On the MIT Mystery Hunt 2026

The MIT Mystery Hunt 2026 marks my seventh year of participating in the Mystery Hunt, making this post my seventh writeup. Each of my previous posts has examined it, and puzzlehunting in general, with a different lens: 2020 on the joy of solving, 2021 on the process of writing, 2022 on the notion of community, 2023 on issues of production and economy, 2024 on continuity and institutionality, and 2025 on solver frustration and effort. I leave the context of explaining what puzzlehunting is to my previous posts; perhaps 2022 has the best brief introduction.

I thought it’d be suitable, for my seventh year, to do a sort of sabbatical, where I write about the Hunt under each perspective. Partly because I didn’t have any other ideas on what to write about. Partly for my own vanity as well; I ask forgiveness for every time I’ll pat myself on the back. But partly because, this year in particular, I have something to say about each of these topics. Anyway.

I don’t spoil any puzzle-specific content. Thanks to ✈✈✈ Galactic Trendsetters ✈✈✈ for comments.

2021: Logistics

One of the FAQs for this year’s Hunt said:

The team I hunt with, ✈✈✈ Galactic Trendsetters ✈✈✈, didn’t do a lot of advanced coordination. We don’t have a lot of top-down coordination in general, I guess; rather, organization emerges when someone wants things to get organized, and people follow. So we didn’t talk about this much until after kickoff, when it was revealed what kind of organization we needed.

Around the first half of Hunt looked like this:

• A single-player game is used to “find” puzzles in a virtual world. The game was an RPG, where you controlled an avatar that walked around, could talk to characters, had tools, and could play minigames. This reminded me of Mystery Hunt 2021, which had a similar mechanic for “finding” puzzles; though that one involved a multiplayer world.

• Puzzles that are found start as locked. Locked puzzles have their title visible, as well as a brief description of what it looks like. Unlocking a puzzle costs a resource called Research Points, typically at a rate of 1 RP per puzzle, though some cost more. A maximum of eight puzzles can be unlocked and unsolved at a given time, though this number went up as the Hunt continued. This is like last year’s Hunt, where you used resources to choose which puzzles to unlock.

• The way to gain Research Points does not involve solving puzzles; instead, you gained RP by doing Research Tasks. Each task awarded some number of RP, ranging from 1 to 8, depending on its difficulty. Tasks ranged from things like “play this minigame”, to “submit a picture of a team member touching grass”, to “go to this event”.

Finding things in the game world didn’t need too much coordination. For unlocking puzzles, we had one person responsible for unlocking, and had a Discord channel where others made requests for puzzles to unlock. Doing Research Tasks needed more coordination than that, though: we didn’t want people to duplicate work, but we also needed to gain RP at a steady rate.

All that to say: I spent the first few hours of Hunt helping coordinate stuff. Putting tasks in a spreadsheet, making sure people claimed the tasks they were doing, getting people to do tasks when we needed more of them done, wrangling sign-ups for events, adding and tagging puzzles on our solving dashboard, that kind of thing. I didn’t get much puzzling done that first afternoon, but I enjoyed what I was doing.

This is the bucket of work I consider as logistics, which is about making sure the right people have the right things in the right places at the right times. A typical Hunt’s logistics include things like assigning classrooms to teams, organizing and distributing physical puzzles, coordinating with CAC to get event room layouts, finding staff for events and hints and interactions, ordering food for the team, and figuring out merch.

When we ran the Mystery Hunt in 2021, our logistics looked quite different. Physical puzzle distribution had to involve shipping things all over the world. Instead of managing physical rooms, we set up Zoom meetings: because some of our members didn’t have MIT Zoom accounts, someone who did had to start each meeting and hand off hosting. I guess staff schedules were kinda similar: we had a spreadsheet of times and roles, and people signed up for shifts.

I didn’t do that much logistics for the 2021 Hunt, but over the years I’ve grown to love this kind of work more. It’s satisfying watching things click into place. I feel powerful when people agree to do things I’ve asked them to. I imagine this is what conductors feel when they’re performing.

2024: Practices

When I wrote about the 2024 Hunt, I took an institutional perspective, analyzing how rules, practices, and narratives shape actors’ behaviors. In the context of Mystery Hunt, practices govern everything. They are enacted via consistent rehearsal and are sanctioned by social disapproval. Tension builds when these rules, practices, and narratives get out-of-sync, and this leads to change.

