Galactic Puzzlehunt 2018 Recap

Before you read, some useful links:

This post will contain spoilers - you’ve now been warned.

I didn’t get a chance to participate in the first Galactic Puzzlehunt last year, since I was busy taking classes and graduating, but this year as a Master’s student, I actually had (some) time to puzzle. I was mostly free for the first weekend, and was otherwise somewhat free in the evenings during the week. I puzzled with “🌐 really the greatest boat of all is this planet we travel on through space called earth.”, an east-coast offshoot of “Team Runpeng”.

Dan Katz and the official recap post already covered most of the basics, so this will pretty much be a collection of my opinions on the hunt and specific puzzles.

Overall Structure

I’ve done 5 mystery hunts so far with Super Team Awesome, as well as a couple of puzzle boats and Palantir puzzle hunts, so I’ve only ever experienced a sequential-unlock hunt structure, as opposed to an Australian-style daily release, which is how GPH was structured last year. I enjoyed the unlock structure except for the bottleneck before the first meta - we got down to 3 puzzles, and were stuck on Sequencing and Special Snowflake. Thematically and from a hunt construction perspective, I understand why the bottleneck needed to exist, but it was frustrating, especially since our team wound up not ever solving Special Snowflake or Sequencing. Our team hit the “hard puzzle wall” at the beginning of each of the meta-rounds Saturday afternoon, and as a result only managed to clean up 2 more puzzles on Saturday.

Sunday afternoon, we reconvened and managed to work through another 5 puzzles, including Norman sitting in a corner and working through Destructive Interference in about 3 hours. The highlight was when we hit Now Get Ready For…, after which someone pretty quickly guessed that it would be Elf on the Shelf related, but which we didn’t figure out the exact mechanic to for another hour or two, after which we all converged on it and had a fun time guessing ridiculous statements, like “Tall Tree Baby”.

After the first weekend, we cleaned up an average of two or three puzzles every evening until Thursday, when I had to pack up for Spring Break. The team managed to solve an additional 9 puzzles on Friday while I was flying home, including the Fortune Cookie meta and the Chessmen meta. A lot of those solves were on the “easy finish” after the “hard puzzle wall”, where easier puzzles were gated behind difficult/long puzzles.

As Galactic mentioned in their wrap-up, a number of teams weren’t happy with the wall of difficult puzzles, our team included. I also weren’t very happy with the late-unlock metas, since we solved each of the meta rounds evenly, which meant that we didn’t see any additional metas until Friday, even though we nearly solved the entire hunt (and as a result, I didn’t actually get to look at any of the last metas). The “width” of puzzles unlocked at any given time also felt a bit restrictive, especially when out of the 5-6 puzzles we had unlocked at any given time, 3 of them were USPC, Pride and Accomplishment, and (for almost 3 days, but not the full week), o ea / Wrd Srch.

Individual Puzzles

Things I liked: “Fun” Puzzles

There were a number of “fun” puzzles that weren’t particularly difficult, but which our team found particularly entertaining.

Wepp Perflontus: This one was one of the two that I solo-solved (along with Fillomino), which was quite satisfying. I’ve been baking and cooking a lot more recently, which actually really helped with coming up with some of the early recipes, especially dough (though I tried yeast and sugar in water for a while before realizing that this recipe called for salt, which is a bit unusual). This was a pretty straightforward logic puzzle which reminded me of NACLO puzzles that I’d enjoyed in high school.

Parlor: This was our first full-team puzzle, which we solved pretty late at night, when we were all pretty sleep-drunk. The puzzle began when someone (Victor, I think), noticed the “bag wet” / “baguette” clue, after which it was a noisy free-for-all of people saying ridiculous things. Most of the other puzzles were solved in mini-teams of 2-4.

Now Get Ready For…: As I mentioned earlier, we got to this puzzle Sunday late afternoon. Since we co-located for solving on Saturday and Sunday, we had a fun time tackling this puzzle as a team, especially after Victor realized that the background images were all slightly “wrong”, and yelling out “Tall Tree Baby”, which has since become a part of our general lexicon.

