Blog: Expanded Universe

MIT Mystery Hunt 2025: grab bag 1

(if you opened this post, have no idea what any of this is, and want to learn more, read Introduction to Puzzlehunts probably?)

meta (of the about this blog variety)

I don't know if anybody other than myself noticed that I never published a blog post about the 2024 Mystery Hunt. I felt like I hadn't fleshed it out to the extent of any of my other Mystery Hunt posts, but I also just ran out of steam, which isn't a good excuse but there you go. I could still decide to publish it though! There are no rules.

Looking at the big picture, I also feel foolish floundering under the tyrannical precedent of my own past actions:1 I've written blog posts about almost every Mystery Hunt I've done, totaling 20,000+ words, and basically no other puzzlehunt. If nothing else, it's hard to justify not having posted about any of the Galactic Puzzle Hunts I've helped write. (An excuse for that is that I write in our hunt wrap-ups instead, but surely there are additional things I'd want to say as an individual.)

Anyway, I'm taking the opportunity of the new blog to write a bunch of disjoint, vignette-sized sections about the Mystery Hunt I just did (with ✈✈✈ Galactic Trendsetters ✈✈✈, as usual), possibly across multiple posts but publishing each post without planning for future ones, instead of one big post with a somewhat formulaic structure like before.

Nobody can stop me. Not even myself.

meta (about the puzzling experience, but not the puzzle variety)

Recently Sniper pointed out to me that it is kinda silly to fly so many hours to MIT and spend hundreds or thousands of dollars on lodging and travel, to have a few days within walking distance of a ton of your puzzler friends from all across the world, and then spend 95% of this time in your room solving puzzles with only your own team.

This is obvious in hindsight, but it didn't really describe me a few years ago and maybe it makes sense that I didn't previously think this way. For my first few in-person hunts, I lived and studied at MIT and thought the coolest thing about the MIT Mystery Hunt was the puzzles. I met some Galactic-affiliated cruft2 incidentally during hunt, and I had heard of some other puzzlers and seen them online, but few people knew who I was and I wasn't confident enough to introduce myself to anybody, much less try to get to know them; not to mention, since I had things to do at MIT before and after hunt, my scheduling was almost never the bottleneck for meeting people.

The first year I participated as an MIT alumnus rather than student (2021) might have been a wake-up call, except that it was a doubly weird year for me since there was a global pandemic and also we were writing hunt. In any case, over the years I've gotten to know dozens of puzzlers on other teams, and for many of them, this is the one time a year I get to see them in person. Revealed preferences: I slept more during either night of hunt than either of the two nights after, because I was up talking to people until 2–3am both of those nights, and got up at like 8am on Monday to talk to a different set of people. I'm very happy I did this!

I don't have a concrete proposal in mind or anything. Opportunities for inter-team interactions are nice, especially when they don't require much effort during hunt to run (like with the early puzzle 📑🍝), but I expect most such opportunities are difficult to scale and to staff, plus they can be harder or at least more intimidating for solvers new to Mystery Hunt (which includes many an MIT student) who might not yet know people on other teams. It would be cool if more people stayed longer before and after hunt to hang out, but I have the privilege of having a lot of flexibility in taking time off to do this, which not everybody else does.

meta (about my team's puzzling experience, still not the puzzle variety)

This was another pretty chill year in terms of how hard Galactic was solving. A few key solvers had been recruited to other teams, and quite a few more stayed home for various reasons, so our classrooms were a bit emptier than in the past.

We organized our puzzle solving with Galackboard. By which I mean, we re-forked upstream Blackboard after it had been upgraded to Meteor 3.0 so it could run on a modern version of Node (which was a huge rewrite — major props to Torgen of Codex Frumious for doing and open-sourcing the hard work!), and manually reapplied/reimplemented a bunch of the changes we wanted, because for better or worse, Galactic is much less logistically/procedurally strict when solving than what Blackboard seems to be designed for. (Simply merging/rebasing was doomed because, among other things, Blackboard had un-CoffeeScript-ed a bunch of its code since our codebases diverged.)

Azalea also brought bingo boards for Galactic solvers to fill out, with tasks about solving different kinds of puzzles or aspects of hunt with different people, to slightly encourage that to happen. I filled out my bingo board fully (though some squares were pretty dubious — the most of which, funnily enough, might be "identify an image").

puzzle with ian or josh ask someone for help with something take a meal break work on a puzzle genre you’re bad at let an additional person join your group working on a puzzle
work on a puzzle genre you’re good at check your work work on a cryptics puzzle go back to a puzzle we’ve been stuck on ask to join a group that is already working on a puzzle
help with the scavenger hunt learn something new about a teammate FREE SQUARE: work on a puzzle go on a runaround work on a physical puzzle
work on a word puzzle call in an answer explain progress on a puzzle to someone else work on a pop culture puzzle use nutrimatic or onelook
introduce yourself to a teammate be wrong about what’s going on in a puzzle ask someone to explain progress on a puzzle to you identify an image puzzle with patrick or brian

I slept a lot.

