2019 MIT Mystery Hunt, Part 4: Puzzles

(This is a recap/review of the 2019 MIT Mystery Hunt, which happened in January. Puzzles and solutions can currently be found here. These posts will contain many spoilers.)

Happy Presidents’ Day! Hopefully you’re celebrating in the traditional way: invading Thanksgiving Town and Valentine’s Day Town. I will celebrate by talking about puzzles I was an author or co-author on in this year’s Hunt.

Puzzles I Wrote

I almost made that heading “Puzzles I Wrote By Myself” to differentiate from the next section, but honestly, nobody wrote any puzzles in this Hunt by themselves. We had a very hands-on editing staff that helped refine ideas, give advice during writing about both puzzle mechanics and content, and suggest revisions before and after testing. My puzzles were definitely improved by the editors who worked with me, and I hope I helped with the puzzles I edited (beyond the handful where I ended up butting in enough that I was promoted to co-author).

* I Knew Weird Al Yankovic, And You, Sir, Are No Weird Al Yankovic (Halloween)

In 2017, I wrote a puzzle called Before & After MASH that I was pretty happy with. It consisted of a bunch of alternate lyrics I sang to TV theme songs, about subjects that overlapped the titles. I heard that some teams played the songs for their entire HQ and solved it “shout out” style, which is the ultimate compliment. I was definitely hoping to accomplish that again, and at some point I randomly wondered what the lyrics to “Shingle Ladies” would be (and answered that question for myself in two minutes), so I decided that was the thing. Whenever possible, I tried to end the snippets right before the title would be sung, in the hopes that one of those listen-alongs would result in people singing the answer together.

I limited myself to single-word artists (no “The” allowed) and tried to use the most well-known artists I could stand. This forced me to use Vengaboys, because while I loved V.A.S.T. in college, they’re (he’s?) way too obscure. I expected to use Radiohead, one of my all-time favorite bands, but the fact that many of their songs don’t include the title in the lyrics was problematic; I was complaining about it to fellow Radiohead fan Tanis and she suggested Roxette, which was one of my favorite tracks to record, so props to her. The background music is all from karaoke tracks I pulled from YouTube, and I recorded vocals over them in my bathroom using Audacity. The worse I sound on a track, the more takes it probably took to sound as decent as it did… my singing voice is sadly not what it once was. At some point we decided to alphabetize by original song title, and I was horrified that this put Kansas and Styx first, which are definitely my worst performances.

When I submitted this, I also suggested an alternative puzzle called Your Own Medicine, in which I would write parodies of non-parody Weird Al Yankovic songs, like “Skipper Dan” and “Frank’s 2000 Inch TV.” This option did not get chosen, because (a) it’s infinitely less appealing than what I did write (except to a small niche group of superfans), and (b) a lot of Al’s originals are artist style parodies, and I’m not sure that a parody of them wouldn’t sound more like a parody of the homaged artist. (“Is that a new version of “C.N.R.” or “The Denial Twist”?)

* Spinning Tops (Halloween)

I think this was the first puzzle in the entire Hunt to graduate from testing. Most Hunt puzzles either start as an outside (or theme-related) idea that is in search of a best-fit answer, or they start by using an answer from the answer list as inspiration. This was the latter; I knew SBARRO was in an early round and thought I could whip up a pizza-themed puzzle pretty quickly, so I did.

The original slicker title to this was Topspin; unfortunately, we discovered in the first test that TOPSPIN is a transdeletion for TOPPINGS, and that sent the solver in the wrong direction. I’m still not in love with the revised title, which doesn’t have much of a ring to it.

* DK8: The Turducken Konundrum (Thanksgiving)

Coming up with the title was easy. The rest was hard.

As soon as I knew we were going to have a Thanksgiving round, I figured we should have a turducken konundrum. I wasn’t sure what that meant though. Initially I figured the solver action would involve stuffing objects into other objects as one does when making a turducken. But this was unsatisfying on many levels. DK6 was about cooking. DK5 involved a bunch of random objects, and lots of previous duck konundra involved moving stuff around grids or another playing surface. I wasn’t really sure what would stand out about this particular set of wacky instructions, which meant I wasn’t very motivated to write it.

Then at some point I had a revelation: a turducken konundrum isn’t a konundrum about things inside of things inside of things… It’s a konundrum inside a konundrum inside a konundrum! I posted this on Puzzletron on September 16 (the post began, “I HAVE HAD AN EPIPHANY”), and started brainstorming various elements, such as who would be solving the Hunt on each “level.” But it was still two months before I found time to sit down and actually start writing steps. Once a draft was written, in place of my traditional “test where I look over the solvers’ shoulders at every step,” I re-solved the whole thing myself in order to write the solution. Then it got two group tests in Seattle and New York, as well as Jen Berk solving it herself on a spreadsheet, which is apparently a thing she likes to do, and we decided it was ready to roll.

I’m pretty burnt out on writing duck konundra (and have been for years), but I’m very happy with how this one turned out. Given the meta theme, if this turns out to be the last “official” duck konundrum, I think it’s a good way to wrap things up.

