha just as you’re about to sleep… but the most significant of them all? Hunt was on campus again.

The MIT Mystery Hunt was held online the first two years I participated, due to COVID throwing a wrecking ball’s worth of wrenches into all human activity, everywhere, all at once. This coincided with my first two years at MIT. All the anticipation I’ve had for running around campus late at night looking for puzzle content since I first heard about hunt has never found a release valve.

Back then, I was a freshman with very little on my plate, and I would spend hours of my spare time wandering aimlessly around campus trying to recreate that feeling (or following an existing runaround, a difficult task since MIT has overhauled its layout plenty in the last few years). I spent both my past IAPs at home out of caution for rising COVID levels. It stung like a wasp covered in fire ants—being an MIT student doing the MIT Mystery Hunt but unable to be at MIT while doing so—but my health took first priority. Hunt just didn’t have the feeling I was looking for.

Cosmic Rays Chandelier

Which is why, when I heard that Hunt was on campus again, I made it a heavy priority to link up with a team that could take me to the runaround. I also wanted to find the coin while I was still at MIT, because such a unique opportunity could not be more enticing.

Which, given how much teams seem to (a) not want to win because writing the Mystery Hunt is a tall order already and (b) (as far as I could tell) not want to have more people out of fear of (a) happening, meant I would have to rely on teammate (the running team, not an actual teammate) to make the call for me. I filled in an unattached hunter form with my preferences and sent it off.

Unfortunately, even teammate couldn’t have made that call for me. Shortly after matching and confirming that the team I matched wasn’t actively gunning for a win, I asked if there were any teams that were (and were open to taking people). As it turns out, the answer was “very much no”. There was only one team in the entire top 20 that indicated even an iota of interest in winning during registration, which does not seem good for the future of Hunt.

Note: ostensibly an image would exist here and at other points throughout this post, but I haven’t made most of them yet. If you want to see what would have been here, the syntax is commented out.

My priorities started shifting. I decided to expand my list of options. After confirming that taking a second matching wasn’t binding, I took the mulligan and got matched to The Team Formerly Known As The Team Formerly Known As The Team Formerly Known As The Team Formerly Known As The Team Formerly Known As The Team To Be Named Later. (That was a mouthful, and I’m abbreviating that to TTBNL from now on should I choose to use the name at all.)

Well, no harm in trying.

I landed back in Boston early Saturday, after what must have been a super turbulent period for flights throughout the nation. The next week was busy between Puzzle Club things and the class I was taking for IAP, but it didn’t diminish my excitement. I went to the MIT Museum twice that week in anticipation of any puzzles that could sneak into the woodwork.

Then came that fateful Friday.

Friday: Allegro molto appassionato

I woke up early that brisk, rainy morning for something for Puzzle Club that it turns out I was not needed for, so I went to socialize with some incoming puzzlers instead. That trip to the Flour on Mass Ave was the first time I’ve walked north of Vassar Street in quite a while, and putting faces to people I essentially only knew by avatar and username was a nice bonus.

I left early to locate TTBNL HQ and set up a position there, memorizing the route along the way, since I was sure I was going to have to make multiple trips between there and campus. After that, it was off to kickoff. Teammate opens up their museum, they’re apparently short-staffed and can’t actually open it, something something puzzles solve everything. Also apparently these puzzles were written by an AI. They then start talking about logistics, Wayne from Puzzle Club sneaks in a reference to a furry convention rule, and we’re off. I see a group of puzzlers I knew gathering outside and take a picture with them before running back north to HQ.

After that, it’s just waiting ‘til the clock strikes 1… 3… 2… 1…

I start by operating off my typical modus: linear searching for the first puzzle that catches my fancy. In this case, it’s Museum Rules, though I do little more than transcription here before thinking I have nothing to latch on to and move on.

I then hit You’re Telling Me. People had already found the major idea here, so I was just around for the work of identification and interpreting the output for our first solve of the hunt.

I briefly examine the new puzzle and suggest some ideas before leaving for greener pastures—in this case Inscription—and clean up a bunch of cryptics within cryptics. We got collectively stuck on ref 5 for a bit, but someone eventually found the right city and we had our second answer.

We got another two puzzles (one of which was in a new round) and I choose to pile onto World’s Smallest Logic Puzzles, which features a bunch of examples of trivial logic puzzles and a number of puzzle types to match them to. Someone immediately cheesed the provided indices to get an instruction; everyone else was left to reverse-engineer how the connection worked. I eventually take that honor and construct an example solution out of the first grid. The rest of the grids fell, I do extraction, puzzle solved.

The team had solved the other three puzzles I had looked at in the meantime, and it was getting close to classtime, so I packed up my computers and headed to campus, with a note to poke me in case things needed retrieving at 5 PM.

Of course, who listens to class anyways? I was just there because I didn’t want to burn my one flexible attendance day on this.

In the meantime, people had started noticing the puzzles were loading slower and slower. They realized the loading animation was a puzzle, solved it, and were now looking at a jigsaw. I couldn’t see the jigsaw at first, but apparently the answer had to be submitted on an individual basis to access it.

