Bpuzzled Recap Part 2: 2020-2021
In Part 1, I wrote about what Bpuzzled is and my experience the first year it was hosted. In this post, I write about its second year. Because the puzzles were never made public, I can’t link to them and won’t reproduce them here. I kept an archive, so ask me if you’re curious about them.
Nov 13: Bpuzzled UMD
The 2020 COVID pandemic happened, so I wasn’t hopeful that I would win another trip to New York City, but was nonetheless surprised when I learned Bpuzzled would return. After doing many puzzlehunts from that spring until fall, I felt way more confident than before. Duck Gizzards, an essentially all-UMD student team, began finishing in the top 20 of popular hunts, so I expected to have a shot at taking first at Bpuzzled. It was also my final year at UMD, so I wouldn’t have another chance.
For this year, Josh and Steven returned as team members, while Grant (another UMD Puzzlehunt writer and regular Duck Gizzards teammate) took Ryan’s spot, since he graduated. Bpuzzled kept the same format: you initially start with 3 puzzles and unlock 1 more for every solve. The answer checker confirms partials (and more), and you are given canned hints every 10 minutes after the puzzle is unlocked. There would be 8 puzzles and 1 metapuzzle that you would have to solve in the 3-hour time limit.
Puzzles
Steven and I got to work on Step by Steppe, which was a simple criss-cross puzzle where each entry was a pair of homophones. On this puzzle I used the answer checker a lot to confirm my entries. Josh completed Dance Like an Egyptian, a puzzle visually cluing famous dances (e.g. YMCA and the Moonwalk), while Grant got started on Rogue Gallery, where he jotted down some mashed up the names of paintings and Batman villains. I joined him, helping ID the villains, but got stuck on the extraction, so we moved on to another puzzle, waiting for its canned hints to come along.
I started to work on Spill the Tea with Grant while Josh and Steven worked on Textris. Spill the Tea clued the names of Drag Queens using some descriptions written like they were for tea. At some point, I returned to Rogue Gallery (after receiving several canned hints) to finish off the extraction, which unlocked Mirror Mishap. After completing Spill the Tea and Textris, we unlocked Grinding Your Teeth and Plant One. Grant already started working on Mirror Mishap, so I started working on Grinding Your Teeth, a word puzzle with letters placed on the teeth of various gears.
Some time while I was working on Grinding Your Teeth, my teammates solved Mirror Mishap and we unlocked the metapuzzle. I dropped my puzzle—I could already tell it’d take too long—and went to look at the meta. The puzzle gave you a grid which was a key for a pigpen-like cipher (clearly inspired by pigpen, but it was their own custom code). Each answer was seven letters long, and the presentation suggested that the letters at each index were assigned a color from ROYGBIV. Then you needed to color in the given grid of letters and decode the shapes using their cipher. We didn’t take too long to figure out the meta, but with two missing answers, it took us some time fiddling around to get the answer to fall out. With the solve in about 2 hours, we punched our ticket to the virtual Bpuzzled finals.
Feb 26: Bpuzzled Finals
The hosts started by sending everyone a Zoom meeting link, and during the competition they would split every team into a breakout room (although my team ended up hopping into a Discord call). All I could see were some people’s names, so I couldn’t really get a sense of which schools were present or how many teams there were. The writing team introduced themselves for a little bit, showed us the instruction video that outlined the format, and sent us off to our breakout rooms.
Recalling my experience from last year, I reminded my team that we could take advantage of the answer checker. Even though it rate-limits guesses at 1 per 2 second, there was no penalty and there is a lot to gain from confirming even the smallest parts of a puzzle. We could also use the hint system to our advantage, by moving on to the next puzzle the moment you get stuck, since the pre-written hints will eventually unlock. And lastly, that we should all try to think about our answers as we solve and prioritize the metapuzzle when we unlock it.
I felt pretty confident, knowing that our team had experience with Bpuzzled’s format and having competed in over a dozen more puzzlehunts than last year.
Puzzles
The first three puzzles were Bonds of the Heart, Coaster Connection, and Headlines. I started to work on Headlines with Grant, solving clues that gave words that either had the substring HEAD or LINE. After an indexing step, we finished and were off to start Bonds of the Heart. My teammates working on Coaster Connection started filling out the crossword clues but ended up getting stuck on the extraction. I told them to move on to the new puzzle, Color Filter Tester, while waiting for the canned hints to come up. Coaster Connection ended up wanting us to write letters rotated (as if they were aligned to a rollercoaster track), which we completely missed even though it was heavily clued by the presentation. We were still leading, so I wasn’t really concerned.
