Time for another long writeup about Mystery Hunt. As always, puzzle spoilers will be hidden behind "Reveal spoiler" buttons.
Our team's leadership decided we are in no way ready to construct
another hunt already, and the easiest way to avoid doing that was
to split into two teams. The idea was to have one large team with
all the dozens of Caltech students, part-time remotes, and those
who love to hunt with huge teams, and another team for people who
want to have more fun solving hunt, participate with a smaller
team, and actually be able to know all of their teammates, and actually
see and enjoy more of the puzzles in each round of the Hunt
(even if it meant we did not reach the end).
For a couple reasons, I decided to join the smaller team. Part of
it was that I was never entirely comfortable with being on such a
huge team. Another part was due to a teammate I really felt like I
first got to know during our construction of the 2024 Hunt, John
Bromels, who many of you will remember from his appearance during
that Hunt's kickoff as Poseidon in the guise of a Southern
gentleman. He is an incredible entertainer, motivator, and even
fundraiser. John was going to be leading the smaller team, and
that helped lock my choice to join that team. The large team ended
up still named The Team To Be Named Later (with some number of
Formerly Known As thrown in there) and we became The Team That Is
Now Named Later.
So I joined The Team That Is Now Named Later, and with only about
25-30 of us on campus out of a total of perhaps 40 people, we got
an on-campus classroom HQ, a break from my years since returning
from COVID on hunting almost exclusively from Le Meridien. This
meant we had to vacate overnight, but that wasn't such a big deal
for me since I am an early-shifted person and was easily one of
the ones to come in and open HQ in the morning when campus opened
up.
We chose not to try to use solver.tools due to its historic instability, and got Jim Hays trained up on Cardinality's Cardboard app which was available in an open repository somewhere. I wasn't involved in the setup, but it worked well. There were a couple things Jim had to do to authorize each user both to have access to the hunt in Cardboard and to have access to the Google Sheets we were creating for each puzzle, but this process was familiar.
Some other things happened before Hunt besides forming teams:
There was an official pre-Hunt, the MIT Mystery Heist, run
by Providence Crime Syndication, which introduced some of the
characters and story from the real Mystery Hunt. It was released
in an on-campus event by MIT Puzzle Club October 14th, and in
December the puzzles were released on the above web site. I didn't
manage to get together with my team for this one, but sometime in
December I soloed it.
My teammates also shared with me the Puzzle Corner in MIT's Technology
Review, a feature which is usually more math/science
problems but in the
Jan/Feb 2025 issue was more of a mini puzzle hunt, written
by longtime Hunt participant and frequent constructor Dan Katz.
This had some puzzles I thought were easiest to get through by
printing it out, and I did that solo as well before hunt.
Closer to actual Hunt, my teammates informed me about
the Bravo Awards (see also
their wiki),
a set of Oscar-style awards for puzzle hunts of 2024, with
nominations and then voting very shortly before actual Hunt, and
winners announced literally days before Hunt. Self-nominations
were explicitly allowed and encouraged, so some of my teammates
and I nominated and voted for some of our things (as well as votes
for all the other nominees since it was ranked choice voting). And
THIS VERY BLOG (or at least the
2024 Hunt
entry in it) won an award, I won the Lifetime Achievement Award due
to my work creating and maintaining the
Hunt Index
and writing four Mystery Hunts, and the
Hell,
MI round
(designed and to a large extent constructed by me) won Round of
the Year. In addition, some of my teammates won for Best
Logic Puzzle
(Jigsaw Slitherlink)
and Best Introductory Puzzle
(Badges
Badges Badges Badges Badges Badges Badges Badges Badges).
Finally, just before Hunt, we got an invitation which was double-sided, on one side inviting us to the Mystery Hunt, and on the other, Robert Finster, the leader of the crime gang from the pre-Hunt, inviting us to a gala for the engagement of his daughter Gladys Finster to someone named Ferdinand Carter, at the same time, of course. This was only an engagement and not a wedding like the wedding of Mario and Peach (that Bowser crashed and kidnapped the Princess, as Bowser does, in 2011), as well as the wedding from the kickoff of Penny Park in 2020, but it was still a familiar setup. We were promised a weekend of 30's, 40's, and 50's glamour, and cocktail attire was recommended for kickoff. I didn't think I had anything appropriate, except somewhere around here I know I have a fedora, but I didn't dig it up since I figured I wouldn't attend kickoff in person anyway. John showed up in a black zoot suit with widely spaced vertical gold stripes that was perfect.