The jigsaw was mostly solved at this point (revealed to be of some kind of puzzle factory), but we were apparently missing a piece. This led some of us to start speculating that we were looking for a building on campus, an idea that was quickly shot down, but that didn’t stop one of us from going to MIT’s nuclear laboratory for clues.

Eventually we get confirmation from teammate that the “missing piece” was not intended¹ and, after a lot of anticlimax, we find the puzzle factory behind it all.

Which promptly 404s because reasons.

OK, for real this time.

With that, the load times were no longer an issue, and I could actually look at the puzzle titled Dropypasta. A bunch of teammates took the dropquotes while I scrolled down and cracked the tournament bracket rules. We slowly pieced together the whole tournament from there.

Right after we solve it, we find a puzzle requiring on-campus presence.

Spoilers for A Trip to the Museum

Conveniently enough, it’s about that museum I visited the day before! And doubly-conveniently, I happened to be at Stata, meaning I was probably the closest person to the location.

No other way to go about this, then—I packed up again and made a mad dash for the second floor of the museum, immediately turning right for the tapestry scroll… to find that the mechanism for moving the scroll had inconveniently broken down at some point.

Great, that’s the one part of the puzzle I actually knew the location of. There was just enough information to get the general idea of the puzzle and not enough to actually pull an answer (until teammate uploaded the tapestry as errata).

I then moved towards a bunch of puzzlers all clustered around some plant. The flavortext did tell me to keep an eye out for one special leaf, and there could be no better indication that that leaf was the leaf to look for.

Unfortunately, I couldn’t exactly get the angle right, so I resorted to simply trying to get the whole triangulation. Which is hard if you’re too short to see what exactly is being captured by the camera at any given moment.

I wandered the museum a bit, tried very hard to take a picture of That One Leaf at the correct angle (before giving up and just putting the best triangulation I could grab in the puzzle chat), failed to grasp the correct extraction for the sunshade (before just taking a video recording of that for later perusal), and solved the virophilia section while not knowing that there were secretly two such sections (I thought the subpuzzles were in the order you would encounter them going through the museum in the opposite direction). After doing another once-over to make sure I hadn’t forgotten anything, I did the tapestry section on the stairs on the way out (because no one was working on it remotely >:V) before grabbing a dinner and rejoining the group.

Someone else eventually got the leaf and the sunshade, but we were stuck on the scroll due to simply not knowing there was a museum component to it. By the time our team reached that point, though, we were long past the Science meta, and also technically had enough information to score a sidesolve.

I came back in time to help out with transcribing our extraction for the Atrium meta before jumping to transcribing some Morse code message. I had pulled out a grid and was about to start decoding the clues before someone beat me to the punch with an automatic decoder. Said decoder had no concept of line breaks, though, so the rest of the file I had opened in audacity turned out to be useful anyways.

We then started solving clues. Along the way, though, we fell into a bit of a trap assuming all the answers were three letters long, and also didn’t know what to do with the long beeps and the random sequence of ten dots.

I moved off it for a bit to clear my brain and helped fill in the grid for Extrasensory. I saw a thing on the diagonal and tried submitting it, which surprisingly didn’t work. Some teammates eventually found other hidden words and submitted the actual answer.

By now we’ve opened up the second meta of the night: Nuclear Words, for the Science round. They were basically done with theorycrafting at this point, making my job to pull out Quinapalus and put the theory into practice. Eventually we have enough data to confidently declare an answer. Two rounds down, N-2 to go.

While I was bashing away, though, someone had declared a newly-unlocked puzzle to be about wargames².

It took me a bit to notice that this puzzle existed, but my eyes had never jumped to look a puzzle so quickly. I took one look at the puzzle and immediately knew what it was really about.

OK, time to grind through identification… oh, I forgot Tineye and Google Lens are absolutely sh-t with board game tokens. And these tokens betray like, nothing to a historically-minded person—I can’t even start reducing the options down to a cluster of related board games. Oh, someone identified one of them… aaaaand it doesn’t fit what I was going for, great, there goes my idea for how this puzzle works.

Spoilers for Business Cards

There were a great many aspects to the puzzle that soured it for me. Ostensibly I should be over the moon that this puzzle exists, but subject material alone does not a favorite puzzle make.

The intended first step—identifying one token from nineteen games—is at best an absolute slog to get through and at worst a complete show-stopper. If the first step is unapproachable, it’s an immediate turn-off for most people looking at the puzzle. Most of the tokens don’t even have a “side route” to identification, i.e. researching subjects known to be from a second field of knowledge to find a potential anchor. Take the Kaunda token used in the puzzle, for instance: Kaunda is a surname used only with one noted Zambian politician family, and in particular Kenneth Kaunda, the first president of Zambia, matches the profile on the token. Perhaps by digging even deeper one can discover his support for black nationalist movements in Zimbabwe, and thus find the board game about the Rhodesian Bush War, where he is a token. (Alternatively, just look up his name on BoardGameGeek.) But for most games of this nature, one piece is not sufficient information to pursue this second line of thought. There are already at least five wargames in the puzzle itself that use what seems to be the standard layout for army tokens, which are only distinguishable by a bunch of random digits with essentially no meaning.