* Crosscut (Arbor)

I don’t know where this idea came from, but I liked the idea of reuniting scrambled haiku, and if the solver was reordering sets of nine bits, I figured iterating that step on nine-letter movie titles would make sense. Some early testers were very grumpy about filling “gibberish” into the crisscross grid; taking something meaningful and intentionally making it scrambled felt wrong to them. My favorite tester quote: “The twist is clever. Is it ‘fun’? I don’t know. But I respect it.”

* Connect Four (Holi)

I test-solve for Grandmaster Puzzles, which is one of the best sources for abstract logic puzzles in the world. (The only reason I don’t simply say it’s THE best is that I don’t regularly solve Toketa?, which is also supposed to be of extremely high quality.) I asked Thomas Snyder shortly after we won Hunt if I could hijack a week of puzzles to use as Hunt puzzle fodder. Thomas gave me a lot of latitude, under the condition that the puzzles were independently solvable and met the site’s usual quality standard.

Several other GMPuzzles testers are also Mystery Hunters, so a tricky obstacle was not making it totally obvious in advance to them that this was part of Mystery Hunt. I attempted to hide this by using a pseudonym, though things got awkward when the testers rightfully questioned the randomness of the letter assignments and started trying to “improve” them. As it happens, one GMPuzzles tester told me after the Hunt that while he was sleeping, the team had found the connection and solved all the fillomino from scratch, not realizing that the tester had already solved them earlier (and had the official solution file).

* Gone Guys (Pi)

I still watch WWE wrestling regularly (I have Raw on now as I write this), but I’ve gotten really into New Japan Pro Wrestling in the last couple of years, thanks to their global streaming service, and I’m fascinated by how much work goes into the booking of the G1 climax tournament. NJPW is a company where wins and losses matter–when a champion gets beaten in a tournament match, the opponent is expected to get a significant title shot at some point–and all the point totals have to add up appropriately for the final day to matter, yet without earlier results making it obvious who will be in contention for the final. Compared to a WWE where currently the chairman rewrites the major shows hours before they air, it’s amazing how much planning (and math!) is put into this every year.

Anyway, I was curious how few results it would take to determine who was who in a tournament block, and that was easy to turn into a puzzle (written entirely by hand, so I knew there was a valid solving path that a human could work out). I heard from lots of non-wrestling-fans who got the aha, knew nothing about who the wrestlers were, but still really enjoyed the logic aspect. And I also heard from one wrestling fan who thanked me for writing a puzzle specifically for him.

* Ore Aft (Pi)

This was another “inspired by the answer” construction, as I asked Tanis for a handful of answers that hadn’t been claimed and one was notably part of a sequel subtitle. As it happens, I’d drafted a Fourth of July + Christmas meta that involved subtitles of fourth movies in series, and that hadn’t made it out of committee, but I still had a huge list of fun movie subtitles. I didn’t spend too much time on this puzzle or think of it as a highlight, but a lot of people cited as a favorite after the Hunt.

* Standardized Mess (Bloomsday)

This was kind of a bold construction, in that I was at first not sure how to clue the aha, and then I decided I would thematically clue it with every single question on the quiz. I was hoping for an effect where solvers would not see this for a while, and then all of a sudden they’d see everything at the same time (kind of like the 2001 Hunt endgame, where there was an entire deck of tarot cards hidden around the room with five in plain sight, and we spent 45 minutes staring at the five, before someone found one card, and then we found a bazillion cards all at once). Judging by Eric Berlin’s blog post, Palindrome had an experience along those lines, but the delay was a lot longer than I intended. Sorry, folks.

An awkward anecdote: In the first test, I made a construction error so that instead of flipping four of the grids, all the grids actually appeared in the orientation that would yield the answer, which they eventually found without doing anything with the quiz. Some members of that group tried to be diplomatic, but others were pretty clear (assuming the quiz was a giant red herring) that they thought the puzzle was dumb. And I agree, that version of the puzzle was extremely dumb!

Puzzles I Helped Write

* Haunted (Halloween)

When Alex Rosenthal found out he was going to be able to do a TED talk about Hunt, a bunch of editors got thrown together to quickly brainstorm the best way to embed information for a 2019 puzzle. The tricky thing was that we knew people would see the talk before Hunt and would look for puzzle information. I think I was the one who suggested a puzzle with a clear answer, so that those people would think they were “done,” but that could be solved differently with a clue phrase clarification to yield a different answer (and the answer AMBIGUOUS was perfect for this state of affairs). I came up with enough details for the mechanic we eventually used that Alex graciously invited me to be a co-author, though he did most of the heavy lifting on clue construction (and, of course, giving the talk.

I initially suggested the title Hunted, which we used in the first test, and the testers thought since it was in Halloween, Haunted would be more thematic. IN retrospect, maybe we should have used H(a)unted.

If you’re looking for additional pro wrestling content, yes, that is a Kevin Owens shirt I’m wearing in Alex’s TED talk!

