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!


Joined In Progress – REDDOThunt

A Singapore-based online hunt called the REDDOThunt began this morning at 10am ET and will be continuing for another 35 hours or so. I heard about this event on puzzlehuntcalendar.com a while ago, but I didn’t mention it here for three key reasons:

1) I wasn’t sure what to expect, not having participated in any events created by the authors.

2) The warm-up puzzles they posted didn’t set my world on fire (no offense to the authors, since they were intended for a less seasoned audience).

3) Honestly, I forgot.

But having said all this, I’ve been working through it as a solo solver, and I’m pleasantly surprised; I’ve solved nine puzzles and the first metapuzzle so far, and so far they range from fine to quite good. I recommend checking them out; if you’re going for points, the first-round puzzle values drop when hints come out, at 10am ET Saturday morning.

Congrats to Aviations Laws for already cutting through the thing like a knife through butter, and good luck to anyone already solving or who decides to join the fray. I’ve got a semi-local work conference to attend tomorrow, so I’ll be solving for another hour or so before sleeping and then disappearing for much of tomorrow.

Recap: Mark Halpin’s Labor Day Extravagana 2017

(This is a recap/review of Mark Halpin’s 2017 Labor Day extravaganza, When First We Practice To Deceive. Puzzles and solutions are both posted. This recap contains spoilers for some puzzles, including the final metapuzzle. If you haven’t already, go solve them first, and tip Mark generously.)

This year’s Labor Day extravaganza was another in the late-summer/early-fall series of “puzzlehunts that happen to be on a day when Dan and Jackie also have a soccer game to go to.” Mark mentioned to me beforehand that the puzzles might be a bit easier than usual, but the metapuzzle a bit more involved than usual, and I’d say that was accurate; Mark’s puzzles often have very creative but challenging extractions, and we got stuck on those steps less often than in previous years. The meta on the other hand, was something we made pretty much zero progress on until we cracked it (with most of the puzzles solved). Here’s a play-by-play of our solving experience.

I started with Rabbit, since I’m familiar with that puzzle type (having solved and written them), and getting INTETINE made it clear that the results would be body parts minus letters. Once I had about two-thirds of the letters, I started thinking about answer extraction; I expected the resulting letters to be reordered since the strings were alphabetized, and head to toe seemed natural, but I wasn’t sure where all of the body parts were, and since the thyroid essentially wraps around the windpipe, I dismissed this and got stuck.

Meanwhile, Jackie had started with Cobra, and as I started to spin my wheels, she announced she’d finished one of the three grids, so I joined her and worked on the third while she did the second. We finished them around the same time and got the answer. After I showed her where I was on Rabbit, she got a few strings, and I started drawing letters on a diagram of the human body (with the ones that were unclear to me written next to each other horizontally). From this we were able to fill the blanks at the bottom of the page, but we couldn’t work out what to do with our list of fish deletions.

To change things up, we decided to work on Ape together. It was easy to get started, and Jackie nailed the “backwards” gimmick on the areas that were giving us trouble, which led quickly to a solution.

Jackie started working on Dog (the cryptic) while I worked on Owl, which didn’t put up too much of a fight… Once I finished it, I checked in with her on the cryptic, on which she was surprised that the entries were all going in normally, since she was expecting letters to be deleted; I pointed out that several words had a letter that COULD be deleted (I think she’d noticed this and dismissed it) and we caught that the first six rows could make ANSWER. Knowing what to do, she was able to pull an answer with only about two-thirds of the grid filled in.

As we were solving puzzles, we were getting animal positions on one of the meta grids… I was filling this in as we went, but at this point, we had no insight on the meta. We did go back to Rabbit at this point and I started randomly trying fish body parts… I was surprised that FIN was right, and when I told Jackie, she thought it made sense as a fish’s “limb.” I was expecting something more akin to the other strings (my previous best guest was FIST, interpreting the string as “FISH minus a letter” = FIS and yielding, as the others did, a string one letter short of a body part).