I had a rather lengthy discussion with another puzzler about this subject. I suggested that adding the whole board to the equation would fix many of my issues with the puzzles; even if Google and Tineye failed, I could get some idea of what the board game was by looking at, say, the map, and deducing what the subject happened to be based off topology or history or whatnot. He brought up that this brings in the potential for people to short-circuit the IDing step and skip straight to part 2, and that it was likely part of the design that people would have to learn the name of the board game at some point. Which would be a completely valid point… except the puzzle also has an indexing step at the end that already requires you to know what the actual board game is, which really just makes gating with the ID at the start of the puzzle seem wholly unnecessary.

I did eventually try to solve Business Cards using only knowledge my team had in its final state, throwing out the PanzerBlitz identification instead of the aha. It is theoretically possible to, at that point, figure out the cards were in alphabetical order, find a list of cards that can be sorted in (quasi-)alphabetical order, guess which ones could possibly have a related board game (mostly the set relating to conflicts like Quagmire or South African Unrest), backsolve the board game by looking up the war in question and checking for the right tokens (since the subtitles will often list the war), and finish the puzzle normally from there. Part of me still wonders how this puzzle even got past testsolving in its current state, because who else would look at the card formatting, immediately scream “Twilight Struggle”, and start backsolving like this?

Also, to get one final nitpick out of the way: the Next War series explicitly explores near-future wars, so it can’t possibly refer to an event during the Cold War. There, at least one historical inaccuracy in the puzzle.

I completely gave up on solving that for the foreseeable future, instead turning to another puzzle that was well on its way with ahas and just needed help with execution: Catenaverbozoa.

Spoilers for Catenaverbozoa

It was a simple concept—assemble fake scientific names whose English translations are anagrams of things depicted in these boxes—but finding the right translation to anagram took considerable effort. We managed to hit it on the nail for sufficiently many of the descriptions in the first half, at which point someone just anagrammed them for the answer.

At this point, I started jumping randomly between puzzles to see where I could contribute. Two physical puzzles had recently unlocked, and I was raring to go at one of them, but they wouldn’t arrive for a bit.

Saturday: Adagio sostenuto allargando

The clock tipped past midnight. It’s been 11 hours since hunt started. No harm in staying around for a bit; there’s still puzzles to help out with.

Puzzle about Tunic script? I’m still unspoiled on that and still have designs on playing Tunic later, so despite the fact that it ticks all my boxes for interest… I really can’t.

Puzzle about garden path sentences? Seems well on its way and—oh, they found the answer already. Guess I’m moving on.

Fans? Hmm, seems to be stuck. Oh, they misindexed somewhere… there we go, it makes sense now. …kind of. I don’t know what this means, or what the stripes on the backs of the fans are. I see more indexing, maybe? I can’t tell. Next.

By now, the physical puzzles had arrived. I immediately moved to the table with the wooden sticks for Weaver, which we immediately turn into a classic weave.

Spoilers for Weaver

Which tells us to immediately weave it under a different set of instructions. Except none of us can distinguish S and Z twill and we’re most of the way through the puzzle before I stop everything due to extract not making sense.

Which now asks us to make a third weave that we also have trouble interpreting.

Which asks us to make a fourth weave by swapping the order of some things around, except we forgot it said TWO BY TWO and were well on our way to trying a classic weave again!

We have to be at the end, right?? We’ve made four weaves with the same wood, they’re already starting to splinter, and I suspect if I do any more I’ll be getting a dozen blisters on my hands. It says… “BE HONEST IS THE MORAL OF THE STORY FOR MIT ADMISSIONS”. Well, it definitely talks about underwater basket weaving…

Three UNDERWATER-related answer submissions and precisely zero results later, I start joking that the puzzle is actually trying to get us to basket weave underwater. Which is, of course, not the desired result for me and my tired hands, but I really wanted to get it over with.

We took a bunch of pictures of our current weave in case it was going to be irrevocably warped by the experience and we took it to the bathroom. I held it under a sink, completely expecting my line of reasoning to be a bust…

…and also completely forgetting that this was teammate we were dealing with.

We read out a message, but parts of it were too warped³ by the act of soaking in water to be able to read clearly. Since it was our only way of continuing the puzzle we were left only with the option of asking teammate for the actual cluephrase.

Which gave us—of course—one last set of weaving instructions, which for once is about making an actual basket.

Unfortunately (uggghhhhhhh), this one was just as vague as the last. We get the exact strips needed to make the basket, as well as the ordering, but unfortunately it gave us zero details about where the rings start and end, or which orientation the final basket weave should be using.

Which meant that I ended up freestyling a basket and our group promptly got stuck on extraction.

I absolutely did not want to deal with the basket ever again, so I promptly left for greener pastures. Our team eventually got a paper copy of the wet sticks, which didn’t have the problem of wiping off invisible paint attached to it, and also more clearly revealed that not all the paint went into letters.

After that whole story, I tagged onto Word Press to help with the identification effort there. Teammate really loves making puzzles out of nonsensical phrases, don’t they. It’s not a habit I like, because I am very bad at coming up with their nonsensical phrases⁴, but what’s done is done, and the chat became a cacophony of puns I wanted absolutely no part in.

It was getting very, very late, but I was committed to holding down the graveyard shift. I start playing New World Symphony on loop and prepare for a long night of delirious puzzling.