Some people make an even bigger event out of Hunt by coming days
before and staying days afterward, but I really don't do that. In
fact, I think I've only ever stayed in Cambridge the night before
Hunt on years when we were constructing and trying to get together
for last minute construction fixups, hiding of stuff on campus,
etc. MIT closing the campus since COVID has made it easier for me
to decide that I will drive down Hunt morning, stay in a hotel only
Friday and Saturday nights, and leave when we finish Sunday or
when it is time to close up HQ Sunday night.
So I wasn't there in person for the How-to-Hunt workshop run by John and others of my team, but I watched the stream via Zoom. John presented a new puzzle designed to illustrate common features of Mystery Hunt puzzles, and introduced an alternative to the ISIS (identify, sort, index, solve) acronym due to it also now being associated with a terrorist group, EASIER (Examine, Associate/Aha!, Sort, Index, Extract, Retry), the last step being the way that many Hunt puzzles make you do a similar thing to what you did the first time a second time in order to complete the puzzle.
I packed up the car Friday morning, drove down and dropped off my stuff with a teammate at the Killian Court loading zone at 11, went and parked legally elsewhere, and walked back to campus to get set up. Having Kresge closed and kickoff being run in a small room with simulcasts set up in two other rooms on campus made it easy for me to decide to stay in HQ during kickoff and watch the stream. You can watch the recording on YouTube.
This other character Ferdinand turned out to be the heir to some Jewelry chain (I think the name Carter was supposed to remind us of Cartier), and with their marriage, Finster's front business was supposed to merge with the jewelry chain. To celebrate, they were going to display the Shadow Diamond (a large diamond with a weird history filled with unlucky occurrences, reminiscent of the Hope Diamond). It was supposed to be in a glass case covered with a dark cloth, but when the cloth was removed, the case was empty. Absolutely nobody was surprised, except the characters on stage.
At this point, the entire video feed turned black and white, and very noir. A detective was brought in to investigate the stolen diamond. And everybody on stage at the Gala was considered a suspect, and it was assumed that only those people had access to it. So the Hunt was to be a Hunt for this diamond, and we, the solvers, were recruited as detectives-in-training at the 2 Pi Noir detective agency (punning on 2 pi as the angular measure of a complete revolution and the use of P.I. to mean private investigator) to help with the case. The now-familiar legalese followed, health and safety, don't do illegal or stupid things, there is no alcohol allowed on campus, etc.
We also learned some interesting features of this Hunt.
There was going to be an open HQ on the 4th floor of Stata, where the Gala was going to continue as a in-Hunt-themed bar (with no actual alcohol, since as they just told us, there is no alcohol allowed on campus). This is where we would go to pick up physical puzzles and do certain interactions. I didn't really do these interactions, but I saw many of the physical puzzles and there were some great, often thematic interactions.
This was going to be, as Death and Mayhem described it during wrapup, a Choose-Your-Own-Adventure type of Hunt. We were going to have a number of keys that would unlock puzzles for us, to be returned when we solved those puzzles. We would have few opportunities to obtain additional keys, meaning that the number of puzzles open at one time during the hunt was going to be strictly limited. We would get to choose which puzzles, out of those we had access to, we would use those keys on to unlock.
This mechanism of letting solvers choose their own path through the Hunt failed disastrously during the 2004 Time Bandits Hunt, in which, after the first round, solvers got to pick the order in which they opened other rounds, but you needed to solve a significant portion of the selected round to get access to the next one, and a majority of solvers picked what was probably the toughest round to do second.
This year's version had several features to help balance the difficulties such a mechanism might entail. First off, we were unlocking by puzzle only, rather than by round, meaning that there was much less chance of getting bottlenecked by a hard round, and also that the rounds still opened more or less in a planned order, with us simply choosing the puzzles to work on within those rounds. And we were going to get a description of each puzzle before we unlocked it, so we did not have to choose by name alone. Finally, though not mentioned in the description of how keys worked (or I missed it), some puzzles unlocked automatically when we got access to them, neither requiring keys nor giving them when solved. This included all metapuzzles and most time-sensitive puzzles, such as ones requiring an interaction.
Also, if you did not have the round for a time-sensitive puzzle open, it appeared under a page called Stray Leads and subsequently moved into the correct round when that round opened. Historically, the way this was handled was by doing timed round unlocks when a timed puzzle was supposed to be released, In recent years we've seen mechanisms more like this, such as in 2022's Bookspace Hunt, in which the scavenger hunt, Book Reports, appeared at the top of the Ministry round but without any way to submit it. It showed up as a regular puzzle in a later round.
