On May 10, the first HMMM (Harvey Mudd Mystery Marathon) was held. This document serves as a recap from the perspective of the event organizers (Celena Chen, Jon Hayase, Cole Kurashige, Ricky Shapley, and Brandon Wada) and covers both puzzle creation and the event itself. If you’d like more context about HMMM or puzzle hunts, read this post first.
A good amount of this recap was written by Celena; I added some content and ported it from Google Docs to a more appropriate medium.
HMMM by the numbers
Number of puzzles: 13, counting interludes and the meta
Duration: a little over 3 hours
Number of teams: 11
First guess: Mr. E and the Bozos, with the guess “Shakespeare” for the puzzle A Rose by Any Other Name
First interlude submitted: Theoishisname
Total guesses: 392
Total solves: 82
Hints requested: 50
Average number of solves: 7.5
Most guesses: Team #ReliveTheDream with 93 guesses (which completely eclipsed the other teams: second place was team 322C with 51 guesses)
First place team: donut play on the tracks (11 puzzles solved)
Puzzle with the most solves: Clearly Confused (solved by all teams)
Puzzle with the most guesses: A Strange Path
Puzzle with the fewest solves: Registration (the meta puzzle, with 0 solves)
Puzzle with the fewest guesses: Registration (also 0)
Our favorite answers
Not everyone gets the puzzle right the first time. Here are our favorites.
For context, our guesslog has columns [team name] [puzzle short name] [guess] [puzzle solved?] [guess correct?].
Interlude I asked teams to send us a poem written without any alphabetic letters.
(note from hmmm resident linguist & writing system enthusiast [Celena]: had this team actually written a poem in korean (a hint of which is in the bottom right), this would not have received credit, as hangul is an alphabet! chinese is logographic, which is distinctly nonalphabetic, so that’s fine. the devanagari script (which i think is what the bottom thing is) is an abugida and not technically a full/true alphabet, which i guess you could argue with me on.)
Asking for creative submissions always opens the door to answers that are borderline. One such answer was so obscene that we are not including it in this document. It did not receive credit. This answer cost its team third place (we don’t know why they did not submit another answer after we told them to).
The website
This puzzle hunt featured a website (which may still be accessible here) created from scratch by the great Ricky Shapley one late night. There are a lot of jank parts, but we love it, jank and all.
One particular feature Ricky is proud of is the way that we keep track of what puzzles a team has solved. The natural way to do this would be to store a column in our database which tracks whether a team has solved a particular puzzle. But this requires keeping tabs on each puzzle for each team. Ricky circumvented this by adding what he called a “policy number” to each team. A team starts out with a policy number of 1. Each puzzle is assigned a distinct prime. When a team solves a puzzle, its policy number is multiplied by that prime. Want to know whether a team solved a certain puzzle? See if its prime divides that team’s policy number.
The observant reader might notice that we are just using the default Heroku URL for our website. This was not meant to be. We bought the website hmmm.page specifically for this hunt, but had firewall issues with our domain. We gave up on using the custom domain right before the hunt started.
The puzzles
Each puzzle has its own story, so we’ve documented them in their own posts. I’m including the names of all puzzles below, but only those with hyperlinks have recaps. We left some puzzles out because we plan on using them in future puzzle hunts.
Round 1 was available at the start of the hunt, and Round 2 was made available about an hour and fifteen minutes into the hunt. Both interludes were unlocked after solving all of the puzzles in the round and the meta was unlocked after solving both interludes.
The parenthesis represent (numbers of solves, number of guesses).
Round 1
Affixation — (8, 71)
A Rose By Any Other Name — (9, 79)
A Strange Path— (10, 101)
Clearly Confused — (11, 25)
In Creasing — (10, 53)
Interlude I* — (6, 7)
* Interlude I wasn’t really a puzzle; instead, we asked teams to send us a poem written without any alphabetic characters.
Round 2
An Eightfold Path — (2, 10)
Harvey Mudd School of Counting — (3, 11)
In Sum — (10, 19)
Sudowoodo — (9, 11)
Werewolves — (3, 18)
Interlude II (Domain Knowledge) — (1, 1)
Meta
Registration — (0, 0)
The hunt
HMMM turned out much more chaotic than the first puzzle hunt we hosted.
The first twenty or so minutes after we got teams settled and the hunt underway were relatively tame. We answered some hint requests and worked on integrating a last-minute puzzle change. Then we were notified of an error in one of the puzzles. One errata later, things started to pick up the pace. Over the course of the hunt, we had three errata total, the most humorous one being changing 9,000,000 to 73.
We observed after about an hour that at the current rate, teams were unlikely to finish the hunt even in our “overestimate” of 3 hours. Our hint system allocated one additional hint request to teams every thirty minutes — in an effort to speed up the hunt, we doubled the rate so that teams gained a request every fifteen minutes. This triggered an influx of hints that consumed us three event organizers. Did I mention that only some of us knew the solutions to some puzzles? Or that we ended up adding the new puzzle (Interlude II) about 1 hour and 15 minutes into the hunt?
Needless to say, we were kept busy.
Overall, the puzzle hunt was a bit too difficult for how long it lasted. Unfortunately, no team got to the meta, which was the puzzle we had spent the most time planning. We had a few puzzles in particular that were a little bit too confusing or otherwise vague for teams. One puzzle, which we left out of the recap since we’d like to reuse it, had two valid answers. We were informed of this after a team sent in a hint request, confused as to what to do next after getting an answer that we rejected.
We learned a lot from hosting HMMM #1 and we’re looking forward to the next puzzle hunt we make.