We suddenly discover a basement to the factory, and I pile onto Think Fast because it looked interesting. Not much happens, we eventually run through everything and discover an interaction at the end, I end up not participating in said interaction because we wanted to wait for HQ to reopen and it happened while I was taking a mid-day nap.

In the meantime, though, I grabbed the puzzle labeled with a single 🐍 emoji and pull through it with one other person. I then moved to Now I Know (which was most of the way through identifying all the pictures), guessed what the puzzle was about off vibes, and end up going most of the way through the puzzle alone (though I still needed to ask a teammate to interpret the final cluephrase after Google Translate spat garbage at me⁵).

I then get pulled aside briefly to look at a beatboxing puzzle, but I don’t notice anything new (“ORANGE BLACK GOLD”? Really, of all the things they could have used they chose resistor code?) and instead tag to City Placement, a puzzle about guessing major American cities and hoping the roads around their town hall match one of the maps in the puzzle. We eventually figure out the maps come in pairs (the city hall and some random fire engine in the city for whatever reason) and that puzzle eventually falls as well.

I see a puzzle labelled “Diagramless”, jump on it to finish off the crossword and then immediately recognize a number of unclued entries as referring to patterns in Conway’s Game of Life, conveniently hidden within the borderlines of the grid. One incorrect submission of the one unclued pattern in the grid later, I try extracting by running our black squares in the Game of Life; I end up with a bunch of gliders flying off into the infinite void and an arrangement of eight squares that read off like gibberish. Completely out of options, I mark the puzzle as stuck on extraction and then get pulled aside for the beatboxing puzzle again, which had made some progress since I last looked at it and suddenly also wanted someone with MIT knowledge. Remember “ORANGE BLACK GOLD”? It’s only clued that way due to an association with the MIT a cappella group called the Ohms, which our team discovered had covered one of the songs being beatboxed.

Unfortunately, I had approximately zero such knowledge, and a formal list of a cappella groups at MIT proved elusive, but I solidified the connection by finding the Logarhythms and reducing the seemingly ridiculous index of THOUSAND down to another three. In the meantime, Charles runs the exact same extraction I tried on Diagramless, gets the same gibberish string of characters… and recognizes it as an actual name.

At this point it’s nearing ten AM, and the tiredness was really starting to take hold. I took a look at one last puzzle, Broken Wheel, and executed a bunch of rotations on the completed grid in a way that seemed sensible to me based on what I was given. I didn’t find an answer, though. I called it a night (despite it being very close to midday by this point) and started a long trek back to my dorm room for a power nap. Little did I know that the answer was right in front of me, and someone else pulled it off my grid about an hour later.

By the time I woke up again four hours later, another 24 puzzles had answers on them, including Business Cards (which had a free answer token used on it), Weaver, and the three remaining metas in the Museum. Which means I was right on time to help out with the museum metameta.

Spoilers for MATE's Meta

At first, of course, people were distracted by why answers had glitches applied to them, or why some were associated with the wrong round. Many theories started to emerge. Slowly, ever so slowly, we start piecing together the puzzle. Someone notices repeated bigrams in answers labeled with art (SEEING DOUBLE RAINBOW). I find an undercurrent of a country theme in those listed with tickets (COUNTRY FRIED MISTAKE), then see two Os each in those next to globes (THEY LEFT BIG INNER CIRCLES). OK, good, information to go off of, and it seems to make a bigram for each for eight letters per row. I try one of the rows and it spits out DIAB??ES, which confirms a Natural History extraction theory as well. We all start piling on the rows, someone figures out how the Science extraction works, everyone starts brainstorming how to break our eights into two fours.

We soon find that each one can be split into two words, preserving internal order, such that each word is an anagram of one of the words from the row above or below. From there it was a rush to fill out the extraction table and get the answer in.

Much of TTBNL imagined MATE going rogue at some point after the museum metameta, perhaps the most predictable twist to come out of an AI-themed anything since the 1980s. Turns out the rigor demanded of writing a Mystery Hunt alone is causing MATE to react as a typical overworked human would—surprise, AIs experience burnout too!

And that’s why you should stop joking about me doing hunt solo. Yes, you. It’s not going to happen.

I then get pulled on The Radio for a third time, after people have started extracting and submitting incorrect completions. I fix an index (phosphate’s resonance structures) and immediately get led to the correct answer, right before the team working on the Office meta figure out extraction there.

We are thus given some juicy intel about teammate’s prior attempts at creating puzzle-writing AIs. Interesting…

I then leave to work on Circuit, which I obtain the basic rule for… to an extent. I was technically correct but there was a stricter rule involved that would have helped with the logic.

I leave that for Subterranean Secrets, which had just performed the on-campus bit and found seven riddles.

I contribute one answer and randomly choose to submit some phrase that showed up randomly in the first letters… which tells us to pick something up from HQ. Just as the person who did the in-person component got back to HQ. Great.

I wait around for a bit, looking through other puzzles in the meantime. The person sent eventually picks up a UV light and reveals seven more riddles hidden in plain sight… just as our team finds a giant pile of junk in the Basement. It is very quickly torn apart through a torrent of ahas, and we recover some hard drives.

Well, what do you do with hard drives besides plug them into the nearest computer?⁶

This reveals a suite of four minipuzzles. I bounced off all of them (besides contributing one answer to the second drive after some teammates found a constellation connection), and there were really only two things left to work on.