At some point I’d picked up Pig and found a few food items for some of the largest enumerations (PINEAPPLE UPSIDE-DOWN CAKE was a break-in) and handed the puzzle off to Jackie, who got further than I did because she noticed the ones that “hopped” over letters. I was working on Cat and managed to make some ten-letter phrases associated with the villains (MR BABADOOK and BRIDEZILLA were first, though I stalled after this because some of the others had looser associations). Around now, our friends showed up to go to the New England Revolution-Orlando City game with us… I spent another half hour multitasking between finding Cat phrases and being social, and I found some more on the car ride over (thanks for driving, Jackie) but after that we took a five-hour break to watch the Revs crush Orlando City 4-0. Apologies to the Kaka fans who showed up for Brazilian Night and saw their hero get totally shut down.

I solved more Cat phrases by cell phone light on the car ride home, and when we got home Jackie and I teamed up on the rest and worked out the answer extraction (which was neat). We then worked together to finish Pig; once we had all of the food items, we were stuck on what to do and thought the flavotext clued Pigpen, but we couldn’t see how to apply that cipher. I came up with the idea of sorting the letters alphabetically by what they crossed (I figured that was unconstrained enough to be constructable) and once we shaded as directed, we realized at the same time that this was where Pigpen came in. Jackie was mildly disturbed that I could remember the Pigpen cipher without references (it’s a lot easier to remember the order of symbols than with Braille and Morse!).

Jackie broke into Gull by finding PEPPFRMINT while I worked on Horse, which I initially avoided because I thought it was a bunch of separate chess problems. Then I noticed the knight’s tour on the second page and found enough entries to realize the items were cards in Knightmare Chess, which I’ve never played but always adored as a concept. Thankfully, Mark posted a research shortcut so I didn’t have to track down the card text, and I realized why there were thirteen items/diamonds and fourteen diagrams… these weren’t fourteen different games, they were boards from the same game! Jackie was getting tired (it was close to midnight if not past at this point), so I helped her wrap up Gull and she went to bed.

Returning to Horse, I worked out that (5,11,6,14), (13,12,3,1), and (4,7) were consecutive strings, and I figured I’d apply the modifiers to the numbers in diamonds (I still wasn’t sure how those affect gameplay, but Mark explained it to me later) and index into the card names, to force the solver to know which cards were used at which points in the game. Rather than finishing the sequence, I tried those strings in different spots and Wheel of Fortuned the answer. I had also been poking at Raven on and off and had “THE MATRIX” and “HUXLEY” (the latter of which Jackie came up with as the only likely novelist/philosopher) but had little more than that. I also realized that the gray circle enumerations in Anansi’s Web fit the answers nicely, so I filled those in, but was exhausted and had no further ideas, so I went upstairs intending to go to sleep.

But then! I remembered I’d briefly thought about whether the puzzle answers were cluing things (CHANGE GEARS seemed to be pointing heavily toward SHIFT, and WYRM toward DRAGON), but I hadn’t gotten anywhere with those. On a whim I decided to go back downstairs and see if those words fit into the web, and they did like a glove. I was able to fill in most of those (albeit with INCREASE instead of SNOWBALL) and Googling “fiver rabbit” revealed the big aha. From there, each step of the meta fell like a domino, and I was able to submit the two meta answers just before 2am. We named our team The Teal Bunbury Appreciation Society in honor of Teal’s performance in that day’s Revs game (Kei Kamara scored a hat trick, but Teal worked his ass off throughout the game and deserved some recognition as well).

The next morning, I noted the full leaderboard [well, I thought it was full, but since writing this recap, I notice it’s been extended to a top twenty] and was glad I stayed up so that we could make the list, at #6… pretty good given the five-hour handicap! I then wrapped things up by working out the answer to Raven, which I’d backsolved from the meta, came from, getting the aha from SWEET POTAT/JOEY.

Recap: P&A 68 (Louis L’Amour)

P&A Issue 69 will be released in less than an hour, and Issue 68 answers have now been posted. For that previous issue, I took the time after solving to write up a detailed recap of our solving process. It might be more detailed than some care about, but if you’re curious how our team of two approaches an issue, here’s some insight! (WARNING: This post definitely contains Issue 68 spoilers, including one for the metapuzzle.)

Today’s yet another day in which a puzzle event coincides with a Revolution home game; the tailgate lot opens at 4:30pm ET, so we’ll be doing our best to polish off the meta in three hours or so. Wish us luck! (Or don’t, if you’re trying to come in above us on the leaderboard.)