Finally, there was the Radio. This was a physical device they provided each team with, a really elaborate artifact done up in the style of a period radio. As a radio should do, it plays audio. But of course it is a puzzle radio, so there were a number of puzzles that used it somehow. They also provided a virtual radio on the web site that could be used to hear the same audio. I don't know the number of puzzles that used it, but I'll eventually figure that out reading through recaps and the solutions. Some of them required tuning the radio to a particular station; it had FM and PM bands (P for puzzle? for P.I.? because PM, like FM, is the opposite of AM? Could be any of those) and the FM was just a regular FM radio, allowing the device to not be useless after Hunt ends, while all the puzzle content was on the PM band, which had stations marked off in fractions of pi from pi to 2pi.
The hunt started, and for once, there was no server downtime at the start of the Hunt due to bandwidth issues. Whoever they used as provider was up to the task.
You will either need to be logged into your team's account or the public access to follow the links on puzzle titles, which go to those puzzles' pages. For public access, go to the main page and click the puzzle to log in as public access.
All the puzzles we got early on were in The Missing Diamond round. From here on out, most puzzle writeups will have marked spoilers.
This was the first puzzle I worked on, a series of crossword clues with a set of number triplets at the bottom. (As a result of the unlocking mechanism, this hunt has explicit, official nonspoily descriptions of each puzzle, but sometimes I'll provide a bit more.
A grid of city names and sharks. Most of the squares were blank, but a number of themcontained city names from around the world, and half as many contained pictures of the same shark, oriented in different directions. When we finished Downright Backwards, I jumped here.
I didn't work on this puzzle, but I saw it come up early and announced that everybody with "Puzzle with name written entirely in emoji" on their bingo card should mark off the square
Except I did work on it, because this puzzle (read the title as copypasta) was about sharing with other teams. The puzzle asked you to send your text to other teams, who would reply with what they got. In that way, teams could find all nine pieces of the puzzle.
This left out people on small teams who don't know other Mystery Hunters, but for most teams the social connections should be there and they should at least know people on some other teams, and maybe ask those people for other teams' segments.
Only two people asked me specifically, so I sent what we got to those teams.
Copypasta refers to text memes, often with many interspersed emojis, that are copied and pasted around the Internet. There have been a couple teams who used copypastas as their names in recent Hunts, for example (see my 2024 Hunt blog). The text in this puzzle was of such a form. And the puzzle appeared in the first round so that many teams would unlock it early and other teams, upon receiving these texts, might unlock it just to participate.
I worked on the first part of this puzzle for a bit, helping to solve the Einstein logic puzzle part of it. I wasn't there when it was finished, nor when we picked up the second part of the puzzle, because I was distracted by....
A Pokémon puzzle.
Aside: I never played any Pokémon games until 2016, when Pokémon GO came out. Interestingly enough, the game came out during an NPL convention, and other members were saying that people were playing it around the hotel. I didn't know anything about it beforehand, unlike some of those who installed the game and starting playing on day 1, and I didn't check it out until aftre the Con ended. But upon seeing it was a game meant to be played walking around the real world, I decided it was a good incentive to get walking more. But I don't really tend to do things halfway. If I stick with something, I try to learn to do it well. So eight-and-a-half years and half a million Pokémon later, I've learned the names, appearance, typing, moves, interactions of typing, etc. of most of the Pokémon that exist.
Oh yeah, the puzzle. There was a lot here. A list of Pokémon, each of which has a name in addition to its species, one of which is marked as our starter and others in the wild that we can catch. There are also other trainers we can battle, and the same number of them as our starter plus all the wild Pokémon. But trying to say we catch these wild Pokémon in a specific order and then additionally beat one trainer with each Pokémon was was too ambiguous, especially only knowing the species of our Pokémon and the occupation of each trainer but not even what Pokémon they have. The weird names had to be important, somehow.
After that was solved, I looked at Introduction to Decryption briefly, and was interrupted by being asked to print several copies of Dropping the Ball. Then we got our first Stray Lead.
This puzzle asked us to send up to three people to take part in a quiz show based on estimation and with MIT knowledge being helpful which we being run every hour on the hour, and so I joined a couple teammates in that endeavor at 4. About 6 or 7 other teams showed up for this first session, and they really only had room for perhaps 12 teams in the room (one of the ones in Building 34) so I wasn't sure what they were going to do if too many teams showed up. Maybe they had other nearby rooms set up as overflow, or they'd limit teams to a single person or something.
We were asked a series of 17 questions, one at a time, with everybody writing down answers and then revealing them together. Most of the questions asked us to estimate some number, like the number of snare drums that could fit on the main stage at Kresge, in a hexagonal packing arrangement, without stacking them on top of one another. But four of the questions were "where is this on campus" and they showed us an image run through a filter that converted it into only a handful of basic shapes like ovals and triangles, each in a single solid color. Between us we did manage to recognize all the images, each of which we marked as a location on a laminated copy of the MIT map which we'd been provided for that purpose.
