2018 MIT Mystery Hunt, Part 2: My Meta Experience

(This is a recap/review of the 2018 MIT Mystery Hunt, which happened this month. Puzzles and solutions can currently be found here. This recap will contain spoilers on many metapuzzles.)

I mentioned in my previous post that to me the most exciting thing about Mystery Hunt is finding the most efficient way to navigate the structure. This leads to my spending just as much time, if not more, staring at a blackboard full of answers as opposed to solving individual puzzles. I think of this as “strategic”… A passing observer might characterize it as “lazy,” which is probably occasionally accurate in moments that I’m not fully awake. (Though I worked on a fair share of individual puzzles, which I plan to talk about in the next post.)

As we solved emotion puzzles from the intro round, various members of the team made various bits of progress on the five metapuzzles. I focused on Fear, as I’d guessed from “medical emergency” that the words and phrases appeared in the Health & Safety guide (as had other team members) and I thought it was pretty significant that the small-word answers we had like “except” only appeared once. I correctly looked at the words after the answer words, and unfortunately, our first four gave the words “locations,” “of,” and “lost,” which sounded like parts of a useful phrase. We tried putting these words in the guide order, but they didn’t make a meaningful phrase, so I kept trying to rearrange them… meanwhile another group of team members thought of looking at the acrostic and got the right answer. We actually solved our emotion metas in a pretty narrow period of time: according to our Activity Log, we called in Sadness at 4:16, Joy at 4:26 (after incorrectly trying BUCK UP), Fear at 4:28, and both Anger and Disgust at 4:40 literally five seconds apart (we didn’t get a Disgust interaction, as Anger brought us our item from Disgust).

We were surprised that solving all of the emotion metas didn’t immediately open an island; we had to go back and solve a bunch of additional puzzles to pump up our brainpower, though we did get our first island (Hacking) open at 4:52. I did some backsolving cleanup and was proud of myself for getting ASCENSION based on some letter information and the book requirement for Joy; I did, however, convince myself that Beast Workshop’s answer would surely have GORILLA in the answer, as a homophone for GUERRILLA, which, like KNIGHTLY, is eight letters and has an L as its 7th letter. Well, it does if you spell it wrong. Let’s move on, nothing to see here.

Presumably due to a backup on interactions (planning to send five visits to every team during the first round is ambitious!), we didn’t actually unlock the Brainstorm until 6:28. Like many teams, judging by what I see elsewhere, we brought back the five words and thought we needed to solve them; eventually someone found the instruction that said to just enter them as a string (although there was some text elsewhere that seemed to contradict that), and thankfully we did that. It was a little underwhelming, especially because this would have been a nice opportunity to “solve” the logo (we tried taking shared letters at the points where colored regions met).

We had been eating dinner when the Brainstorm happened, and afterward I went back to trying to fill in letters into the chutes and ladders grid for the Scout meta. Before dinner we only had two answers that overlapped, and they fit into two places. I had noted (to Josh Oratz, I think) that in one place they could make ?XO???????? if you read upward, and I brainstormed out loud “EXOSKELETON?” After dinner we had a second overlap that only fit in one place and blocked one of the options for the first pair. Josh was thankfully still around and said, “Look, it is EXOSKELETON!” which I had not noticed. After calling this in, I fit in the rest of our answers and determined that the other two should be of the form L????A?? and PE??????L. The latter seemed like it could only be PERSONNEL, PERENNIAL, or PERPETUAL. The folks working on Murder at the Asylum pointed out that the answer was likely to be nine letters long, and PERSONNEL seemed thematic, so we guessed that at 10:47pm. It was incorrect. We’ll come back to that.

I hadn’t left HQ in a while, so I decided to participate in the Scouting Challenge with some other alums. We definitely missed some pictures early on, because when we finished we had the string TTNOONE. We were going to retrace our steps when I suggested the answer might be TRUST NO ONE, and we called that in and confirmed it, saving us some time. When we returned to HQ, James Douberley was coming by to visit… I whispered “Trust no one!” to him, and he laughed politely with a look that suggested he didn’t know what the hell I was talking about.