So I linked up with the crew solving the Hall of Innovation, which had made considerable progress in the meantime.

Spoilers for the Hall of Innovation

The Hall of Innovation had opened up while I was asleep, and involved a number of “gizmos” scattered around the factory floor. Whenever a gizmo was interacted with, it would change the content of some puzzles in the Innovation Round in a predictable manner. This also changed its answer, which could take the form of gibberish, the empty string, or simply not exist at all.

The team has, at this point, documented the effects of the nine gizmos on most of the puzzles in the round, had the correct extraction for five of the seven puzzles, and even deduced configurations that would allow for some of the puzzles to output reasonable answers (notably Image ID and Numberlink, with Word Search well on its way). In the meantime, though, they were trying to extract from everything else in order to reach the meta.

This happens around 22:30, which immediately confirms the prevailing theory and also is immediately solvable without having the other five answers. Whereupon the answer checker tells us to factcheck the puzzles first.

With the list of answers in hand, we obtained valid configurations for Maze and Word Search within minutes of each other and start focusing all energy on Connect the Dots…

While this was going on, I was keeping tabs on the progress of Reactivation. One by one, the team figured out the names of each of the four AIs on the drives. And just as we had finally figured out the full extraction mechanism for Connect the Dots…

Sunday: Subito prestissimo affrettando

This section is defensively spoilered, since it contains late-hunt spoilers.

…the power goes out around us, the gizmos stop working, and we receive a rather threatening visit from teammate informing us that Mystery Hunt was cancelled.

Well that sucks, we can’t even work on Innovation anymore with the gizmos bricked. Guess everyone’s piling on that one puzzle Tyler’s losing his mind over… oh, wait, there’s a new round? Full of… puzzles we’ve solved before… and that one puzzle titled with a Gemini emoji we never solved, apparently.

A few minutes of submitting used answers later, we have five of six answers to the round and an unlocked meta that told us to find a bunch of laser-cut triangles outside HQ. Oh, and also we met a couple of the AIs we revived along the way.

I immediately pile on the triangles. It was immediately clear what we needed to assemble, and I was in the thick of it, first compiling a letter distribution chart to find patterns (“no Ds? weird”) and then directly initiating the assembly. I start drawing diagrams to explain the puzzle to confused passersby. At one point, we had a bunch of leftover tiles we didn’t know how to fit with our mostly-complete layout. The group that had returned to working on ♊ came by, looking for backsolve information; we gave them the tiles and some candidate letters to transcribe. In the meantime, we found an alternate way to place one of our answer words, which resulted in me borrowing another two sheets of paper to transcribe as necessary. Just as we do that, we hear a sudden cry coming from the other side of the ballroom—♊ had just backsolved their answer, and that affirmed our alternate placement. Just as I finish transcribing our new arrangement, I see the extraction… and that’s when we realized that this round went a lot deeper than expected.

Oddly enough, the tiles also came with a weird threat on a small slip of paper. We penciled that down for later in case it was meta-relevant. (It wasn’t.)

After that triumph, I take a look at the eleven or so new puzzles we had. Each one felt very intimidating in some way or other; I decided to fall back to the one puzzle I did know how to do, the Innovation Round, and join the two other people working at it in spite of the shutdown.

They were still able to manipulate the two gizmos controlling Connect the Dots due to having not reloaded the gizmo after shutdown, so we slowly work out how Connect the Dots works and fix the clock arrangement and the grid to one degree of freedom each. It is at this point that I get the epiphany about the extraction for Cryptex, and find a satisfying arrangement of gizmos. Cipher, with its four gizmos now tightly constrained by other puzzles, could now simply be eyeballed for a solution.

Unfortunately, the factory was still shut down, so we couldn’t simply submit the meta answer then and there. We worked out what we wanted the arrangement to be in analog, then left a note about the situation.

It was now half past 2 AM. I saw many more hours of hunt ahead, so I abandoned my plan to stay up through the rest of Sunday and started packing up to head home.

Though I wouldn’t exactly catch my Zs just yet.

As I arrived home, I decided to check in on one of the voice chats; someone was hammering out code to extract things from Pixel Art, and had already found steganographic information from everywhere under the sun, including short-circuiting one of the steps to find a hidden message in the fourth red bit early.

Spoilers for Pixel Art and the Expedition Round in general

I suggest a sample parsing of that hidden message that turns out to only be partly correct; someone corrects me on the blue parse, and the two together manage to produce a cogent image cluing the answer to this puzzle…

…which also explicitly made the correct newlines and punctuation part of the answer.

Oh right, forgot to mention. teammate made this whole round, the Cosmic Discovery Expedition produced by this ASCII text cat, accept answers that included newlines, whitespace, punctuation, and the whole rest of the nine yards of ASCII.

Gotta admit, that kind of curveball takes impressive amounts of audacity to pull off. It’s incredible shock value.

With that last solve complete, I went for a second power nap…

I walk back into HQ at around 10:30 AM. One solve has happened in the meantime, which was about the velocity I was expecting for the puzzles I saw. Teammate also reenabled the gizmos for us, so I cleaned up the Innovation Round as planned. This results in unlocking the meta for the factory floor… which was immediately back-burnered due to a) our priorities landing exclusively on the new AI rounds and b) the fact that it referenced the Hall of Innovation round page, which also no longer existed due to teammate wiping the museum in-story.