We went back to HQ, and shortly our puzzle updated with a set of images filtered in the same way as the four we were asked about during the quiz. I recognized a couple as album covers, even though I couldn't actually name any of the albums. My teammates got a few, but what we noticed was we were scored 0 to 5 on each question, and we got a better or worse version of the album cover based on how well we did. To try better the next time, we tried to write down the questions from memory and research their answers the best we could, finding the exact numbers for several of them.
Since we were still researching, we skipped the 5:00 session, and sent other people back at 6. They reported back that about half the questions were different, and they thought they did even worse than we did the first time. What's more, they got a completely different set of albums. And they didn't even write down their memory of the questions as we had. Between the setback of the questions changing and not knowing what we were supposed to do with the albums also changing, and also simply other puzzles being more attractive, we didn't send any more teams until Saturday, when we sent two more teams and then solved the puzzle.
When I got done researching the estimates, I joined other They Might Be Giants fans on my team working on this puzzle. And no, I'm not even marking that as a spoiler. Three of the four words of their name are right up there at the start of the puzzle title.
Aside: I go way back as a TMBG fan. I learned about the band in 1990, when a group of friends I met at college (Rice) introduced me to them. Back then, Flood was indeed a brand new album for 1990 like the theme song says, and there were far, far fewer TMBG songs than there are today. The actual number is hard to count, because it was more than just the ones on their first three albums. There were several EPs, their demo tape which was widely circulated as a bootleg, and songs that had only been released via Dial-a-Song, which was simply an answering machine that they invited you to call, with a rotating selection of their songs used in place of the message asking you to record a message. The original Dial-a-Song phone number became so famous due to the band using it in advertising that it was used in this puzzle as a clue to the band for those who missed it from the puzzle title.
Anyway, the song titles were clued, sometimes obliquely, as titles of fictional research papers, and they'd already gotten the easy ones, and I went back through and tried to help figure out some of the tougher ones, like...
The next puzzle I worked on was this word search.
This was the first meta-puzzle I worked on in this Hunt.
I didn't really work on this puzzle, but it's here because of how it ties in with what I did do—and didn't. There were 6 minipuzzles and a meta, but there were also 4 covered-up minipuzzles.
I worked on this puzzle from start to finish, the second to appear in this hunt with a name written entirely in emoji.
The second meta in this hunt that I worked on. I think we had only five of the answers when we started, and that kept us from seeing what we really needed.
We opened up The Background Check round with Knights of the Square Table at this point. It was an incredibly difficult consecutive + knight-adjacency sudoku with two breaks in the sequence of letters to put in, breaking up what letters could be considered consecutive. I let somebody else work on writing a program to solve it.
We also opened up The Stakeout and The Paper Trail and got a couple puzzles in each and a couple more Stray Leads that I didn't look at before bed.
A bunch of puzzles opened overnight, some of which were solved before I ever saw them. I got up to take the 6 AM shift as our strategist, an unexpectedly busy role with having to choose which puzzles to unlock with keys on top of the normal things. There were only a few of us around at 6, so after we got the room set up again, I didn't prioritize doing any interactive stuff until more people arrived.
I chatted with Barney, the preceding Strategist, on Discord before MIT opened, which was also at 6, so I could catch up on status. The Australian crew had solved a lot of puzzles overnight, 11 during his shift as Strategist, though that was in part because The Stakeout opened, a round of fish puzzles, and Barney opened a bunch of those for them as they seemed safe puzzles we were unlikely to get stuck on. But we solved other puzzles after I went to bed, and it was more like 20 solves while I slept.
This puzzle was partly solved overnight, but it had an interaction to do afterward. (Mild content warning for the puzzle, though it came without one.)
The person who came to mind, though, as MJ Andersen, who played Dionysus last year. It so happened he came in before 8, and I offered him the task, which he readily accepted. I also had him pick up a physical puzzle I unlocked at that time.
This was the physical puzzle I had MJ pickup. It was a miniature cereal box, like the single-serving boxes from the variety packs I remember as a kid and may still be available somewhere. It didn't have any cereal inside, just four mini crayons like the ones in kids' restaurant packs. But the entire box was covered in minipuzzles. I let other people work on it while I continued strategizing.
We unlocked this Friday night, and it was one of the things the Finsters who visited us around 8 AM suggested we focus on. I was being Strategist, and I hadn't seen the puzzle at all since it became available after I went to bed, so I mostly just directed some other people to look at it. Barney let me know that people had done the runaround part before campus closed, but the info was scattered between the spreadsheet, the puzzles' Discord channel, and the general on-location Discord channel, so as a first step I asked the people working on it to try and collect all the information into one place.