We had a second island open by now, and I wasn’t really following the Pokemon structure other than hearing that there were puzzles and harder evolved puzzles. I participated in the solving of You Know What’s Missing (without urinating) and Mass Aid (uggggh) and helped wheel-of-fortune out the answer to the Advertiser meta (from 2/5 answers, I think) after someone else worked out the extraction. Around 11:30pm I decided it was a good time to go back to the hotel and sleep.

I probably got about 5.5 hours of sleep in and returned to HQ around 6:30am. I was shocked that we still didn’t have a third island open, particularly since the organizers had bumped up the requirement for said island due to points flowing too quickly. They had actually called us, since we were (in their words) “very close” to opening a third, to make sure we wouldn’t think that was unfair. I didn’t realize that was going to result in six more hours with only two islands! We were close now, so after being sad about missing the MLS puzzle overnight and circling the word CONIFER in CONFINED AQUIFER, I noticed everyone had given up on Murder at the Asylum. I talked through some of it with the half dozen people in the room but eventually decided I needed to go out in the seating area in the hall to try to concentrate on solving it frontward. It’s a great puzzle but it took a loooong time; from the Log, it looks like it was hours, though I didn’t think it was *that* long. I never fully worked out the murderers/accomplices, but I got enough information to squeeze out the answer ALTERNATE. That’s weird, that’s not PE??????L or L????A??. The printout I made of the chutes & ladders board had been thrown away, so I filled in a new one and discovered I’d left out a letter. With ALTERNATE confirmed, I realized that the PE??????L options were now valid for 10000-Puzzle Tesseract, which lots of people had put time into without finishing. We tried all three, and the last one PERSONNEL was correct at 9:37am… almost eleven hours after we’d tried it on the other unsolved puzzle. Major. Backsolve. Fail. On the bright side, this was just enough brainpower to open up Games Island.

This is getting quite long, so I’ll wrap up this post and continue in Part 3.


2018 MIT Mystery Hunt, Part 1: General Thoughts

(This is a recap/review of the 2018 MIT Mystery Hunt, which happened this month. Puzzles and solutions can currently be found here. This recap may contain spoilers, but I’ll try to avoid them where possible, especially for puzzles I recommend.)

Going into this weekend as a member of Setec Astronomy, I did not expect my team to win the Mystery Hunt. Some of my teammates were taking a more relaxed approach, although I pretty much only have one intensity setting at Hunt; for me, the most fun thing about a puzzlehunt of the size and complexity of Mystery Hunt is figuring out the most efficient way to get from Point A to Point C (for coin) through strategic puzzle-solving, meta-cracking, backsolving, and time management. Navigating the Hunt itself is the most interesting puzzle of all.

On Sunday, after grappling with the final Sci-Fi Island metapuzzle for about five or six hours (for those on the construction team who asked, we were NOT stalling; we were missing a key insight for more than half that time, and then once we implemented the right thing, it gave us garbage output due to a technical error we didn’t find for quite a while), it became apparent we were about to win the Hunt, and we had a conversation about that. My philosophy, which I acknowledged is not every participant’s philosophy, is that if you choose to be on a team strong enough to win the Hunt, you owe it to the community to write the Hunt if you’re the fastest team to reach a point where you can finish. Members of our team had lots of different perspectives and conclusions, but the vast majority agreed, for one reason or another, that we should finish what we started, and we surprisingly found ourselves in charge of next year’s Hunt. Again.

We learned a lot writing in 2017, and we have yet another data point in seeing this year’s Hunt and how it worked from our perspective (and I hope to hear how it went for other teams). There are some things we’ll try to do again, there are some things we’ll try to do differently, and there are some things that will be much easier having done them recently (I know TK is super-psyched that most of the Hunt website code already exists in a form he understands). We’ll probably lose some members, we might gain some members, and our Hunt may be very different than the one many of you just solved. But like every construction team, our goal will be to produce an event that everyone enjoys, and we’ll do our best to achieve that goal.