I jump onto 5D Barred Diagramless With Multiverse Time Travel, which seemed fairly self-explanatory—the diagramless part was mid-way through being assembled when I found it, so I helped pull it through and then used the context to construct some chessboards.

Spoilers for 5D Barred Diagramless

We start brainstorming a lot of theories about what the green cells meant, which involved plenty of crackpottery on my part before we settled on the obvious answer of “checking a king in the past”. We then start debating the rules of 5D Chess; I had watched the video on the topic before and decided to break off into my own grid after deciding the debating was going nowhere. However, by all metrics possible, we were only finding five possible squares despite the puzzle claiming 22. Double-checking my work ends up with the same result, and I eventually drift off the puzzle to find other work to do…

As it turns out, we were still very much on the right track. We simply missed a quirk that, while present in regular chess, takes on a whole new character in the 5D field of creating timelines and boards: a “discovered check” can occur when copying a board to a new timeline if the pieces line up correctly, causing every move to that board to be a valid checkmate.

I continued to ruminate on this for a bit until I got distracted by a puzzle from the last unlocked round, Interpret Perplexing Texts, to which I contributed some amount of gruntwork.

At this point, teammate has realized that hunt was not progressing at the planned velocity and has started raining free answers upon everyone. With this in mind, the team refocused towards cracking metas (though they didn’t stop people from focusing on puzzles they found interesting and wouldn’t take a lot of time). We had rounds open for all four AIs at this point⁷; I was asked to assemble a pod for the Space Modules meta, which we now had four answers to⁸, soon to be the full complement of five⁹ since people were starting to wrap up Terminal. I did end up getting the group coordinated, but none of the many, many ahas in the puzzle came from me, as I would eventually be pulled away… again.

Remember how I was the one arranging triangles in The Legend, the first Hole meta? Well, people had unlocked the second level Hole meta, called The Scheme, in the meantime, and the giant triangular-shaped arrow on that meta page suggested, perhaps, reusing the old Sierpinski triangle?

Unfortunately, every idea we tried on this front went nowhere.

In between attempts to make sense of the triangle, I took a brief look at Mosaics, a puzzle in the Ascent round. I also went back to Space Modules and helped out where I could—double checking the extract and spotting the final sort early. We eventually found the puzzle referenced an XKCD comic, with a seemingly sensible answer sitting on top.

We submitted it. It didn’t work. In frustration we submit the whole transcript of the comic and also the easter egg that came about as the result of said comic (both incorrect for more obvious reasons). I leave again to help brainstorm for The Scheme.

…and but thirty minutes later, we get some news—the answer checker was just broken and didn’t accept our rendition of the sensible answer (presumably due to casing).

Another thirty minutes go by just brainstorming through a dead end. Then we get another email.

The collective facepalming reaction was immediate, and soon enough we see more puzzles on the other side. Oh hey, one of them’s about Quiz Bowl!

…uhm. Wait. These questions make no f-cking sense. Also there’s 48 of them. Never mind, we’re tokening this at the first opportunity.

I guessed that this cycle was probably going to become a running theme.

We immediately rammed through three puzzles in the new third level in a bid to find the meta as soon as we could. I spent a spell looking through the other two open rounds (the newly-unlocked Ascent meta and the game forming the backbone of the Quest round (long story)), but couldn’t quite find any good inroads on either. Another hour and we took a fourth free answer to unlock Lost At Sea, but we were still stuck on a wrong idea until a forward solve for Flooded Caves pulled in an answer with one clear referent and maritime flags to match.

Identifying all the ships and their hull letters led to a relatively low-intensity ordering step that we proceeded to overconstrain due to a misunderstanding of how indexing was supposed to work in this puzzle. Incidentally, we had also axed an identification due to a combination of this perception of indexing and The Error That Cannot Be Named; even after restoring such and doing the indexing properly, we still couldn’t get any phrases that worked.

Eventually, we decided to grab the last answer, due to there probably being more in the round after this, and found a ship to match. Someone took a look at our work and immediately randomgrammed the letters into the correct answer.

As expected, this unlocked yet another layer full of puzzles. The problem? All of them were literally blank.

Confused, we immediately burn a free answer token on one of them. Fortunately we had people who had worked on Collage and recognized the connection, which finally, finally brought us to the meta for the round.

This was it. No more puzzles to unlock, just three metas (and a metameta) between us and the end. I had started a run on Quest in the meantime, which I got stuck on due to a lack of answers, before turning to look at Ascent, which was already at 8/9 bought answers.

Spoilers for the Ascent Round and its meta

This round’s gimmick was that all answers were written in a foreign language (with the puzzles themselves also being generally centered around said language/culture). I didn’t know much about the feeders—the only puzzle I even partially worked on in the round being a fill-in crisscross featuring German compound words—but I was ready to give the meta a shot.

By now, everyone else had gotten through the part requiring Vocaloid knowledge and had tentatively assigned each answer to a vocaloid based on the blessing they took in some song (IDK Vocaloid stuff). The leading theory was that each answer was a mistranslation of one of the Japanese words, and that translating the intended Japanese term backwards would yield something.