= = = = =

P&A 68

This issue (the meta, at least) was a fairly quick solve; Jackie and I had the first meta solve, and I think it was in just over ninety minutes. We didn’t complete the issue until later that night, but that was partially because we were distracted by grocery shopping, soccer, and Japanese wrestling. So, you know, a typical Saturday.

The first two puzzles I worked on were The Key-Lock Man and Showdown at Yellow Brute, and on both of those I got stuck on the answer extraction. Meanwhile Jackie solved The Man Called Noon (with some assistance from me with regard to time machine trivia), and afterward I talked through my progress on those first two puzzles and we worked out how to finish them, giving us a total of three answers.

Next I solved Treasure Mountain while Jackie worked on Long Ride Home, and after I finished the former, I helped her finish the latter. The Long Ride Home answer was key, as the relationship between TETRAGRAM and MARGARET jumped out at me, and the other answers we already had confirmed the meta mechanism. The answers we now had gave us ??L?C?S??TO?. The context given in the flavortext made SUITOR look pretty likely, but we weren’t sure about the first half. Given the reference to “regressive” in the flavortext, which I didn’t yet realize was a mechanism hint, I thought A BLACK SUITOR might be more progressive if Louis L’Amour’s books were filled with white people (I’d never heard of him before solving this issue), but that wasn’t correct… we then co-solved Lando, and even though the A that gave was already something we’d considered, this narrowed things down enough for us to get the metapuzzle answer with only 6 out of 12 puzzles solved.

Now that we weren’t on the clock, Jackie shifted her focus on other things and I started solving for the complete. I took a look at the letters we needed to produce for the meta, and the only thing that jumped out as a likely answer was FAKED, but I didn’t feel like randomly trying that answer on every puzzle. (Though for what it’s worth, with an automatic answer checker, I see no ethical reason not to do that. Discuss.) I solved Trouble Shooter, which wasn’t bad once I broke in. Then I figured out what to do on Trail of Memories, but once I figured out the extraction, I counted the number of dots, dashes, and spaces, and confirmed that in the right order, they could produce FAKED, so I submitted that and skipped the rest of the puzzle. I like to figure out how every puzzle in the issue works if possible (which we failed at in the Christie issue), but if I know the answer and how it will be extracted, I don’t feel much of an urge to fill in all the blanks.

On Last of the Breed, getting ESP from the top of the grid was enough to get the answer via a wild card search, and once I started making progress on Iron Marshal, I solved the full puzzle, because how could you not? Tall Stranger was tough to break into since I’ve never seen the TV show the puzzle was based around (and there was a lot of data to sift through before deciding what I needed). I thought both of these last two puzzles were good pop culture puzzles; even if you didn’t know the shows involved, both were based around show concepts that were interesting and easy to digest. I thought the hardest puzzle in the set by far was Hills of Homicide, as even when you have a long phrase translated, reconstructing a Playfair square is not at all easy. I think I got three out of six letters (probably the first, second, and fourth judging by the puzzle), and that was enough to solve the puzzle knowing the meta constraint.

Elegant meta and some nice satisfying puzzles, even if the theme this month did nothing for me. As previously noted, the answers this month haven’t met Foggy’s usual standards for entertaining meta answers, but I suspect there’s an additional constraint on those that we haven’t seen yet.

Upcoming: Live and Online Puzzlehunts in September and October

We interrupt your much-delayed breakdown of Bar Exam for a quick rundown of a half-dozen puzzle events coming at you in the next two months. Well, three of them are coming at you wherever you might live; the other three are local, so at most one of them is physically coming at you, unless you live in three places at once, in which case I’m both happy and sad for you.

Hat tip to Dan Egnor’s puzzlehuntcalendar.com, which lists most of these events plus others, and is an excellent resource to keep tabs on upcoming puzzlehunts.

= = = = =

Live Hunts:

The Microsoft PuzzleHunt, probably the event most similar to the MIT Mystery Hunt, goes down September 16 and 17 on Microsoft’s campus in Redmond, Washington. Unlike the Mystery Hunt, MSPH has strict rules on team size and Microsoft affiliation, so unfortunately you can’t just cobble together a team and show up, even if you live near Seattle. On the other hand, if you’re local and you know folks who work at MS, it may or may not be too late to guilt them into inviting you.

If you live in the Boston area (as all the cool kids do), BAPHL 16 will be happening on Sunday, October 8 somewhere in that vicinity. Nathan Curtis is organizing and crowdsourcing the puzzles, and registration opens soon for teams of up to six.