But enough about 2019, what about 2018? My favorite Hunt of all time, the 2011 Hunt (Video Games) felt in many ways like an improved form of the 2009 Hunt (Zyzzlvaria) taking a lot of the bolder innovations of that Hunt and presenting them in a more effective way. Similarly, I thought that this year’s Hunt felt very much like an improved version of the 2013 Hunt (coin heist); the rounds had complex structures, often with multiple tiers and/or multiple metapuzzles, and the metapuzzles themselves often involved elaborate procedures as opposed to a simple aha about the feeder answers. I was thinking this long before we opened the Sci-Fi Island (which was our fourth), so I already made a connection to the 2013 Hunt before another six-sided-cube-with-overlapping-metapuzzles appeared!

There were at least three big improvements over 2013, however: (1) Puzzle quality across the board was a lot higher. Puzzle concepts were bold and creative, and execution seemed clean. 2013’s Hunt only ended on Monday because the organizers handed out puzzle answers like candy; this year the first completion happened later than Death & Mayhem (Life & Order) expected, but it happened organically (by the end we could afford to “buy” two free answers, and we just bought one to make our last meta solve easier). (2) While the metapuzzles were often very difficult, they were reasonable. All four of the top-level metas were complex but none of them took hours of processing; if we had all of the answers and knew what to do, they would not be a slog. (3) I found the theme (at least the main theme) very engrossing; Inside Out is fantastic source material, and the production values in the kickoff and final runaround were ridiculously high.

If you can’t tell, I liked this Hunt a lot… there were a couple of two-hour periods where I felt useless and a little frustrated, but the Hunt never felt unfair, and most of the time it was a lot of fun. My biggest complaint was that the island themes felt pretty random and a bit disappointing. The Hacking round, which we opened first, seemed like a labor of love clearly designed by people familiar with MIT hacking culture, and the tier filled with physical puzzles was very cool. We opened Pokemon next, and the structure of that round was neat, but “Pokemon” and “hacking” don’t really seem like items in the same category. We could tell from the “preview text” that our third island was going to be Catan-themed, and ultimately I didn’t really buy it as “Games Island,” since it wasn’t really about general games. And as for the Sci-Fi Island, the cube structure’s been done, and a space fiction theme that unites lots of different existing properties is a great Hunt theme… which is why we used it in 2009. The point made at wrap-up that these were different chronological stages of Terry’s life made them feel more related, but I didn’t grasp that at all during the Hunt when it matters most. I was definitely expecting the islands to be either closer to the Inside Out plot or themed around other Pixar films, and I think either of these would have resulted in a more cohesive Hunt. This was a great Hunt, but the stitches in the overall structure were showing.

One intriguing decision by the organizers was to allow solving teams to choose the order of the round unlocks, which hasn’t been done since 2004. (For anybody who heard James mention the “Vatican Effect” at the wrapup and didn’t know what that means, in 2004 you could choose which round you’d unlock next, and the organizers thought solvers would likely open rounds in numerical order… but the highest-numbered and hardest round was visually closest to the opening round on the visual map, and so many teams chose that thinking they were supposed to.) This “choose-your-own-adventure” approach has been brought up in multiple years when I wrote, and I’ve always opposed it out of fear that solvers who choose the “wrong” order might get bottlenecked and get screwed out of the opportunity to win by random bad luck. It also gives the organizers less control over when solvers encounter key concepts… When we were dealing with the scavenger hunt, we wondered how frustrating it would be to open said scavenger hunt in one of the last unlocked rounds, since we had to send team members home to gather stuff. From talking to other teams at wrap-up, I can confirm it would be very frustrating.

This got discussed a little bit at wrap-up, but I was hoping for a lot more information. What order did you open the rounds and why? And do you think it helped or hindered your progress? I’m glad someone had the guts to try this in a year where I wouldn’t be responsible for the result, and I’d love to hear how it affected your experiences.

Coming soon: More posts on my personal Hunt experience with the metas and sleep (or lack thereof) and some notes about puzzles I particularly liked or wanted to throttle for one reason or another. If there are things you particularly liked or disliked this year that you hope next year’s team will preserve or change, chime in via the comments; I can’t guarantee any/all requests will be honored, but I’m pretty confident that next year’s constructors will at least hear about your feedback.