Spoiler below includes the answers for the four location metas in The Missing Diamond.
You can skip the above and go on with reading just about The Thief below.
People started this cryptic overnight. They had figured out the right idea, and mostly filled the grid, but were bad at collecting information, maybe because they were doing it in the wee hours. I worked on it off and on during my Strategist hours.
Somewhere around this time, we got a National Weather Service advisory for snow coming on Sunday. It said 6 to 10 inches. This was potentially very bad. If we got a blizzard and it was still blizzard when everybody got kicked out of campus at 1 AM, well, that was going to be a mess. But it was Sunday. And I didn't have the more detailed forecast, and I could look it up later. I was puzzling and didn't want to stop at that moment.
This puzzle opened automatically during my shift, a bit after 8. They were supposed to call us within an hour of it opening, and I figured John or Bella would get the call and talk to us, but they both slept through the calls, and shortly before 10 Nick had them change our callback number to his phone and we scheduled our interaction for 12:15, by which time many people were in HQ to participate.
There is not really any way to spoil this one, as it was a live event that can't be solved now. We had to send one escape room enthusiast to a room, but the rest of us were told to watch the puzzle page. We got video only (no audio) of our representative in an escape room. We were given the tasks to perform (more of which opened up as she made progress), and we had a control panel that let us send commands of the form "___ the ___" using words from a small selection provided to fill in the blanks. Some sort of display in the room updated periodically with the words we had pressed most since the last update. Of course, the words provided didn't exactly match the tasks required, and sometimes we had to break down tasks into steps we could clue for her. One example was that we needed to water the plant, but we weren't given "water" or even "feed" as a verb. The only water-related verb was "quaff" so we had to send "quaff the plant" and let her figure it out. Eventually she got it all done.
Another puzzle that opened up automatically, a bit after Control Room. This was one of the puzzles using the radio. There were a few of these Friday night that involved learning how to interact with the radio in various ways to make it play various musical notes, and Todd was our lead musician on this so I waited until he was around to give him the puzzle.
Ten clues written in large letters with Vs in place of Us, and ten more clues the same way but in smaller letters.
I worked on this cryptic early after opening it. We got most of the numbered clues and figured out what 10 and 15 were, but we had almost none of the perimeter clues (I solved the shortest and we had one or two others). As a result, we didn't have the theme yet when I stopped to work on Control Room, and I went on to look at another puzzle afterward.
This was a bunch a Nikoli-style logic puzzles. Several of us were working on these separately. I jumped in to take out some of the very difficult ones my teammates were unwilling or unable to finish: Japanese Sums or Products, Hungarian Tapa, and Index Yajilin. The last of these I only saw a single example of in Google and it was on a site which didn't provide the rules, but I inferred from the example that the numbers were the sum of the distances to the black cells each arrow points at.
Eventually we got enough of them solved to index into the final grid...
When I got here, my teammates had already identified the source and solved most of the numbered clues, and then seemingly abandoned it.
I was working on this one at the start, a set of nine boxes you can click on that change among three color states, and also change the colors of another set of nine boxes in a fashion not immediately understood, as well as a row of nine black or white lights below them. I identified a bunch of the behavior and some of the rules exactly, but I stopped when I couldn't figure out any more. It was hours later before someone finished it.
I spent a while trying to fill the 4x4 grid and failing...
I worked on this with Nick and one other person, I forget who, shortly after they got the physical component and were still assembling the pieces. I am, after all, one of the 3D-thinking specilists on my team.
Solving the six screens worth of crossword clues was a bit of slog, but I understood quickly how they were meant to be entered on the shapes. Explaining that to my teammates took a little more doing.
More of my teammates stayed up until 1, or even later, solving from hotel rooms, so Sunday morning was very empty for most of my shift. The Australian crew had solved few puzzles overnight, though the locals got a few.
I took the time before heading into campus to check the detailed forecast. Yes, 6-10 inches of snow was forecast for Cambridge, but we were near the boundary with 3-5 inches. And it wasn't a blizzard. That snow was forecast to fall starting around 1, but with the main part of the snow at 5-10 PM. Probably 1 inch an hour at the peak, and winds were forecast 9-12 MPH during that time. Not a blizzard. I also checked Cambridge's web site. There was no parking ban scheduled. With the NWS advisory having gone out almost 24 hours before, if they were going to issue a parking ban, they would already have scheduled it. So they didn't think it was a bad storm either. This reassured me that we were getting "just snow" and I made my plan to solve until the main blast of the snow was done, and then pack up my stuff and go home.
The group that arrived in the morning with me noticed something about The Illegal Search round.
This one opened automatically while we were getting into the room in the morning. It was another radio one: Play a duet with another team on the radios. So I just had to wait for our radio man to show up and hand it over.