The day before that is DCPHR 2, the result of southern gravity dragging the BAPHL format down to Washington DC. The website is short on details about the upcoming edition, but it does include the archived puzzles from last year’s event; I haven’t solved them yet, but they look fun, and month from now the Halloween theme will feel timely again.

= = = = =

Online Hunts:

One of my favorite online events of the year is Mark Halpin’s Labor Day Extravaganza, which is usually themed on a piece of classical literature or mythology (the very first one was actually related to Labor Day, exploring the Labors of Hercules). This year’s drops the Saturday before Labor Day, September 2. Mark’s puzzles are very tightly constructed and often very difficult; the size of the Labor Day hunt is usually comparable to an issue of P&A, but I still consider it one of the hardest puzzlehunts of the year. It’ll be even harder for me this year because I have to leave for a soccer game four or five hours in… consider this an invitation to win in my place.

(I forgot to add Mark’s Labor Day puzzles to the sidebar until today, so there’s a backlog of almost a dozen excellent extravaganza that you may not have seen before. Enjoy them, but be sure to tip appropriately.)

Speaking of “issue of P&A,” the next issue is released one week later, on Saturday, September 9. It’s issue #69, and I’m sure Foggy will treat that occasion with class and dignity. I will be trying to finish it very quickly, because later that day I’m going to, you guessed it, a soccer game. You guys are killing me.

Finally, Foggy’s much more expansive project, Puzzle Boat 4, sets sail on Saturday, October 14 (the Kickstarter I linked to earlier this year met its goal and then some). Unlike P&A, Puzzle Boat is closer to a Mystery Hunt in size and complexity, so you’ll want to recruit some friends in advance. Thankfully, I don’t have a soccer game to go to on the 14th! Instead I have a wedding. So I was going to assemble my team on the 15th, but on the 15th I have a soccer game. (I wouldn’t mind this as much if the Revolution were remotely good right now. Get well, Kelyn.)

Bar Exam, Part 3: Extravaganza Team Dynamics

(These posts will discuss the creation of Bar Exam (aka Willy Wonka and the Puzzle Factory), which was the extravaganza presented at the 2017 National Puzzlers’ League convention in Boston at the Revere Hotel. Puzzles are posted via the link above; future parts will include puzzle spoilers, but this one will not.)

I really am going to try to have this recap done before I start teaching in the fall. (Originally I was intending to post all four or five parts once a day over the course of one week… You can see how that plan worked out.)

I mentioned previously that the NPL Con extravaganza poses an interesting challenge in puzzlehunt-writing, in that people come for the convention, not just the extravaganza, and thus they’re not necessarily excited about puzzlehunts. Another side effect is that they come to the Con individually, rather than in pre-selected groups, which means somehow they have to be divided into teams. To be frank, this process is not fun… and this is coming from someone who enjoyed figuring out the table assignments for his wedding.

Since I started attending the convention in 2001, in most years solvers have been asked to self-identify as Runners (teams that want to race to the finish as quickly as possible) and Strollers (teams that want to savor the puzzles). This was intended to prevent bad dynamics within a team, where, for example, one team member might want to submit an answer or help solve a puzzle faster while someone else might want to complete that puzzle at his/her own pace. Despite this effort, I know multiple people who have been on runner teams with teammates who wouldn’t accept help or show them the puzzles they were working on, so that self-identification is not always accurate. (I should emphasize the two levels are not intended to be skill divisions, though some people think of them that way.)

On top of that, like almost any organization, the NPL has many members who don’t get along with each other due to various interpersonal issues. When competing in most puzzlehunts, it’s easy to avoid these sort of issues by choosing not to be on a team with someone you don’t like; but again, when solvers are showing up solo and need to form teams, you have a veritable minefield of personality clashes on your hands.

A couple of years ago, the extravaganza designers just let solvers form their own teams, and I don’t actually know how that turned out, apart from the fact that the next year a first-time group constructed and were encouraged to build the teams (as far as I know, since that’s what they did). For my part, I really liked the compromise Mark used when he constructed with Darren Rigby and David Shukan (hey, I just noticed that extravaganza is online!), in which solvers could sign up either solo or as a pair. That way you could guarantee yourself the opportunity to solve with at least one person you like, but you’d still likely solve with some people you don’t normally solve with. We were going to assemble these solos and pairs into teams of four, but test-solving indicated we should have teams of five instead if we wanted teams to complete the event in the intended period of time.