Recap: P&A 70 (aka Lion, Schmion, Backsolved)

(P&A 71 was released today, which means the answers are now online for P&A 70; so here’s my spoiler-laden recap of our experience solving the last issue, themed around Madeleine L’Engle.)

For a change, we weren’t racing to be done for soccer reasons; the Midnight Riders actually had a pub trivia night (well, afternoon) at 2, but we knew we wouldn’t make it to Boston by 2 and it had already been a busy holiday weekend, so we decided to punt trivia and focus on puzzles. As it turns out, we finished just after 2 (starting at noon Eastern), so this was a very quick issue for us.

My printer prints pages in reverse order, and so New Year’s Eve caught my eye right away. I got the aha immediately (from the flavortext and spotting the anagrammed “INFINITY”). Jackie had picked up Earth Day, but I asked her to help spot a few of the remaining math words, and after switching from the incorrect ordering (vertical) to the correct one (horizontal), we submitted the answer from six out of nine letters at 12:06. As usual, I went to the Stats page to try to identify the low-hanging fruit, but at that point our solve was the only solve! So I flipped through the stack and New Year’s Day seemed approachable; that was solvable with only about 2/3 of the grid filled, submitted at 12:14 (after incorrectly submitting the plural form). As it turns out, the first two answers we got were probably the two of the most helpful in breaking into the meta, but we didn’t know yet that they were paired!

Now there were a few other solves on the Stats page, which led me to Mother’s Day. The third clue on that one was the break-in, and I submitted the answer at 12:19. At this point, Jackie had the gimmick on Earth Day and most of the grid filled, and we worked together to figure out the answer at 12:25. We then teamed up on Flag Day, which wasn’t too bad (12:29). Now there were six puzzles solved on the stats page, and the only remaining one was Back to School, a word search. Faced with the choice between a word search and an unsolved puzzle, Jackie wisely ran toward the unsolved, grabbing the logic puzzle Independence Day while I dealt with Back to School. It was tedious but not too difficult, and I finished it at 12:42. We now had six puzzles solved, and according to the Stats page, no other puzzles had been solved by anyone else.

Jackie kept working on the logic puzzle, while I focused on Christmas Eve. I ID’d a lot of pictures and didn’t see a common thread, so I employed a standard strategy for this sort of puzzle… Google various combinations of likely interpretations and hope the results yield, say, the lyrics to You’re A Mean One, Mr. Grinch. Once that page came up, I was able to ID all of the pictures except the Eel and the termites. (Fun fact: Early in our relationship, Jackie wrote a short Christmas puzzlehunt for me that featured a Grinch song parody; the lyrics were about me, and the extraction involved looking up the long bits at the end of the verses. I bring this up in case anyone doubts that I married the perfect woman.)

I was stuck on extraction and so was Jackie (having finished the logic portion of her puzzle) so we worked together to finish the extractions for Independence Day (1:07) and then Christmas Eve (1:09). We then worked together on President’s Day, and after filing the grid with what seemed like garbage letters, I wondered aloud, “What does this have to do with presidents?” and she had the aha moment that the letters (mostly) corresponded to last names of presidents. Once again we tried reading vertically before we read horizontally, but then we had an answer at 1:25.

Nine answers and no ideas on the meta… I had noticed MON in LEMOND and thought from the date pages that we might be using days of the week, but I didn’t know how that would work with the clocks, and there weren’t day abbreviations jumping out of the other answers. Jackie got to work on the anaquote lyrics while I started chipping away at the cryptic clues in Valentine’s Day. I was noticing some overlaps between clue answers (PANACHE/PANACEA, for example), and when Jackie had about half of her lyrics identified and was trying to do the same creative Googling we’d been successful with before, I thought about the title and noticed the Days of Christmas gifts in the artists. Initially we counted wrong when indexing and got a key letter wrong, but after correcting the L, we got from ?EC?LA?MUSI? to an answer at 1:55.