This was another of those automatically-opening puzzles. It invited us to make an appointment to submit the Infinite Scavenger Hunt, but I couldn't see that we'd ever received such a thing. When we got done cleaning up all the other overnight detritus and resetting our room to its working configuration, someone called in and confirmed that we didn't have to have anything prepared in advance, and that it was a non-traditional scavenger hunt, so we had some people go do it.
We discovered this physical puzzle overnight, so it was more of my task management to send somebody to open the puzzle and send somebody to retrieve it as soon as we had anybody going to the Gala for any reason. I only worked on initial investigation of the puzzle after we got the thing, over 100 feet of ribbon with 19 paper tags attached... or it was supposed to be, but ours only had 18, so we had to send somebody back to get an unbroken copy.
The flavor text said it's also not Bananagrams, but the overnight crew that worked on this puzzle figured out what game it was. They just weren't very good at it or didn't try hard enough.
Ah, a Mystery Hunt tradition.
And we've also had a bunch of "10000" puzzles (including one elsewhere in this year's Hunt that I did not look at) and other puzzles in this vein. So yeah, I looked at this one.
We didn't finish this puzzle during the hunt. I came back later to continue, but the spoiler block here shows our status when I looked at the puzzle while doing this writeup.
###########
###...*####
##....#####
#......####
#...#...###
#.......#*#
#*#.......#
####......#
######....#
#####*...##
###########
Except that one or more of the three-letter entries at the edges are
connected to another three letters of another of these blocks at the
places I have marked with a * to make 6-letter entries.
I came back to this puzzle on the evening of Sunday, January 26th after finishing my first draft of this post, and my subsequent work is at the end of the post.
I think I took a Stakeout puzzle at this point because I hoped to finish it quickly and take another look at the World's Largest. I was solving this dogpile-style in the spreadsheet with several teammates.
This puzzle was one which is mock-backwards. We have a bunch of mostly arbitrary two-word phrases which are all anagrams of each other, and there are blanks next to each for the words of a clue phrase to clue that anagram. Many of them are highlighted in yellow. We also have a list of all the words for the clue phrases and simply have to find a way to use all the words once each to make suitable clues.
Seeing no new ideas on World's Largest, I helped do the data extraction from the video for this puzzle, which turned out to be about
I didn't really know those, so I moved on at this point, glad to have helped somebody along.
This might be a minor spoiler to some people, but it was meant to be clear to solvers. This physical puzzle consisted of a large bag of laser-cut acrylic tiles that were all the same. They were in the shape of the Spectre aperodic monotile discovered in 2023. (See the name right up there in the title?)
I haven't told this story before, but this is a good place to tell it. Prepare for the longest aside of this post.
The "hat" tile, the first aperiodic monotile, was discovered during 2023 by people I met previously at the Gathering for Gardner, an event for Martin Gardner enthusiasts. Some of you probably already know what "aperiodic monotile" means and heard about the discover of the hat, and you can skip the next four paragraphs. If not, but you have heard of Penrose tiles, just skip two paragraphs.
For the rest of you, first understand that we're talking about tilings of the plane: arrangements of repeated shapes that completely fill an infinite plane in all directions without overlaps or holes. There are many familiar periodic tilings, such as the lattice of squares on a sheet of graph paper (if we assume the lattice keeps going infinitely beyond the edges of the paper). These are periodic in that there is a repeating unit, in this case a single square, that tiles the plane simply by applying the same two translations, that is, by making copies of this unit and, keeping them in the same orientation, sliding the copies in two directions by multiples of the shortest move needed, so as to fill the plane. It's possible to take such a tiling and make it aperiodic, for instance, by shifting certain rows of squares in an aperiodic way, but that's not what we mean when we're talking about aperiodic tilings, because the square can tile the plane in a periodic way. We're looking for shapes that can tile the plane, but only in an aperiodic way.
It wasn't until the 1960s that mathematicians proved that there were sets of tiles that could tile the plane only aperiodically. The first such sets were huge, containing thousands of different tiles, but these numbers were quickly reduced to just 6 different tiles. It was in the 1970s that mathematician Roger Penrose found the first set of only two tiles that tile the plane only aperiodically. Martin Gardner reported this in his "Mathematical Games" column in Scientific American magazine in 1977. And ever since then, people have been trying to find one tile that tiles the plane only aperiodically.