Once you have solvers signed up, the most effective team-generation process seems to be:

Step 1: Randomly sort everyone into teams.

Step 2: Look for personality conflicts (both known and suspected) and switch people to fix these issues.

Step 3: Look for teams that seem particularly strong or weak in terms of solving ability and switch people to fix these issues.

Step 4: If you’ve changed anything, repeat Steps 2 and 3 until you don’t see any issues.

As noted in a previous post, we renamed our divisions Competitive and Casual rather than Runners and Strollers. Since we were forming fivesomes, we also tried to avoid making competitive teams that were two pre-selected pairs and an individual, for fear that the individual might feel like a fifth wheel. To guarantee the competitive teams were equally sized, we asked for a few volunteers to be willing to be in either division, and we formed a few sixsomes on the casual side.

After announcing the signup sheets were live, I was approached by a solver who said they wanted to sign up a full group of solvers who would be a casual team who could be competitive if they wanted to, but would be casual so they could solve together. They asked how they could all be on the same team, and I said they couldn’t under our current plan, since it would be unfair to other teams. After feeling (unnecessarily, in retrospect) guilty about this, we announced a restructuring of the rules where casual teams could sign up groups of more than two if they wanted to.

Shortly after opening that flood gate, about a half dozen full teams of five solvers, many of whom have won Mystery Hunts and have consistently been runners in the past, signed up as casual solvers. Based on previous years we expected about twice as many competitive teams as casual teams, and in practice we had the opposite. This bothered me a lot, since I thought the point system was one of the most unique elements of our event, and I think a lot of people opted out of it to avoid solving with random partners. For the most part I was thrilled with how our event turned out, but this aspect of things definitely left a sour taste in my mouth.

I don’t know the right answer to formatting signups for future extravaganzas. If the divisions are intended to be differentiated by solving style, and people are instead choosing a division based on whether they get to pick their teammates, the divisions are broken. Now having said that, if choosing their teammates is a priority for solvers, that’s an argument for the “just choose your own teams” model. But if solvers who know and like each other all team up, that leaves the new members who don’t know people and the people who cause the most personality conflicts left over to be grouped up… that doesn’t seem like the best model to ensure that new members have a good time and come back to future conventions. As you can see, I have a lot of opinions about this, but I’m also grateful it won’t be my problem to solve over the next few years.

I hate last-minute setup, and we spent lots of time discussing all of these dynamics and planning for them before the event, but unfortunately we couldn’t actually build the teams during that process; we didn’t freeze the team registration sheets until Saturday afternoon, which means that we didn’t get to start building teams until after the flat competition. Delighted to have a legitimate excuse to miss the Con photo (otherwise my excuse would have been “I hate the Con photo”), I retreated to Todd’s hotel room on the almost-top floor of the hotel to start building the teams, while my partners loaded up a luggage cart to bring our equipment to the ballroom, and Kelyn Rowe scored his first assist for the US national team. (Note: If deciding who should be sent home after the Gold Cup group stage were a puzzlehunt, Bruce Arena would come in DEAD LAST. Curse you, Bruce Arena. Curse you.)

Once we had a viable team arrangement, we needed to format it and print it in a way that would allow teams to figure out what team they were on. Unfortunately, I needed to go to the front desk to print, and the elevators in our hotel were notoriously slow… so I’d like to thank the concierge for being helpful and patient when she was approached by a panting, sweaty puzzlesmith who had just run down twenty flights of stairs with a USB thumb drive. After I finally made it back up to the 6th floor where the event was happening and posted the team lists, I momentarily panicked when several people walked up to me saying they weren’t on a team… Thankfully, they were all on a single team that had somehow been left off the printouts, so we only had to fix the sheets rather than our team plan.

(At a recent crossword tournament in Boston, when I was telling this story, Andrew Greene suggested buying a cheap printer, using it to print the team sheet, and disposing of it. I’ve been on a Mystery Hunt team that did this for a weekend, and I’m kicking myself for not thinking of it for this; I definitely would have paid thirty bucks to be able to print in Todd’s hotel room without extra steps.)