Lips are Movin’: This was another puzzle that we tackled as a full team, and was a reasonably straightforward solve. Turns out we aren’t that good at identifying songs though, because it still took a couple of hours once we actually opened it. One really interesting phenomenon that I experienced during this puzzle was that auditory cortexes were fully engaged even though the songs had no audio whatsoever. It was incredibly distracting for other people to be talking in the background, which was also a point of amusement.

Things I liked: “Hard” Puzzles

Make Your Own Fillomino: As I mentioned, this was another puzzle that I solo-solved. We unlocked this Monday night, but no one felt like taking a hard look at it until I did on Tuesday evening. I approached the construction by first placing all of the tetronimos, constraining them with 1’s, 2’s, and 3’s, and then filling the remaining region with as few constraints as possible. I realized that alternating rows of large numbers could be pretty well constrained with very few constraints, since they sort of force each other to grow directly out. After constructing a solved puzzle, I removed numbers until I met the max limit of constraints, while trying to keep things fully constrained. After a few tries and small adjustments, I managed to get it, after which the final extraction was pretty straightforward.

USPC: We unlocked this one pretty early in the “hard puzzle wave”, and Dabin and I chipped away at individual puzzles throughout the week, whenever we felt like we couldn’t make good progress on other puzzles. I solved Geometric Distribution, Kaku Rogue, Magic Order, and Route, Dabin and I collaborated on Almost Yajilin, and Dabin solved the rest (except Cave LITS, which we didn’t ever solve). We tried extracting given the USPC extraction instructions, but didn’t realize for an incredibly long time that the commas and parentheses should be included for the blanks in the puzzles. Victor took one look at it on Thursday night or so and immediately guessed that, as well as the extraction technique, and by then we’d done enough legwork that everything else pretty quickly fell into place. I really enjoyed learning about different types of constraint puzzles - I’m a fan of Sudoku’s and the like, so I really enjoyed the legwork of the puzzle. The biggest complaint I had for this puzzle was that after doing all of the legwork, the extraction clued the current year, which led to hex grid extraction, after which we got “SAME YEARS”. Instead of an actual answer, “SAME YEARS” was just another step in the extraction, which felt a bit rude after a puzzle which required so much legwork. It seemed like one step too far in an unreasonably long puzzle.

Adventure: This was another one that we unlocked pretty early, again in the “hard puzzle wave”. We explored the grid as a team, expecting that it might be on a hex grid, or on a nonplanar surface like many text adventure in past hunts. We pretty quickly realized it was a keyboard, and after managing to stumble upon a combination of letters which let us press a button without immediately resetting, we realized that we would be required to build a quine. Norman and I theorycrafted how to accomplish that for a while. I figured I might try to write a DFS search to solve it, given that the pruning would be pretty good and the depth would be pretty limited, Norman thought he could probably work it out, but we decided to put it aside for the night since it was pretty late. The following evening, we didn’t really look at it, and finished up puzzling around 2am, with everyone pretty exhausted. I woke up the next morning to a message from Norman sent around 5:45am, that he’d solved Adventure - apparently, he had an insight soon after everyone else left and thought he had to work on it until he’d solved it, which he did about 3 and a half hours later. Norman got a lot of kudos for that.

Puzzles that frustrated me

Our team felt pretty strongly that some of the puzzles were poorly clued, not particularly clean, had a number of red herrings, or were just very unconstrained.

Stackables: Though the mechanic was interesting, the final arrangement was not particularly high-fidelity, so extracting “pawn” was extremely difficult (we used filled voxels to visualize for the final extraction instead of just lines, which seemed to be clued by not being able to avoid springs - we assumed you were a voxel traveling through space, which made the “A” of pawn very difficult to identify).

o ea / Wrd Srch: We got to the final phrase of “lk cn r n rgmnt”, to which we filled in “like a canoe or an argument” (pointed), “like a cane or an argument” (extended), “like a coin or an argument” (circular), and a number of other possible answers of varying cleanliness, before finally hitting the correct phrase/answer combo. Some of those wrong answers seemed incredibly clean, and having the only way to reject them be to make a wrong guess was very frustrating.