Jackie went to move her laundry, and I still had cryptic clues to solve, but I figured that with ten out of twelve answers, we should really be able to break into the meta. After some staring, I found MIN/UTE and SEC/OND (now having SECULAR MUSIC was very helpful) and figured that SECOND/MINUTE/HOUR/DAY/WEEK/YEAR would fit the clocks and calendars. I filled in the ten answers we had and wanted the letters to somehow fit the sequences of wrinkles… they did not. But the other ends of the answers did seem to spell something, and with ten out of twelve answers, the meta answer fell at 2:04.

Based on the meta mechanism, the remaining answers needed to be of the form R*AR and T*Y. Since one of the remaining puzzles (First Day of Spring) prominently featured lions, I tried ROAR as a wild stab… Success! Jackie returned and we teamed up to finish Valentine’s Day; we knew the first and last letters from backsolving and worked out the second and fourth (and first) letters as intended, which was enough to figure out the answer and secure a complete.

January Update: Two months later, we never figured out how the lion/lamb puzzle worked. I’ve looked at the solution now, and it appears the lions and lambs are just showing you which letters the hidden Boggle words start and end with; if that’s the case, I cry foul at the fact that some of the corner squares are indicated by icons orthogonally adjacent to the letter, and some diagonally adjacent (see, for example, the two lambs in the bottom center grid). I don’t see any reason to place them inconsistently unless they’re trying to be something more involved than “next to a letter”).

Miscellaneous Updates (or Keeping Up a Blog is Hard)

Sorry again for the long delay between posts; the end of the semester can be busy for academics like me, and it only gets busier when (a) you have to report a whole bunch of students for exam cheating, and (b) you’re hosting Christmas for the first time (we made a prime rib roast and didn’t poison anyone!). Most of my own puzzle solving has consisted of logic puzzles from past WPC/WSCs, some online programming challenges,  and test-solving some top secret Microsoft Puzzle Hunt puzzles for a friend (which were very good!).

The MIT Mystery Hunt, which is for me the highlight of the puzzling year, is in just a week, and I’m sure I’ll have lots to say about it afterward. In the meantime, let me take care of a few loose ends and bits of news:

  • There’s a new issue of P&A out tomorrow, as Foggy tends to schedule around the Mystery Hunt in January. This means that I’ll be posting a solving report on the last issue soon, and it also means you should purchase and solve tomorrow’s issue, as it’ll make for a nice Mystery Hunt appetizer. For the first year in a while, I won’t be solving it at a math conference, so I look forward to getting to use my own printer instead of staking out a FedEx Office store.
  • Another pre-Mystery-Hunt tradition is Kevin Wald’s excellent pre-Hunt cryptic crossword, which takes its theme from the previous year’s Hunt. Kevin’s cryptics are often very densely structured in general, and I’m always impressed at the connections he finds within the pre-existing Hunt theme/structure (especially when it’s one I helped write, and thus one I’ve spent a lot of time thinking about!). You can find this year’s puzzle and previous years’ here.
  • One of our Hunt team members, Tom Buehler, did a ton of filming during the 2017 Hunt (which we wrote), and he’s just released the final cut of his 52-minute documentary. I think it’s really good, and it’s worth watching if you’re getting psyched for this year’s Hunt and want to walk down memory lane (and get some insight into how we handled the re-staged events for the teams that outran our intended schedule), or if you’ve never participated in an on-location puzzlehunt and want to know what they’re like (there are some really nice visual segments showing how some of the puzzles are solved, so it’s a nice watch for newbies).
  • There’s a ClueKeeper hunt happening now in conjunction with an upcoming film called Solver. The main page for the hunt is here, and the puzzles are appearing on Instagram. I found the first six puzzles pretty uninteresting (though to be fair, they’re intended for a more general audience). The seventh, by escape room podcaster Errol Elumir, is more intriguing, and I think I know how it works, but I haven’t been able to crack it yet.
  • Embarrassingly, I still have yet to post about the puzzles from our July NPL Con extravaganza, despite writing three novelettes about the logistics of putting it together. I really will get to it at some point, hopefully before next year’s extravaganza.

I intend to be at the opening ceremony for this year’s Mystery Hunt, assuming I’m not totally foiled by MIT’s more restrictive parking policies (as of June, it’s no longer a free-for-all on the weekends, so beware). If your a reader who knows what I look like and we’ve never met, feel free to say hello!