What made the discovery of the hat even more interesting was that it was discovered by an amateur mathematician, someone like myself, who consulted some professional mathematicians to help him prove it. And it's a relatively simple shape, a polyform. Polyforms are polygons formed by combining copies of a single simple shape edge-to-edge, such as the polyominoes formed by groups of squares. There are also polyiamonds based on equilateral triangles and many other polyforms. The hat is an octakite, a shape made from 8 kites, the kite in this case being one-third of an equilateral triangle divided by the perpendicular bisectors of its edges meeting in the center of the triangle. I have personally worked on tiling problems using tiles as complex as this one. This is absolutely something that I could have found myself, if I had only been looking at the right shape.
Polykites like this tile the plane in a way where the individual kites remain aligned to a grid of equilateral triangles each divided into 3 kites as I described above. The triangles tile the plane in the usual way. The hat, as one of these polykites, forms tilings like this. Despite this underlying regularity, the positions of the hat tiles themselves are aperiodic. While you can find patches of hat tiles that look similar, that similarity breaks as you extend to larger groups of tiles, and the hat cannot tile the plane in a completely periodic way.
It was while my team was constructing the 2024 MIT Mystery Hunt that this was discovered, and within days I put in a placeholder puzzle idea to write a puzzle based on it. As soon as I found the time (when not working on other puzzles) I started investigating the hat, learning how it tiled, and how I might use it as the basis for a puzzle.
Through my G4G connections, I saw presentations from the authors of their proof that this shape tiled the plane only in aperiodic ways, and how they investigated the shape and developed the proof. The first to investigate the claim by the discoverer that this shape tiled the plane only aperiodically subjected it to two tests by different kinds of tiling programs. One program looked for ways to fill a cell, one of the repeating units that can tile by translation, building up to larger and larger portions of the underlying kite-triangle lattice without finding one, to the point that if there was one, it would have set a record for the largest minimal cell. The other program was designed for shapes that don't tile the plane, and attempted to compute the Heesch number of the tile. This is effectively how many times you can surround a tile with copies of itself, each layer completely enclosing the previous layer without any holes or overlaps. A shape that tiles the plane has an infinite Heesch number because you can just keep going indefinitely. Among shapes that don't tile the plane, the largest known Heesch number was 6. His program had gotten up to an arrangement of hats that showed its Heesch number was at least 16. At this point, he knew the discoverer had found something special. Either he had completely blown away the record Heesch number, or he had broken the record for the largest minimal cell, or he had indeed found a single shape that only tiles aperiodically. Then they spent months trying to figure out how the tiling actually worked.
They discovered some interesting behavior in the tiling. The hat tile is chiral, meaning it differs from its mirror image, and the tiling requires a fraction of the tiles to be in the mirror form compared with the large majority of other tiles. These mirrored tiles cannot touch, and the tile shape forces certain other tiles to be in particular orientations adjacent to the mirrored tile. These forced tiles, along with other nearby tiles that may have multiple but limited numbers of choices, combine to form what the authors of the proof call supertiles. The entire tiling can be broken up into copies of just two different, repeated supertiles, but it's still not periodic because the supertiles form another aperiodic tiling system which works by combining them only in certain ways to form even larger supertiles, and so on. Each level of supertiles has a different shape, but they all tile in the same manner. Because they form these ever-larger supertiles that continue to tile only in this way, there can never be an overall periodicity in the tiling.
So I started developing an idea similar to some of the fractal or "infinite" puzzles such as 2013's In the Details, a fractal word search, or 2005's Maze, both by Derek Kisman. The idea would be that you can look at only a small portion of a tiling of hats, but you have to figure out where you are among the supertiles many levels up by examining different areas. The small areas, I determined, should be constructible by defining the placement of the supertiles first, and dividing them into smaller supertiles, discarding ones too far from the area of interest, and iterating until we get down to the individual hats, then drawing the resulting hats in the selected region.
The idea I had was that we'd give solvers an applet that lets them see the hats at a position they could specify, and some number of coordinate pairs they could plug into this applet, probably accompanied pre-made pictures of the hats at each position. Something would tell them how many levels of supertiles up they had to look, say 5, and the two kinds of supertile at each level would be assigned 0 and 1, giving a 5-bit binary result for each coordinate pair. By doing the reverse, starting from the supertile 5 levels up and choosing which kind of tile I wanted at each level, I could find coordinates for any binary string I wanted.
I wasn't entirely satisfied with this idea, but there have been other mathematical Hunt puzzles in the past like this that basically turned into a program you had to write to solve a particular mathematical puzzle, and I set it aside, hoping to find a better idea.
Before I had any such idea, there were more mathematical discoveries. First, there was a second tile that tiles aperiodically in the same way as the hat. Then, these two tiles were discovered to be examples of an infinite family of tiles, no longer polykites, which can be described in terms of a continuously varying parameter. And finally there was the spectre tile, a single tile that tiles the plane only aperiodically but without the need for the use of mirror tiles. The state of the art in aperiodic tiling was changing faster than I could write a puzzle!