Bonds of the Heart gave a word search grid with rows that you had to reorganize. The given left-most column gave a cluephrase, and implied the right-most column would do the same. From here I threw the left-most letters into an anagram solver and was able to mostly reconstruct the grid. In the word search, you found words (which were clued) that were parts of sets of things that included hearts—SPADE, DIAMOND, and CLUB for example. Then you would find the names of each set in the grid (SUITS), pointing you to the answer.
Grant started helping with Color Filter Tester as we unlocked Shooting Hoops. Steven and I jumped right on Shooting Hoops, as it was a puzzle about some NBA games and we are both basketball fans. The puzzle gave the scores of some games and robots whose initials were the same as one player on each team during those games. Drawing lines connecting the box scores and robots, we obtained semaphore giving the letters AFGTT, which didn’t anagram to anything nice. We sorted by various things, then finally game dates, which gave us FGATT. It didn’t look good, but I threw it into the answer checker and it told us that we needed to interpret that as “Field Goal Attempts.” Neither of us had ever seen it abbreviated that way but solved the puzzle and moved on. Glancing at the leaderboard, I noticed we gained a decent lead in solve counts over the other teams, so I knew we had good pace as long as we didn’t get stuck.
Meanwhile, my teammates solved Color Filter Tester, a puzzle about removing letters from words using certain sets of letters (e.g. Roman Numerals). They also started working on Homemade Caesar Salad right after, which was a criss-cross. I helped my teammates figure out the extraction, which involved Caesar shifting some letters by resistor colors, which could be related to cutting board colors. Some teammates started working on the last two puzzles, Prank Bit Shank Fit Fishtank and Double Vision, a wordplay puzzle and a Pokémon puzzle respectively, but when we solved Homemade Caesar Salad, we all jumped to the metapuzzle.
The metapuzzle was mostly pure, the shell provided an order but there was no other given info. I didn’t figure anything about the feeders before we unlocked the meta, so I tried to scan the flavortext for any possible clues. My teammates also stared for a while and I started googling words in the flavortext. I wasn’t desperate yet, but since there wasn’t a penalty, I figured it wouldn’t hurt to start guessing related terms. Best case scenario: the answer checker could confirm something, bailing us out. And it did. I was shocked that typing in something as conspicuous as KEYHOLE (in a puzzle themed after unlocking padlocks) would actually come up with anything. It confirmed a subanswer that I was meant to extract out of a feeder, giving me all the context I needed to piece the mechanic together. For each answer, you could replace a letter with KEY, to form a word. We extracted a letter from each feeder, giving us a hyperlink to a final clue. Solving this gave us the final answer and secured a first place finish in about 1 hour and 45 minutes. I was happy that we won, but more than that, I was relieved we didn’t get walled on any puzzles.
Retrospective
After finishing Bpuzzled, we sat in our Discord call, waiting for the competition to end. Solving the meta locked us out of the puzzle page, so we couldn’t see the leaderboard anymore to watch other teams progress. A member of the writing team hopped into our breakout room to congratulate us and give us a choice among a selection of prizes (we all took the Nintendo Switches we missed out on last year). When the competition ended, they had a short ceremony announcing the prize winners, followed by an author Q&A session.
Similar to the previous year’s Bpuzzled events, I thought the puzzles were well-made with a couple neat ideas. The meta for the finals was clean, and there weren’t any puzzles I was unhappy about. During the finals they started to rely on the answer checker a bit too much to clarify some unclear steps (many weren’t your typical sparse “Keep Going!”, but more descriptive). However, within the context of a hunt aimed at beginners, it worked for Bpuzzled.
On a separate note, many of these puzzles from both years incorporate significant amounts of art, which I really like because it prevents all the puzzles from looking plain or uniform. Personally, I think there could be an overall style guide to make the hunt more cohesive—the font choices are different from puzzle to puzzle, even if it doesn’t add to the theme, for example—but every author puts in some effort to stylize their puzzle, so it gives the impression that there’s a lot of variety and texture to the hunt overall.