Recap: P&A 69 (aka Damn You, Bender)

(P&A 70 was released today, which means the answers are now online for P&A 69; so here’s my spoiler-laden recap of our experience solving the last issue, themed around Isaac Asimov.)

I’m writing this report a week after the fact, but I jotted down solve times, which should help jog my memory. But before I recap our solve path, let’s have a little chat about QR codes.

QR codes seem to show up more and more frequently in puzzlehunt puzzles, and I haven’t heard anyone else describing them as the bane of their existence. I’ve probably run into half a dozen puzzles where I figured out how the puzzle was supposed to be interpreted as a QR code, but the actual interpretation seemed impossible. The things that drive me crazy are:

1) Unless the black squares are perfectly black and the white squares perfectly white, scanners don’t seem to work.
2) All of the scanning apps I’ve tried don’t indicate any difference between “I don’t see a QR code in front of me” and “This QR code is invalid.”
3) Despite the fact that there are a million apps that scan codes and give output, there seems to be no place online to manually enter a QR code grid and convert it (or explain why it doesn’t convert).

To skip to the end of the report, I figured out what to do with Bender the day before I actually solved it, because I could not get the %$#% QR code to translate. When you know exactly how a puzzle’s supposed to be solved, but you can’t solve it for technical/transcription reasons (and you’re not even sure which)… that’s not much fun. Fortunately the rest of this issue was more enjoyable, we started when the issue was released, at 1pm ET. Jackie solved with me, but joined about 15 to 20 minutes late.

My printer prints documents in reverse order by default (which is sometimes annoying for logic puzzle competitions, in which the light warmup puzzles tend to appear first), so Watson was the first puzzle off the presses. It was a fairly straightforward break-in, and I solved it solo and submitted the answer at 1:08. Checking the stats, I saw that solvers were getting AquaCom quickly, so I broke into that and solved it at 1:25. I translated some of the Call numbers from phonespell without knowing what to do with it, and then worked on Rosie. Once I had the gimmick for that, I was able to extract an answer at 1:41 with only about half the grid filled.

Meanwhile, Jackie had joined and had been working on Th Stpfrd Wvs, occasionally calling out a crossword clue she needed verified. (I’m old enough to have actually read the “Cheers and ____” column in TV Guide.) I came over as she was extracting letters, and was able to wheel-of-fortune out the answer at 1:44. I think at this point we worked on Bubo together for a while and got the clue phrase KINGLY FOURTH; we saw that several of the winged creatures/items had “king” varieties and assumed we had to take fourth letters of these. This didn’t lead anywhere, because it was very very wrong. (We didn’t correct that thinking for a long time.)

After some frustration, we switched to Astro Boy. Jackie had the insight that the names could be spelled by tracing the graph edges, and I suggested copying the same letters to the right side, since the circles were in the same place. Together, we got to the answer at 2:16. Actually, at some point during that time, I think I split off and looked at Tik Tok while Jackie was getting the Astro Boy aha, because I finished Tik Tok four minutes later (at 2:20), and I know the deciphering for that took more than four minutes.

Once I looked at Optimus Prime, it was pretty straightforward and we finished it at 2:31. Meanwhile, Jackie was solving more phone numbers in Call and had the insight that they matched in pairs. I joined in and figured out what to do with the area codes, and we got the answer at 2:39. Then I looked at R2D2 for the first time, which fell by 2:53. Jackie bit the bullet and started chipping away at the cryptograms in Vision. She read me some of the clues, and I not only recognized who they were but was pretty sure they all had eyepatches; since Jackie had actually read the flavortext, she figured out that left vs. right would be important, and we submitted the answer at 3:32.