These other tiles were harder to work with. While it was probably still possible to do the thing I was thinking about, it was harder for me to visualize the math because they didn't have the underlying lattice of kites forming triangles that would be the basis of a coordinate system I could locate all the tiles in. So, I thought, maybe I should do something else entirely using spectre, or maybe I should ignore the later enhancements and use the first aperiodic monotile that I had an idea how to use.
The best way to proceed wasn't clear, but what happened instead was that I got involved in writing the Hell, MI round and had to drop the idea entirely. That is why you did not get an aperiodic monotile puzzle in the 2024 MIT Mystery Hunt.
So, now, back to this puzzle. My teammates solved the jigsaw they had been provided with: a couple hundred spectres and a few larger pieces that formed a frame that allowed the spectres to fit together in exactly one way. Along with the pieces came a printed sheet of cardstock with an outline exactly matching the interior edge of the frame pieces, and 10 lines of similar but not identical lengths. I had been working on other things, and by this point Cardinality had found the coin and we were just playing for fun.
When I stopped trying to slog through Inspectre, the strategic focus was on the three metas for The Background Check that opened for us together a bit after 4 PM Sunday.
By the time I looked at it after dinner, with hints being made more widely available now that the winners were through, we had hinted our way to solving all the feeder puzzles, and we were left with a collection of very bizarre answers. The round page associated the puzzles with three different newspapers, and we believed each newspaper went to a different meta. We were most confused by The Mark which appeared to be a hex grid to spell out words Boggle-style, though since "24 unique letters are used" and we have only 21 spaces, it appeared some of the "pieces" we want to use contain multiple letters.
Eventually we reached a contradiction, in that a sequence of 3 letters appearing in our answers needed to be connected by both ends with three different pairs of letters, and regardless how many of the letters were combined on the same space, with all the letters being unique it simply was not possible. So we called in a hint, which was responded to quickly. The result of that response was to reveal the most outlandish thing Death and Mayhem had done for this Hunt, and perhaps the most outlandish any Hunt-running team has ever done.
This, needless to say, really soured my mood. But the result had given us info to proceed with on all three metas.
After calming down from the incredible BS that Death and Mayhem pulled with these metas, I looked at the info we got for all three of them and ended up working on this meta.
With HQ closed at 10, I made ready to leave. Clearly, we weren't going to finish tonight. The snow had not kept up with the forecast times; it didn't start at all until sometime after 8, and it was still light. So I packed up my stuff, pulled my car up to the loading zone, loaded, and drove home.
My teammates continued working on puzzles wherever they were, through the night, and past wrapup. I stayed home and watched wrapup online, neither wanting to pack into a crowded room nor go out again in snow that (at least in my neighborhood) hadn't really been cleaned up, and fight for parking spaces that might also not be cleared. I cleaned off my car but did not move it on Monday. And I didn't really go back to working on puzzles, opting instead to start this writeup and take care of all the things I had neglected in order to do a whole weekend of Mystery Hunt.
My teammates paged me on Wednesday evening to take a look at this puzzle.
On the evening of Sunday, January 26th, after finishing the first draft of this post, I went back to continue work on this puzzle. You can see where I was before earlier in the post. And this is really long, so I have broken it into multiple spoiler blocks.
I started work on a general solver (having in my earlier analysis turned the grid into text file mapping out the blocks) but I did not have time to finish it Tuesday night. Wednesday and Thursday I was looking at other puzzles and did not work on this one. By the time I resumed, we had finished every other puzzle in the hunt.
My end-of-month NPL duties interfered with my time on this puzzle as well, but by the morning of Sunday, February 2nd I had written a solver which solves the entire grid to a unique solution.
Then we were trying to make sense of the solution.
We got more people to look at it on the afternoon of Monday the 3rd, and we worked out what the flavor text was trying to say:
My teammates continued working through the Hunt and John brought this one to my attention on Wednesday, January 29th. It's a hexagonal Battleships variant, with a bunch more variants piled on, including movement rules that make the grids somewhat dependent on one another.
Unfortunately, John wasn't able to join me in the solving Wednesday, but I got them into the spreadsheet and got two of the 10 grids completely solved and progress on others.
The puzzle comes with a note advising against solving it in a spreadsheet. I discovered that that was because Google Sheets is glitchy as heck working with the hex grids.
Thursday night we met up and finished the puzzle and it was surprisingly straightforward for such a bizarre Battleships variant. and I guessed the extraction correctly before any of Thursday's work.
When we finished Bermuda Triangle, I went on with John and others to look at this meta since it was the last answer we needed for it.
This opened up another hidden round of puzzles which my teammates managed to finish during the day Friday.