Wheels spun for about an hour before the final answer for the meta was bought, immediately upending the matching we had assumed—the Cherokee answer meaning “anxiety” was originally assigned to the blessing for whirlwinds, while the incoming answer was literally a kind of windstorm. And despite the obvious matchup there were multiple attempts to undo the change.

This still led to no new revelations, so I jumped to the Hole meta, which was well on its way.

Spoilers for the Hole meta

It was pretty clear that each of the “people” in the meta was meant to match to two of the answers, so we start matching. A couple of the pairings were… questionable at best, especially when it came to shoving BRITAIN into a category.

Eventually, though, someone found the kicker: each pairing of answers had one “real” and one “imaginary” answer. Now using the numbers made sense, as components of a complex number to plug into the equation.

However, which number went where was still a massive subject of debate. When I took a look at it, I noticed that PICO, assigned to chemistry at the time, was actually referring to the fantasy console PICO-8; this opened up a lot of ground, though the matter of reassigning TRIFORCE, in PICO’s old spot, was slightly more contentious. I wanted to slot it in Hausdorff-imaginary due to the Sierpinski triangle, others wanted to put it with the jigsaw puzzler. And no one could agree on the placement of BRITAIN. Some wanted it with Jigsaw Puzzler (4). I personally was in favor of Housekeeper (9 royal houses, in the same way as the associated imaginary answer defined “house”). Eventually someone found out that Britain had a Hausdorff number attached to it, and I shot the notion down in spite of knowing about the coastline paradox and all it entailed because I still thought TRIFORCE fit better in that slot.

Eventually, though, we had enough letters we were sure of to make the right pun. Two down, two to go.

I then returned to Ascent, which had made almost no progress in the meantime.

More spoilers for the Ascent round and meta

It seemed like people had started making crackpot theories and pursuing them in other sheets. One was trying to associate the languages to the clues based on vibes only—which, I will admit, I wouldn’t put it past teammate to do¹⁰.

Eventually, though, I catch an airwave about someone mentioning exactly how they thought 炎 “fire” (from a blessing) got mistranslated into ΦΛΕΓΜΟΝΗ “inflammation” (one of our answers, written in Greek), and specifically noted that the Japanese term for inflammation was 炎症.

Somehow, that was what triggered something in my brain—at no point did I think of translating the answers into Japanese.

I run off this hunch and translate the other answers we were sure of the matching of into Japanese, pursuant to the first part of the icon, and find a similar pattern for everything—even the seemingly unrelated connection we were sure of (the Sanskrit term for “animal”) matched the corresponding blessing on one kanji.

Knowing this, I translate the Cherokee we had displaced earlier, which I didn’t know the word for. Once again, one kanji—and once again we had to reorder, because the answer written in Arabic script, which we had marked as “consumption” in Persian, was already occupying that slot. Curious, I translate the word out of Arabic instead; it comes back reading “bank”. A perfect confirmation.

In the meantime, I caught another airwave. Someone else was explaining an inherent pun in the title of the song they used—”love” and “sorrow” both translated to the same pronunciation in Japanese, just using different kanji to distinguish the meaning. Both words were present in the flavortext, which served as a confirmation mechanism.

I wasn’t paying attention to that much—I had switched to transcribing the Japanese translations into Rōmaji for the benefit of everyone else. When I reached the bottom of the list, though, something just… clicked.

My attention had shifted to the second part of 動物 “dō-butsu” (animal). Read another way, it could be 仏—the Japanese word for the Buddha. Which conveniently fit the corresponding clue of “historical figure”.

Just to check whether I was going crazy, I tested this theory on the next two I could. 銀行 “gin-kō”, I soon realized, could become 紅茶 “kō-cha”, black tea (despite 紅 itself being “red”). 波長 “ha-chō” was 蝶 “chō”, for a butterfly. And as one final check, I translated the butterfly back to Dutch.

VLINDER. Seven letters, extract the first.

I sent the word out. The remainder of the puzzle fell within minutes.

It was good to know my latent Japanese knowledge was good for something.

Solving that left exactly one puzzle to crack: the meta for Quest.

Spoilers for the Quest round and meta

Quest was a… weird round. At its core is a procedurally-generated roguelite, involving walking between rooms, fighting monsters that gave minipuzzles as challenges, finding keys in chests locked by puzzles (of course), and using those keys to unlock doors to access more puzzles. Eventually, with enough keys, one would encounter a boss gauntlet consisting of some number of all the monsters in the dungeon, as well as a cryptic message…

“Count your foes carefully to map your way to victory. You might meet your match in this room!”

Everyone interpreted this message in myriad different ways. I, for one, immediately took to matching monsters to answers using the description of the task and whatever else I could find. In the end, of course, I found this endeavor futile and moved back to mapping everything.

Monday: Tempo rubato

Time soon slipped over into Monday. Still no sign of anyone having found the coin, and still no end in sight for this last meta.

This section is defensively spoilered, since it contains late-hunt spoilers.

Nothing to do but keep mapping and hope things line up. I had enough knowledge of how the game worked at this point to solo this task. At the end of the first run I tried to map, though, I was quit out of the game for unknown reasons. When I tried to start up another save, the same thing happened right at the start, which irked me enough to ping everyone about it. I eventually got one complete run.