We finished our last two metas, The Murder in Metropolis and The Background Check, on Sunday night at 10:30pm and 10:40pm respectively, with quite generous hints from D&M. So we were pretty far from plausibly finishing on time without hints by when HQ closed at 10pm. Props to them for setting boundaries for closing up and sticking to them though.

cool puzzles (!)

The funniest moment of the hunt for me was in Shell Corporation 3: Superior Stonework:

Spoilers for Shell Corporation 3

I asked around for how to help with metas and was told, without much context, that we had to solve a short Vigenère ciphertext whose key was a six-letter word. (Just try every key! said multiple people, some meaning it more seriously than others.) I decided to try doing so finally downloading solvertools, which I knew had a cromulence function. It turns out solvertools also does Vigenère decryption! As a result, writing and running my solve script was about 20× easier than installing solvertools (protip: it uses a ton of data; download it ahead of time). My script output the answer almost immediately.

Only afterwards did I understand/appreciate that this was probably not the intended solve experience, in that we were meant to get some information about the key from the feeders.

In general the Shell Corporation metas were very cool! I didn't work on them that much though.

Absolutely Not Balderdash is, even with its obviously contrived clues, a miraculously constructed puzzle. (I have to say that the miraculousness of the construction doesn't translate 1:1 into the fun of solving it — which is not to say solving it isn't fun, just that the fun is an 8 or 9 while the construction is a 10.)

Garden Anecdotes is a cute, novel "classical cryptography" puzzle. I had the pleasure of making the final successful attempt to assemble the final answer, about which somebody wrote, "I was here while Brian pumped3".

O, Woe is Me is a cryptic crossword. I heard mixed reviews from others about the actual theme of the puzzle, but I find it incredibly funny. It simply would not occur to me to write a puzzle about it, and if it did I would have no idea how to do so mechanically.

Mild spoilers for O, Woe is Me (just eliminating a false hypothesis) Also, a small part of my brain immediately wondered whether the puzzle was about "owo". It really didn't help when the first entry we put into the grid was "cowork"! (I know the same thing happened to other teams.) Though, I never fully believed it because I had searched wordlists for words and phrases with the "owo" substring a while ago to construct another puzzle, and thought I would recognize more than one definition half if the puzzle really were about that. (Indeed, it wasn't.)

I thought World's Largest Crossword Puzzle was very cool, though it was saturated with programmers so I didn't really work on it until we got stuck on extraction.

Spoilers for World's Largest Crossword Puzzle We had obtained a 5×5 metametagrid of 6×6 metagrids of bits with constraints, and ubuntor had written a z3 script that directly found a solution for all 900 bits, which we had been staring at without any success.

I transcribed the metagrids' constraints, realized there were few distinct ones, and asked somebody to find all of each metagrid's solutions, with the theory being that the metagrids were themselves logic gates we might assemble into a larger metametagrid than what the puzzle provided. The latter theory didn't work out, but with the metagrid solutions in hand I highlighted the bits that were connected between metagrids in our 900-bit solution to double-check that they matched, then realized that the puzzle might be designed to draw solvers' attention to those connections.

In hindsight I think those connecting bits would have been more salient if we hadn't z3-bombed the metagrids and metametagrid in one step; in particular, if we were "worse at programming".

I did not do The 10000-Sheet Excel File or _land but heard good reviews and will probably want to post-solve them. (There are many more such puzzles; this is just the "more than one recommendation" shortlist!)

puzzles that have caused amendments to Have You Tried?

World's Largest Crossword Puzzle goes here too but I won't recap it.

Incognito:

Spoilers for Incognito The first person who solved the logic puzzle put down the intended guilty party as intended, but when we failed to extract, we argued about whether housemates and neighbors were the same thing. I felt clever when I proposed that "neighbor" and "housemate" were meant to be mutually exclusive categories rather than synonyms, and that Eddie was actually guilty because his innocence was described as an "inspector's conclusion" and the inspector had been misled or was a co-conspirator somehow, while every other innocence followed logically from the crossword's direct assertions. (I am told other teams went down a similar false path.) Of course this was totally wrong and the original solution was the intended one. Another rabbithole I went down was second-guessing which names in the puzzle were first and last names, especially in an attempt to make the names spell something with their first letters. I went to sleep on Saturday thinking about this puzzle, and woke up thinking about this puzzle, with still no luck.

After a day (and 30,000+ solver-minutes, according to Galackboard) we ended up backsolving, after which I found our mistake in like two minutes, which was that nobody had bothered fully parsing 37D so we didn't realize it was clueful for the murder.

Any Coat Will Do:

Spoilers for Any Coat Will Do I don't actually remember the state of the puzzle when we were stuck on it, but it does seem like we might have found the intermediate cluephrase at some point, failed to parse it, and discarded it because it does look like a phrase that's easy to fail to parse.

  1. I want to say the hobgoblin of little minds again but I titled a blog post with that too recently so have a bizarre novel string of words instead↩

  2. MIT slang for alumni↩

  3. pump (v.) to do a lot of damage quickly to an enemy in a video game (?); metaphorically by extension, to succeed at something, especially rapidly or over a short period of time↩