Exceptional Expedition: I spent way too much of my life on this puzzle. In my opinion, there were too many pieces and it was not obvious which ones paired with which others - we pretty quickly entered giant spreadsheet hell, which was pretty frustrating to deal with as only a 10 person team with other things to think about. This puzzle would have been better suited to Mystery Hunt, where you could have a larger team, with no non-puzzle things to worry about, and thus more effectively manage a large spreadsheet. Additionally, the fact that some dinos (but not others) came from the mentioned lagerstattens was a huge red herring, especially when archaeopteryx was mentioned in November. We thought that we would have to connect the different months in a loop for a long time, and the extraction would be something like “index by Moh’s hardness of month A into the dinosaur genus of month B that should have been in the lagerstatten of month A but was observed in month B instead”. We also missed some of the cluing until very late, which probably exasperated our frustration. Also, Steno’s law is that the newest fossils/lagerstattens are on top, not on bottom, so the final extraction was upside-down. Overall, the puzzle felt a bit too much like a series of disconnected pieces slapped together.

Pride and Accomplishment: this one was a collection of entirely unrelated puzzle mechanics with no cluing, puzzle theming, or notable flavortext. There was no cluing for what things you should look at, and no relation between what you looked at for any of the different “sub problems”. This was blindly “look at this set of words and try to extract something”. If you happened to guess the mechanic, great, but if you didn’t, there was no flavor text to direct what you might look at, and solving one “sub-puzzle” didn’t give any additional cluing for the others, except perhaps what not to do. Also, the fact that there were 151 adjectives (minus 2 overlap between the 2nd and 3rd adjective slots) was a minor red herring cluing “first generation pokemon”. Finally, the pixel art on the periodic table pieces was another minor red herring (the rainbow highlight on 4 elements was very distracting).

Cookie Clicker: There was no way to progress aside from random guessing, with only binary feedback on whether you were right or wrong. When you were wrong, you had to start from the beginning, which was also very frustrating, especially for the longer sequences. We de-obfustucated the code after spending 3 or 4 man-hours listening to annoying beeps. We found puzzle pretty unsatisfying. Normally when you have an “aha” moment on a puzzle, you’re more encouraged to work on it, but Cookie Clicker was the opposite. Once we figured out how it worked, it was just an annoying chore to blindly feel out the sequences.

An additional rant

An additional note on Cookie Clicker: I wanted to highlight what I liked about Word Search to more clearly illustrate why I hated cookie clicker. Word Search was an example of a “blind search” puzzle done well. You received non-binary feedback on your guesses, which let you actually make progress. You could attack the puzzle from different angles (one angle, finding the length of the word, another angle, finding the letters). There was opportunity to be clever, by trying one-off words (the solution gives the example of pin/spin and pin/pine, we utilized a very similar mechanism). You could make ratcheting progress - you never had to start from the beginning, and you never had to repeat yourself.

The key difference here is the feedback mechanism - Word Search had helpful, directional feedback, while Cookie Clicker only had yes/no binary feedback. Tachi, a 2.009 product from 2017 that was a game for blind people, was an excellent example of this - they used a sloped game box to direct blind players attention to the textured cards in the middle, and the exact same principle appears here for puzzles. Directed feedback is absolutely critical for blind-search puzzles.

As a secondary commentary after reading the author’s notes on Word Search, it could have appeared in the “Pokemon Island” round of Mystery Hunt 2018, where the “evolution” was thousands of words you needed to guess, forcing you to write code and implement the algorithm with the knowledge gained from the first, smaller solve. The author’s notes seem almost apologetic for this puzzle, but I thought it was one of the most satisfying in the entire hunt.

Final Thoughts

I ranted above a lot about some puzzles I didn’t enjoy, but overall I thought the puzzles were very reasonable, satisfying, and it was fun to do a less-large-scale puzzlehunt than Mystery Hunt, and especially fun to see the majority of the puzzles without feeling overwhelmed and working alone. I’m really grateful to all of Galactic for putting on another puzzlehunt, and I look forward to participating in the future - just, maybe with a bit more emphasis on fun, unique puzzles and caring a bit less about difficulty or solve times for the top teams. Thanks for reading!

Tags: #thoughts