From roughly 2pm on, we’d also been looking periodically at the meta with no luck, and we weren’t yet sure whether we should be filling things into white boxes or gray boxes or both. Even with ten answers out of twelve we weren’t sure what to do. I figured out that the letters in Bender translated to either BLACK or WHITE, but then struggled with making the resulting QR code scannable. (I colored in the BLACK squares with pencil… not good enough. Colored over it in sharpie… not good enough. Then I manually filled in a grid in Excel… still no, probably because I had a couple of squares wrong.) Meanwhile Jackie had what seemed sure to be a useful insight, as it was possible to fill an answer in the gray boxes reading across one row and an Asimov title (I forget which one) across the corresponding white row so that the letters matched. But after pursuing this further, we couldn’t find any more instances that worked. I went back to QR codes, and Jackie figured out that the first words of our answers (with two missing, of course) seemed to fit into the top boxes, so we should be generating something else for the bottom boxes.

We were already running late for yet another Revolution home game, so Jackie was packing up the car when I yelled (not literally) “Eureka” from downstairs. I had already pictured Silver Surfer in my head when we got WEB SURFER as an answer in a robot-themed issue (I know he’s not a robot, but he looks like one) and I got lucky enough to get the phrases “Silver Surfer” and “platinum blonde” in my head at the same time, allowing us to submit the meta answer at 5:00, make the top ten, and have me be much less cranky at the Revs game than I would have if I were still thinking about the meta. In the car on the way to the game, I was attempting to backsolve the remaining answers, and I realized what the Bubo answer was and why without the puzzle in front of me. But after numerous attempts to force the Bender answer (my most promising option was POTATO BUG), after the game I resigned myself to translating the entire grid into letters and automatically translating it rather than coloring by hand. The resulting grid was only a square or two off from my previous Excel file, but it scanned instantly, and the resulting challenge (resulting in a complete for the issue) was trivial compared to the QR code-grappling that preceded it.

I’m curious to know if I’m the only person crazy enough to be trying to work out the Six Authors meta-answer with five out of six issue answers. I have some incredibly vague theories, and while I don’t think cracking it would allow me to guess the sixth answer in advance, I think it might give me a bit of an advantage on the November issue… [NOTE: I wrote this paragraph in September, and in this month’s issue, Foggy revealed that there is no pre-written meta for the six answers, but he’s inviting readers to create a meta they feed into, Spaghetti-style.]

Watercooler: Endgames

Well, so much for posting a watercooler every Thursday. It’s been an insanely busy few weeks; on the puzzle front, BAPHL 16 was quite fun, Puzzle Boat 4 was fantastic (track down a team if you haven’t, you can still register and solve), and I’d like to congratulate the United States team for earning silver at the World Puzzle Championships. I qualified for the World Sudoku Championship for the first time this year (held right beforehand in the same location), but unfortunately a week in India in the middle of the semester wasn’t feasible, and the chaos of the last few weeks convinces me this was the right decision, if a sad one.

Thanks for those who shared ideas/experience about puzzle testing on my last discussion post; quantity wasn’t what I hoped for, but quality was top notch. It’s definitely interesting for me to hear about practices for events I haven’t written for, like MSPH or Puzzled Pint.

Today I’d like to talk (and encourage you to talk) about endgames; by this I mean the last portion of a puzzlehunt that wraps up the story and usually has some amount of puzzle content. By nature these tend to occur in large-scale hunts– I wouldn’t call the final metapuzzle of a P&A issue an endgame, per se– though I’ve definitely done some BAPHLs where the last meta or metas felt like a separate phase of the event, especially if that phase occurs at a new location.

Naturally, since I have the most history with Mystery Hunt, those are the endgames I have the most opinions about, and there have been some controversial Hunt endgames. In 2015 (20000 Puzzles) after a very slick, clean Hunt, the final endgame was a sequence of time-consuming activities, including a Family Feud game and a task to take something like a hundred selfies. Setec reached the endgame when there was a glut of teams and were thus given a delayed start; I went to sleep instead of participating, and several members of my team wished they did the same.

On the other extreme of time commitment, in 2007 (Hell) we gave teams a construction task that was (in my opinion, and I didn’t write it) one of the most elegant puzzles I’ve seen in the Mystery Hunt. We thought that in itself would be a satisfying conclusion, and since teams had done another runaround earlier (and had “found the coin” five puzzles into the Hunt!) we basically surrendered the coin when teams showed up with a valid solution to the puzzle. The reaction to this was almost universally “That’s it?” I’m not a huge fan of the final runaround, but obviously a lot of solvers look forward to it.