This year’s Mystery Hunt raises questions about many of these practices, some of which were brought into the spotlight in the past few years. We get a lovely opportunity to watch how institutional change happens—or perhaps, doesn’t happen. Some aspects to consider:

Choice in unlock order. In 2025, teams had keys, which they can use to choose which puzzles to unlock next. This received praise. As I’ve discussed, this year had a similar system—for the first half of the Hunt. This also received praise.

HQ location. For several years, the running team’s HQ was in the Bush Room, 10-105. Last year, Death and Mayhem chose an “open HQ” model, moving it east, to the Stata R&D commons. This received praise. This year, while Cardinality also used the Bush Room, they also had an open HQ in the Student Center. This also received praise, though a bit less than last year, I think?

The scavenger hunt. An old practice is for the Mystery Hunt to include some sort of “scavenger hunt” puzzle, where teams do a certain number of tasks, often involving creative prompts. This is a frequent target for change: 2020 used it to collect donations for local charities and 2025 gave it the twist of being randomized. This year’s scavenger-hunt-equivalent was spread out over various Research Tasks. This has received lots of praise, not only from me. So, it’ll be interesting to see if a similar thing happens in future years.

Hints. The first formal hint system spawned in 2015, where teams could ask yes/no questions on puzzles. No formal hint system was in place between 2016 and 2019. Since 2020, each Hunt had a formal hint system again, where teams could ask for a freeform hint on certain puzzles. This changes slightly year-to-year, as the writing team gets to decide when hints are released: after it’s unlocked for a while, after enough teams solve it, or after a manual decision. I see this as a relatively stable practice: I haven’t seen much change to formal hinting since then.

However, informal hinting—the kind that happens outside of these formal systems—is a practice that’s changed quite a bit year-to-year. In the 2025 Hunt, teams could go to the open HQ and ask for informal hints on puzzles, provided they weren’t one of the leading teams. At some point on Sunday, the “floodgates” opened, and hints were more freely available to all teams. By the end of Sunday night, the HQ was crammed with people asking for hints, with longer, back-and-forth interactions being common.

The narrative response to this was quite pronounced, as you could tell if you look at one of Dan Katz’s posts and searched for the word hint, which appears plenty of times in the comments. To flatten a lot of discussion, there’s a decent amount of people thinking that hints were given too aggressively that year, though also plenty who praised the abundance of hints.

This year’s Hunt continues in the practice of informal hinting, but not to the same extent as last year, with a huge open HQ with lots of hinters. However, there was still a definite “floodgates” point, where a lot more hints became available as well. The hint reception seems, at least to me, to average more positively. Sure, it’s a bit of a regression to the mean kinda thing. But I imagine that Cardinality’s approach to hinting was at least partly influenced by the discussions that happened last year.

2020: Taste

The bulk of my 2020 post talked about puzzles. When I read that post now, I see some ways in which my puzzlehunting tastes have changed, and some ways where it’s stayed the same:

• I participated in an event in 2020. This year, Cardinality ran ten distinct events, many, many times through the weekend. I participated in Puzzmon Yoga, which I enjoyed for being quite silly. In general, though, I’ve gotten less interested in doing Hunt events; I only did this one because nothing happens on Saturday morning.

• I alluded to Galactic’s meta squad, suggesting that I wasn’t on it. I enjoy working on metas a lot more these days. I’m a particular fan of pure metas, which is why I made meta-data and puzlink.js. That doesn’t mean I’m any good at them, though: I burned many hours trying to solve Mixed Messaging, Cryptozoologist, Let Her Cook(book)!, and the other first-half Kingdom metas. Pretty much none of my work ended up being useful. I love the Kingdom metas in general—Cryptozoologist is a particular favorite—but I’m a bit angry at myself with how I did on them.

• I talked about a puzzle that required MIT-specific knowledge. Puzzles related to MIT are still one of my favorite genres, because I love getting to show off my MIT knowledge. This year, I enjoyed crushing Go Back to Square Fourteen in 15 minutes; hell, I wrote a whole blog post on the puzzle’s topic.

• I discussed a puzzle that involved heraldry. In general, I still love puzzles about useless esoteric knowledge, though I feel like I didn’t get to use any of mine this year, outside of Go Back to Square Fourteen. When will the next square dancing puzzle come out?