We hoped to start right at 8pm (plus time for entry), and because of the frenzy to print team lists and ready the room, we didn’t. But the good news is, most things went much better and faster than the worst-case scenario we were planning for, so the delayed start didn’t matter much in the end. Also, I misplaced my laptop bag during said frenzy, and I was too panicked about finding it to look at a clock and get annoyed about time.

On the bright side, after all this work, not a single solver complained to me, before or after the event, about their team. So everybody was either happy or too polite to tell me they weren’t, and I hope it was the former. Next post, let’s actually talk about some puzzles.

Bar Exam, Part 2: Structure

(These posts will discuss the creation of Bar Exam (aka Willy Wonka and the Puzzle Factory), which was the extravaganza presented at the 2017 National Puzzlers’ League convention in Boston at the Revere Hotel. Puzzles are posted via the link above; future parts will include puzzle spoilers, but this one will not.)

Celebrating my birthday (in lieu of gifts, please vote against ignorance in 2018) by continuing my NPL extravaganza breakdown, which last paused in the middle of what I thought was going to be the “theme and structure” post. Before I move on to structural stuff, a few loose ends in the theme department:

1) If you weren’t at the Con, you might wonder why we had such a generic title. As it happens, the extravaganza title and a brief blurb appear in the official Con program, and being showmen, we wanted a big reveal… so we named the event Bar Exam, and gave a very law-flavored description that was technically accurate (referencing things like “becoming a partner” and having a “good defense” that also apply to the children’s journey through the factory). So much like a Mystery Hunt, most solvers didn’t know the theme until Wonka was introduced.

2) One of my favorite ways to contribute to a puzzlehunt is writing the opening skit, and (to pat myself on the back) between Mystery Hunt and Bar Exam, I’m on a roll this year. But with both events, the script wouldn’t have worked if the performers hadn’t sold it, and Mark, Brent, and Todd did a great job with their characters, and Todd wrote a lovely parody of “Pure Imagination” that we edited down for time.

This is also the second event in a row where multiple people came up to me and said, “You just did that theme so you could play [Character X], huh?” when in fact the character I played was not my first choice! (Though Wonka was a really fun character to improvise at the answer-checking table, especially when actual kid solvers came up to me. I was originally going to do an Ed Wynn voice like I used for the Mad Hatter in the 2014 Mystery Hunt, but fortunately I tried it in the privacy of my home and determined I would have been hoarse after three minutes.)

One of the biggest challenges of any puzzlehunt is creating an event that will work for solvers of varying goals and skill levels. Anyone who’s written the Mystery Hunt will tell you it’s virtually impossible to hit a target, even if you can agree on what the target should be. Many Hunts have gone controversially long before the first team finishes, and this year plenty of solvers were surprised by how early the first team found the coin; I’ve been on construction teams in both instances, and the outliers don’t happen on purpose.

The NPL convention extravaganza is like an extremely scaled down version of this problem; the extravaganza begins at 8pm plus epsilon, where epsilon depends on how long it takes the constructors to ready the room and for people to find their seats, and most solvers proooobably want to finish some time between 10 (maybe even 9:30 depending on taste?) and midnight, with the target varying based on how much people like to struggle, how much they like to sleep early, how much they want to play after-hours games and/or drink, and so forth. The audience is also a little unusual in that they’re not as targeted as you’d see in the Mystery Hunt and regional puzzlehunts. People who participate in DASH probably have some specific interest in puzzlehunt-style puzzles, whereas many (not all!) NPLers are more crossword enthusiasts than anything else, and they may scowl at you if they think you’re forcing them to solve something that’s too logicky or too aha-based. And from previous extravaganza-writing experiences, I can tell you that some (not all!) of them are not shy about complaining if they don’t get what they’re hoping to get.

Last year’s event, written by Erin Rhode and her partners in crime (I’m not sure exactly who helped because I think some initial participants dropped out) erred on the easier side, which reduced the danger of things running long (probably a good direction to err for a first-time group) but also meant my team polished off the puzzles in a little over an hour. So we approached this year’s extravaganza with one primary goal: Can we create an event that satisfies solvers that want something light or something heavy, and in particular, can we do that without our having to linger in the ballroom past midnight?