On the other hand, putting too much into the endgame risks a situation where too many solvers don’t see the coolest stuff in your Hunt. 2013 (Coin Heist) was a Hunt with a lot of problems (I think that’s been accepted) but it had a very ambitious heist finish with high production values that even had tasks for large teams that couldn’t send everyone on the runaround. I think it was the most creative element of the event, but because the Hunt ran long, I believe only two teams got to experience it. (On the other hand, some cool elements of endgames can still be shown to people who don’t make it there in person; the GlaDOS interaction at the end of 2011 (Video Games) was a great audiovisual experience, and thankfully it translates somewhat well to video.)

Due to the number of teams participating in the Hunt now, it’s ideal to have an endgame that supports multiple teams at once. I’ve completed several Hunts now where there was a queue and we had to wait for another team to finish before we could start… this is not fun, because you usually have nothing to do during that time (though it’s a good excuse to get people to clean up, or in 2012 (Producers), to get everyone on your giant team to participate in a gigantic award-winning production number from Man of La Mancha). My first Hunt was 1998 (Enigmatology) which ended with two teams looking for the coin in the same weight-limited elevator, requiring the two to make a deal to search in shifts to prevent anyone from getting hurt.

I get super-stressed-out during an endgame if I think there’s any chance of being passed… At the end of 2016 (Huntception), our team was broken into a whole bunch of subteams (which, by the way, would have completely sucked were we a small team) and my subteam got lost; Left Out was beginning the previous task as we completed it, so I was panicking that our group of three people was literally going to cost our team the entire Hunt. In 2009 (Zyzzlvaria), I know that one team (the name-changing team I later joined and constructed with in 2014) had figured out enough about the endgame in advance (based on information we designed not to let teams figure out enough about the endgame in advance) that if they had finished their last meta earlier, I am completely confident that they would have blazed past Luck, which would have been exciting for them and supremely unpleasant for Luck.

For me, I’d like the Mystery Hunt endgame to be a brief brisk victory lap with interesting Hunt content and low pressure.  I was really happy with what we ended up putting together this past year (Monsters et Manus); we knew we wanted to have a significant middlegame accomplishment, and I thought it was a great concept to have our RPG middlegame take place with figures on a tabletop hex grid, while our endgame took place with actual solvers standing on a life-size hex grid. We ended with a short-ish runaround which was a lot more dramatic with musical scoring, and because we had three sets of hex tiles and Mystereo Cantos costumes, we ran more than a dozen teams through it without (as far as I know) requiring any team to wait for a significant amount of time.

So what do you like in an endgame? What don’t you like? And what endgames have you found memorable (for good or bad reasons) in the Mystery Hunt and other puzzlehunts? Have at it.

Watercooler: Puzzlehunt Testing Practices

Let’s see how this goes…

One thing I really want to do with this blog is get conversations started about different puzzlehunt construction practices. I’ve had lots of experience constructing in certain areas (Mystery Hunt, BAPHL, NPL Con) and none at all in others (Microsoft Puzzle Hunt, DASH, BANG, The Game, etc.). Different formats and different audiences present different challenges, and I’d love to hear about everyone’s experiences from all angles.

As a starter “watercooler” topic, on my last post eudaemon asked about puzzle testing, specifically in reference to online hunts. In the last year, we’ve seen some online events that were, in my opinion, extremely clean (Galactic, REDDOT) and less so to varying degrees (SUMS, Cambridge). Of course, testing is just as important, if not more so, for live puzzlehunts… a posted PDF may be easier to edit on the fly

I’d love for people to chime in with stories and opinions about puzzle testing. To constructors, what’s worked well for events you’ve helped write? What hasn’t? What are good practices in general, for the people who test and/or the people who organize testing? And for solvers who perhaps haven’t written before, how can you tell when a puzzle probably has or hasn’t been tested, and what do you think would help?

Thursday is usually my “work from home” day, so I’ll try to get in the habit of posting these on Thursdays; though we’ll see based on participation whether there’s actually a demand for weekly prompts. Comment away!