Teammate soon announced that “"”errata””” (hints) would be distributed for the AI metapuzzles at 2:30 AM. By this point, I was very much tired of continually running through this game, and so I just spent time listening to a violin concerto while helping others with the boss gauntlet.

Eventually the clock struck 2:30 and the hints came out. I remember being sorely disappointed that the two metapuzzles we had actually cracked were given direct cluebats while all we got here were a list of things that were irrelevant to the puzzle.

I walked over to sit with someone who had a bit of a theory to propose: their group had constructed some supposed mappings between puzzles and monsters on their maps, and each puzzle they saw more than once had a consistent index to go with it. I immediately dismissed the idea, but they convinced me to pull out my mapping to run a comparative analysis with their theory.

A lot of things did match up—not everything, but at least more than could be attributed to coincidence. Enough to be convincing. Enough that I felt comfortable adding my own data to the chart.

I walked back to my table to do some more mapping, perhaps to obtain a good sort order, but by that point I no longer needed to—someone had randomgrammed the meta answer using the indices.

All four metas had fallen—and that left for us only the final capstone, which was, after that colossal amount of effort put into metas, a nice breath of fresh air.

We clocked in the final answer at 4 AM. Teammate immediately called us to schedule a runaround. Immediately afterwards, we received an announcement e-mail ominously stating that “the coin will have been found” by Monday morning.

So, this was it.

I stayed around the lounge for the next two hours, cleaning up my corner of the room and then just lounging around talking to teammates and watching a Schrödinger’s card game play out. I had hardly eaten anything and had been running off “meals” of milk and instant (uncooked) ramen for all of the last 48 hours, but I had never felt more energized.

Soon enough, we were off to Lobby 7. And I would finally be able to go on a runaround.

I brought out a penguin plushie I kept in my bag the whole hunt. This was something I really wanted to be memorable.

In the end, the runaround was less “figure out where to go” and more “here’s where to go, now do X” than I would have liked, but this sentiment is minor compared to being able to experience the denouement of an entire year’s worth of logistical difficulties.

And as we unlocked the door to the Bush Room and turned the gears of the puzzle-making machine, I could feel the determination welling within me.

We are going to write a hunt. And I am going to do my damnedest to ensure it’s as good as it can be.

Coda: Poco a poco ritardando

It was now snowing outdoors, almost as if Mother Nature herself had shown up to “congratudolences” us. Unfortunately, I did not prepare sufficiently for precipitation, so I ended up cradling the puzzle box I was now holding the whole way back to the hotel for the team meeting, and then all the way back to my dorm room.

I still couldn’t catch Zs; there was still all of wrap-up to attend, and people to meet, and everything else… Unfortunately, I spontaneously fell asleep in the Student Center while waiting around and wound up appearing at wrap-up thirty minutes late. Whoops. Sleep has a habit of catching up—especially today, where I ended up retiring at 7 PM, significantly earlier than usual.

Hunt is draining, yes—and even more so this year with how long it ran on. The old guard can pull out paragraphs upon paragraphs on how Mystery Hunt has gotten too long for its own good and there should be a conscious effort to pull it down for the sake of everyone’s enjoyment. In the end, though? Fun, especially here, can only be defined as a product of multiple wildly unpredictable factors on a per-person basis—there are people who enjoy blitzes and people who enjoy marathons, people who enjoy simplicity and people who enjoy complexity. And, looking back on it many weeks later, can I honestly say I didn’t enjoy the hunt as a whole?¹¹

See y’all next year.

_xxx_
x_xxx
_xx__
_xxxx
xxx__

Based on my interactions with the other jigsaw puzzle after the hunt, I think I know what the bug is: the piece started low enough on the page that it was hidden outside clickable range. Zooming out does not fix the issue; while it reveals that the piece exists due to scaling its position, it also immediately rehides it by restoring every piece to its original position and size. Zooming in, on the other hand, explicitly exacerbates the problem. ↩
Not the movie, the genre of board games. ↩
ba-dum tss ↩
Teammate’s last two hunts, particularly the metas, loved using nonsensical phrases for no particular reason, to my absolute frustration as I dumped a grand total of seven hints (two regular, the first of which did essentially nothing, and five followups) into one meta in 2021 due to teammate also being notably clammy when it came to hinting metas in particular. ↩
I suspect the main problem was not realizing that the X was actually a silencing diacritic. ↩
Yes, I know the potential hazards of plugging a random USB into a computer. ↩
Since the titles of these rounds are actually spoilers for Reactivation, I will be referring to them by the words “Hole”, “Expedition”, “Ascent”, and “Quest”. Note that clicking on any given link risks spoiling some part of Reactivation. ↩
Besides the aforementioned Pixel Art and 5D Diagramless (which we burned a free answer on), we had also solved World’s Largest Logic Puzzle and burned a free answer beforehand on Invisible. ↩
The scavenger hunt, while in this round, did not contribute to the meta, instead giving two free answer tokens. ↩
See: Sword Swallowers. No, I’m still not over that, and probably won’t ever be. ↩
~~Yes and it’s all the fault of that f-cking owl.~~ ↩