• I praised Safari Adventure as the best meta of the 2020 Hunt, though in retrospect, what I loved was the weird round structure. This year gave us many rounds with weird round structures, of which you can find much praise from others. My only serious interaction with these was with Fate’s Thread Casino, which amused me for sharing a gacha mechanic with Galactic Puzzle Hunt 2024, and perhaps Atlas of Mosaics, which amused me for sharing a piece-placing mechanic with Galactic Puzzle Hunt 2025. To clarify, I thought Cardinality used these mechanics in quite different ways than how we used them in GPH, but the surface similarity is kinda funny.

While not being directly puzzle-related, perhaps the strongest constant since my first puzzlehunt in 2020 is how much I enjoy competing. When I read the section The beginning of the end, I still feel the palpable energy of being so close to finishing the Hunt. This year, we got told on Saturday night that we were in the lead by a tiny margin, which felt thrilling. In the end, we ended up finishing second.

2025: Frustration

Last year, I wrote about the experience of doing puzzles, and a particular topic I discussed was frustration. The strongest feeling of frustration I felt this year was while working on the Kingdom metas, as I talked about above. I didn’t even make any progress toward resolving these feelings. Someone else came up with the insights we needed.

The total frustration I felt from this year’s Hunt was less than what I felt last year. Which is to be expected: my experience with the 2025 Hunt—particularly with The Killer and The Grand Illusion metas—was negative enough that I wanted to make my previous post about puzzle-related experiences in general. The 2026 Hunt is an outlier in the opposite direction, though: it’s the Mystery Hunt I’ve felt the least total frustration with, so far.

The thing about being stuck on the Kingdom metas was that I didn’t have to be stuck on them. There’s plenty of other things I could’ve done to pull our team further ahead. Other than the Kingdom metas, the two puzzles I remember being stuck on are:

• Tribute Puzzle, where for a long while we knew what to do and couldn’t do it. By the clock, we were only stuck for around 30 minutes, which in puzzlehunt scale is quite short. It felt a worse mostly because, for most of that I was working alone. I also wasn’t that interested in the subject matter: I was working on it for the sake of progressing the Hunt.

• Balancing Act, my favorite single puzzle from this year’s Hunt. It’s one of those fantastic word puzzles that doesn’t seem tractable until you finish it. As I talked about in last year’s post, there’s a kind of difficulty that’s satisfying to resolve, and Balancing Act lands squarely in that bucket.

These two are the exceptions that prove the rule: in other years, this list would be much longer. This is an experience other teammates shared. For example, Brian said “this hunt felt unusually smooth.” Which felt good in-the-moment, yes, but somehow less satisfying after the fact. The feeling I’m left with reminds me of something I read in Level 51’s Galactic Puzzle Hunt 2024 writeup:

2023: Ephemerality

A central theme in my 2023 post is ephemerality and its consequences for the Mystery Hunt. In that section, I suggest that a puzzlehunt could choose to embrace ephemerality, delivering an experience that can’t be repeated. Much of what I enjoyed about this year’s Hunt is difficult or impossible to reproduce after the Hunt.

One thing I mentioned in the post was copresence, which refers to sharing the same space at the same time. The entire Atlas of Mosaics round makes use of this copresence, by giving each team a digital space of hexagons, that members could annotate and place pieces on.

Copresence is also inherent to events, and this year’s Hunt had ten distinct events over the weekend, most of which ran several times. I heard that Advanced Pictograms was quite fun, for example, and it also exploits physicality with its use of props. I’d hazard that Cardinality might have run the Hunt with the most events, but this is not a straightforward statistic, since interaction puzzles like 2025’s Control Room and In Communicado Tonight could also be considered events.

This quantity of events comes with the tradeoff of having less people available to run other kinds of interactions. Most of this year’s story was delivered via cutscene, rather than having cast members act out a scene live. I discuss this tradeoff in more detail in my 2025 post.

Another property that can lead to ephemerality is statefulness, where the presentation of a given puzzle depends on previous actions taken. Most puzzlehunts are stateful to some extent; for example, you might be unable to view a puzzle until you solve certain other puzzles. This kind of statefulness is, most of the time, easy to replicate after the Hunt ends: you can choose not to open a puzzle until you’ve unlocked it.