It was pretty clear that we weren’t going to keep both a super-solver team and a more casual team busy for the right amount of time unless those teams potentially solved different amounts of puzzles while still feeling satisfied with what they accomplished. There was a food-themed extravaganza years ago that dealt with this by providing “snacks.” If a team finished their puzzles early, they were given snack puzzles, which were less puzzlehunty and more Weekend Editiony, in the style of “solve as many of these twenty things as you can without extracting a final answer.” But that year, ranking was still based around solve time… the snacks were just for fun. We wondered if we could make those snacks matter for the teams that wanted to work on them, without overwhelming teams that might not have time to complete them. I’ve also been recently interested in hunt mechanics where you have a main tier/map/round that branches off as you solve stuff, as we looked at structures of this flavor for rejected themes in both the 2014 and 2017 Mystery Hunts.

I forget who proposed a point system, but perhaps it was natural that it arose since all four of us had been prolific writers for Shinteki’s Puzzle of the Month. For anyone who wasn’t at the event, here’s the structure we ended up with: Rather than the traditional “racer” and “stroller” team designations, which are intended to prevent people who like to aim for the finish line from solving alongside people who want to finish puzzles completely and savor them, we named the teams using the relatively isomorphic terms “casual” and “competitive.” (Solvers didn’t form teams in quite the way we expected… much more on the task of creating teams and some opinions about handling it in the future will come in a future post.) We told teams that competitive teams would have an opportunity to earn points to determine their ranking, while casual teams would just be working to finish and would not be racing.

We gave teams all nine “main” puzzles in their initial packet. These nine main puzzles had answers that could be used to solve Mark’s beautiful “great glass elevator” metapuzzle, which featured a physical elevator for teams to interact with. Each time a team solved a main puzzle and submitted the answer, they earned one of nine corresponding “bonus” puzzles. Bonus puzzles didn’t require any knowledge of the main puzzles that spawned them, but we tried to relate them thematically when possible (for example, solving Wonkavision Labs unlocked a puzzle about the Gum-Stretching Room, which is similar to the taffy-puller where Mike Teavee is sent after Wonkavision shrinks him). Teams would also receive the metapuzzle when they solved seven main puzzles; we intentionally wanted to give this out before all main puzzles were solved because the metapuzzle begins with a transcription step that might not be easily parallelizable, and we didn’t want team members to be stuck watching.

Casual teams could solve the bonus puzzles if and when they wanted (so we weren’t withholding any content from them), but their main goal was to solve the nine main puzzles and the meta. Competitive teams, on the other hand, had an incentive to solve those main puzzles and solve them quickly: Bonus puzzles would each be worth 20 points to the first team that solved them, then 19 for the next, and so forth, until they would stay at 10 points for all teams after the tenth. This is a scaled-down version of the scoring system from Shinteki POTM, and from my perspective, that system isn’t that interesting for a single puzzle, but when you combine it into annual standings, it creates an interesting dynamic challenge that rewards solvers for consistent accomplishment. In a sense, we set up nine simultaneous unlockable POTMs for competitive teams, which a “big board” showing everyone how many points were still available for that puzzle. This created an interesting strategy option, since teams could see which puzzles were ticking downward in real time, which would help them determine paths of least resistance… but on the other hand, those paths of least resistance were becoming less valuable as they became more evident. Competitive teams still had to finish the meta to “qualify,” but their ranking would be determined purely by point total.

Competitive teams that solved seven main puzzles may have been surprised to be given a “redemption ticket” for the meta rather than the meta itself. Some of the constructors became concerned that, with the structure we’d chosen, it would benefit competitive teams to put off solving the meta until they were finished earning points. We worried that teams might not realize that (which I felt was just part of strategy) and/or that teams would be tempted by the cool elevator sitting on their table that they shouldn’t touch yet (which I acknowledged could be an issue). The compromise was to *allow* competitive teams to get the meta, but to force them to give up their ability to earn bonus points if they wanted the meta. This meant that, as intended, competitive teams generally didn’t redeem this ticket until they solved all the bonus puzzles or the “bonus puzzle time limit” expired, whichever came first). I initially thought this was a silly thing to worry about, but as it turns out, this was a really elegant compromise, and I’m glad my wiser colleagues convinced me it was worth introducing one extra step per team.

Still to come in the future: Team dynamics, puzzle commentary, and how this whole thing went in practice. (Spoiler: Really well from our perspective, thankfully!)