* If At First You Don’t Succeed (Thanksgiving)

Setec held a weekend retreat in September to try to jumpstart testing and construction. As part of the retreat, Chris Morse and I led puzzle construction seminars, and my seminars had a format where Tanis gave me a handful of unclaimed answers and we brainstormed puzzle ideas for them. This was very productive, though I ended up as a co-constructor on most of the ideas, so it made my plate very full in a short period of time. (Fame is Fleeting and Compromised also arised from these sessions; I edited both but didn’t co-construct.)

This puzzle was basically triggered by the observation that the two words of the answer started with the same trigram, and that the remaining letters could be clued independently. We considered giving this without the accompanying trigrams, but given that it was in an early round, we wanted to err on the easier side. If it was in a late round, we probably would have done the opposite.

* Twelve! Eleven! (New Year’s Day)

I like puzzles with lots of mini-ahas even once you know how the puzzle works, and this was one of those. I recruited Jackie to help brainstorm list ideas, and to help me narrow down a larger list of lists to the best set of options.

* Middle School of Mines (Arbor)

This was mostly Philip Loh’s baby (a sequel to his 2017 puzzle School of Mines), but I gave the suggestion of creating a final logically solvable “ubergrid” where all the absent instances of one number would be used as an extraction. Then I wrote that final puzzle, which was enough to get me an author credit.

I now find the fact that he called this “Middle” School of Mines troubling. Philip, we’re not winning again. Stop that.

* Rules of Order (Holi)

Of the three puzzles I co-wrote with Jackie, this was the one that was primarily engineered by her, which makes sense because arithmetic dynamics is what she does for a living. She had the idea of getting solvers to identify Julia sets, and I suggested also identifying Julia Roberts movies at the same time, and we weaved those ideas together into the final product.

The original draft gave the movies in extraction order, but shortly before testing, I had the unnerving realization that if you knew nothing about Julia sets but identified all the movie character names, you could do a computer search for ten-letter strings that pulled a letter from each name in order, you could narrow things down to a couple of options, one of which was clearly thematic and likely correct. To combat this method of short-circuiting, we came up with the “order by magnitude” mechanic, which ensured you’d need to know which letters you were taking *before* knowing what order they should be placed in. Hopefully this stopped people from bypassing the math.

* Protection Plan (Pi)

This was another retreat brainstorm based on the answer ZOEY. Somebody remembered that Zoey Bartlet had a fun Secret Service codename, and in doing some research we discovered that Secret Service codenames of fictional characters is a fun and extensive data set. Someone suggested connecting the data set to the game Codenames, and that’s when it started to feel like a full idea to me. There had recently been both a Secret Service codenames puzzles (with real codenames) and a Codenames puzzle (emphasizing the red and blue elements) in Hunt, but we decided the mechanics in this puzzle were different enough from both of those to justify its existence.

* Letters From the Battlefield (Patriots’)

Mark Leach, a fellow Midnight Rider, beat us to putting an MLS puzzle in Hunt, but once I knew Patriots’ Day was a round, Jackie and I were definitely writing a New England Revolution puzzle. Spelling an answer in chronologically ordered away game locations in consecutive years was very constrained, and once we had the game set, Jackie and I both generated lists of notable game elements, but then I wrote the majority of the old-timey game descriptions. This was tedious. I hope this was more fun to solve than it was to write.

Kelyn Rowe shows up in a lot of these descriptions, partially because he’s my favorite soccer player, and partially because he just happened to do a lot of notable things in these games. He was traded this off-season. 😦

* Broken Concentration (Bloomsday)

This was another retreat brainstorm, with Matto Mildenberger leading the charge in suggesting that this answer should be a “concentration” game where the undersides of the cards change every time. (A good way to brainstorm Hunt puzzles: What variation on Puzzle Type X will cause the solver to swear at the constructor thirty seconds after opening the puzzle?) Most of the mechanism got hashed out in that room, and then we just needed to implement it. Sincere apologies to anybody who revealed all the pairs in the hope of getting extra information at the end.

* Getting Digits (Bloomsday)

Chris Morse wrote this puzzle as a 2017 backup puzzle that didn’t get used, although in that form it was just a list of clues for each segment. At the beginning of the year, I wrote a manifesto about how to avoid both Identify-Sort-Index-Solve and “list of clues” puzzles, so if we used this, I really wanted it to look like its own thing, so as an editor I proposed the “scroll readout” presentation we ended up using (and the START clues that would add an alignment step), and Matt Gruskin made it look awesome, as he did with many puzzles (Broken Concentration, for example).

Chris was indisposed during much of the period where we were writing the START clues, so I ended up writing those and got promoted to co-author.

* Keeping Tabs (Bloomsday)

Yet another retreat brainstorm, and yet another Matto inspiration, who was the one to note that UKULELE has four unique letters and four strings. He wondered how many other instruments have this property… enough, as it turns out. We ran into a bit of a block trying to figure out how to clue the bank words in a way that felt like it belonged in the puzzle. Someone suggested a series of couplets being “sung” by the ukulele player, and I predicted that was too constrained to be well-written. Justin Werfel quickly proved me wrong with a great set of rhyming clues.