The Land of No Name gives an example of statefulness that’s hard to simulate manually. You can look at the Herman’s designer notes for a description. He posits that solvers found the round fun for two factors: the ability to shortcut the process of solving a puzzle, and the feeling of momentum that solving each puzzle brings.

Compare the Land of No Name with the Glasses round of Galactic Puzzle Hunt 2020. In Glasses, which was also the intro round of that hunt, each puzzle had a blurry version with partial information, and a clear version that unlocked after an hour. This enabled the shortcutting that Herman talked about, and a feeling of momentum comes from unlocking new puzzles with each solve.

Unlike Land of No Name, however, Glasses is less ephemeral. To simulate the experience after the Hunt, you can choose to look only at the blurry version until an hour passes. Any state is only puzzle-wide. To fully simulate Land of No Name, you would need to remove unavailable letters from all the puzzles, as the state is round-wide. And I’d argue that this ephemerality gives each Land of No Name solve more meaning—but perhaps I’ll explore this some other time.

2022: Community

The 70 people on Galactic’s roster this year range from five people who joined us for the first time, to the forty-ish people who’ve been on the team for at least five years. Around two-thirds of the team are MIT students or alums, most of which live, or have lived, on floorpi. These range from people who graduated in 2010, to people who will graduate in 2029. Fifty-ish people did the Hunt at least part of the time from campus, and the rest did Hunt remotely over a Discord server.

Consider the following questions.

Who are the ✈✈✈ Galactic Trendsetters ✈✈✈? Are we more than a group of people associated with floorpi? Are we more than the members in our team and the relationships between them? If we gain a few new members every year, how do you make them feel welcome, when so many people already know each other? In my 2022 post, my explanation of our culture amounted to listing memes and in-jokes: is that all there is? If residents of floorpi keep joining Galactic, and people keep inviting their friends, do we eventually become too big? How do you help team members hunting remotely feel included with our on-campus members?

Okay, you can stop considering.

I don’t know the answer to most of these questions, and I wish I did. I care about the experience of newer Galactic members, even if I don’t show this care as well as I want to. There’s one question here I can answer, though, and that’s an explanation of culture. I don’t think I did enough justice to the topic in my 2022 post. I want to give it another shot, because I care about this too.

One way to look at Galactic is through our shared interests. For any given thing, you can probably find someone in Galactic who’s interested in it, but there’s some things that lots of people are into: gaming (broadly construed), trivia, geography, crosswords, languages. If you pick a different large puzzlehunt team, and look at their shared interests—well, okay, you’d probably end up with a similar list, but it’d probably be in a different order.

Another way to look at Galactic is through our traditions. There’s small ones, like how we pronounce our team name: one person says “whoosh”, people might join in for “Galactic Trendsetters”, and by the end, everyone in earshot says “nyeeowww!” And there’s big ones, like writing the Galactic Puzzle Hunt together every year.

The people in Galactic interact in a way I’m struggling to put words to. My best description for it might be oscillating between throwing memes around and taking things seriously. That’s part of why I think we’re competitive in puzzlehunts: people like engaging with bits, or taking absurd ideas one step further, or finding humor out of something dry. It’s a mode of conversation that reminds me of two-in-the-morning chats in floorpi lounges.

2026: Gestalt

Jared Singer’s performance poem Love opens with:

Contrast Singer’s notion of love with the Mystery Hunt.

The Hunt is not experienced in parts. It is a large and messy whole. Like every game, it comes with its own magic circle, delineated by kickoff and campus and wrapup. Within this circle, only the puzzlehunt exists.

The Hunt, as a whole, makes no sense. It is an event with thousands of participants, whose organizers change every year. It costs tens of thousands of person-hours to produce, and it is free. It is a miracle that the Hunt happens at all, and this is a thought we must constantly remind ourselves, lest we grow cold.

The Hunt is composed of parts that, in isolation, can be understood. It is complex, not complicated. And so I have shined lights on it, from many directions, trying to understand. I consider these posts as the partial products of my undertaking.

I do not want to mislead. By nature of form, any long-enough description of the Hunt will split it into parts, each which can be analyzed in isolation. Look at any writeup or wrapup. See how it tries to decompose: puzzles, metas, events, rounds; travel, meetups, kickoff, wrapup; story, art, tech, ops; teams, communities, economies, institutions; people, relationships, moments, feelings. The Mystery Hunt is built from these things. The Mystery Hunt cannot be replaced with them.