Recruiting event puzzlehunts are naturally limited in their scope, since their audience is mostly people unfamiliar with puzzlehunts. Thus, the puzzles end up at best tropey, relying on codes and ciphers, or at worst unfair and frustrating, if written by an equally inexperienced team. However, they do serve as an introduction to puzzlehunts for many students, occupying a specific niche, despite being unavailable to the general public. I’ve been pretty invested in the topic of writing “beginner” puzzles ever since writing for UMD Puzzlehunt, so it’s interesting to study how the runners of recruiting events approach writing a puzzlehunt for beginners, even if they mostly end up using overdone mechanics. I hope these events continue to get support from the companies that sponsor them, since they can really benefit the hobby.
Writing these posts made me realize how different my approach to and experience doing puzzlehunts is nowadays. In general, my team (Duck Gizzards folks plus a few new members) is very conservative about guesses—and I don’t really like to backsolve unless we have a pretty confident guess and made a serious attempt at a forward solve. (However, if the Bpuzzled-style answer checker system made its way into an online hunt, I’m sure we’d spam it.) I’m also pretty apathetic to my team’s ranking in online hunts—I never visit the leaderboard page until after we finish. I still care about improving my own solving ability and finishing as fast as possible, but the latter mostly so that I don’t get distracted after the first weekend that hunts start on.
Thanks for reading this two-part post about some hunts you didn’t do. Thanks to Ryan for providing feedback on these posts. Here’s a bonus write-up:
Bonus: APT/Mastercard Puzzle Tournament
I’m not sure when the first one was, but Applied Predictive Technologies (APT) also held puzzlehunts for college students. In 2019, this became the Mastercard Puzzle Tournament (after APT was acquired). These were competitions only between teams within a school, and I never kept the earlier puzzles, so there isn’t much I remember about them other than doing really poorly my first two years. They didn’t hold a competition in 2020, but in 2021 they held a virtual event where it was a competition with other schools. I did keep these puzzles, since it was a PDF.
I don’t think this event was well-advertised to other schools, since the pool of colleges and teams was pretty small. I ended up teaming with Josh and Jesse (a friend who had done puzzles before, but doesn’t do them as frequently).
Since this hunt was just given as a PDF, I started by staring at the meta puzzle page for the first few minutes. There were 12 feeder puzzles, each marked with a difficulty level, so we all started working independently. I looked for the easy pickings, first working on Time to Escape, a standard semaphore puzzle and Pizza Pi, a puzzle that had you manipulate tangram pieces. I printed out Puzzle Pi and spent most of my time trying to recall that they were called “tangrams,” so I could google the shape formations. Jesse started working on Five Fillings, which had a bank of words minus their vowels. I hopped in to help figure out that the extraction involved binary and went back to looking at the untouched puzzles. Josh worked on Chopped, a puzzle cluing fruit and vegetable puns. Then we worked on Hot Cross Puns, another pun puzzle that had us enter the words in a criss-cross and solve for the extract positions using minesweeper clues.
Jesse moved on to do Note-o-gram, a nonogram, and Josh started on Labyrinth, a puzzle where given some images of the interior, you had to spatially orient a 3D sketch of a house. I thought this one was pretty unique, since it was essentially a visual logic puzzle. I went to look at the metapuzzle with 5/12 answers and was able to make headway, placing our answers on the given grid. Josh solved Labyrinth and gave me just enough letters I needed to pull out the answer.
We ended up getting first, and backsolved all but one remaining puzzle, a duck conundrum style puzzle. Josh volunteered to work on it and earned us a 100% completion before the time ran out. Another UMD team with Grant, who teamed with Sam (Duck Gizzards member and UMD Puzzlehunt author) and one of his friends, got second place. The first place prize was a $100 prepaid Mastercard.
The puzzles we solved were nothing special yet solid, but the meta included some literal red herrings in the form of symbols of fish along with numbers that you actually used to extract. The symbols weren’t meant for anything except being red herrings. Although it didn’t impede us, I’m not a fan of that choice. At some point afterwards, we also solved Hexorflow, which required reading hex codes as words (using the numbers as letters in various ways). The parsing on this step was a bit tough without hints confirming the step, especially because you needed to concatenate codes together to get some instructions. On the other hand, I think Labyrinth was really memorable, since the main solving didn’t rely on typical beginner puzzlehunt tropes. Even though it was rated as the most difficult puzzle, I feel like its barrier to entry was low, which could make it a good puzzle to introduce to people.
Thanks to Ryan for proofreading/editing.