Recap: Galactic Puzzle Hunt 2018

(This is a recap/review of the 2017 Galactic Puzzle Hunt, which happened earlier this month. Puzzles and solutions can be found here. This recap may contain spoilers, but I’ll try to avoid them where possible, especially for puzzles I recommend.)

Much of my puzzlehunting bandwidth recently has been devoted to things I can’t tell you about (or that I potentially could tell you about, but it would disqualify you from Mystery Hunt and give me an excuse to try to rope you in for constructing), but I’ve participated in two very ambitious hunts this month: the second Galactic Puzzle Hunt, and the first Cryptex Hunt. I have a lot to say about the latter, including my experience trying to solve the Finals 1.0, and why I didn’t pick up a daily prize until the very last opportunity, but that event is still ongoing (and recommended). The Galactic Puzzle Hunt wrapped up earlier this week, so let’s chat about it, shall we?

This year’s event departed from last year’s Aussie-style structure to a more Puzzle Boat/Mystery Hunt-like format. The overall theme was Cookie Clicker, and while you could up your number of cookies by manually clicking (which I believe was a reasonable way to open puzzles at the very beginning), solving puzzles was necessary to earn enough cookies to open puzzles at a reasonable pace. Then after the first round (in which the puzzles were much more approachable), cookies split into three separate currencies, with every puzzle transparent about how many cookies it would add of each type (and every next-puzzle-in-a-round transparent about what it would require to be opened). Metapuzzles were the last puzzles to unlock in each round, and so even if they were short-circuitable, you were unlikely to have much of an opportunity to do so. We got bottlenecked toward the end with six puzzles in a state where solving any one puzzle would open nothing, but solving another would unlock two metapuzzles at once. Once we finally did this, we basically insta-solved both metas.

I for one approve of this structural change, partially because that’s the style I grew up on, but also because I find the daily releases of Aussie hunts a bit artificial, and it also makes you more likely to be stuck with only one thing you can work on. Of course, I say that having happened to have most of the weekend free; if I’d had any major conflicts, I’m sure I would have been delighted to solve all/most of the puzzles for a particular day and then getting non-puzzle-stuff done until the next release.

This was a *really* good puzzlehunt. I might have to decide whether it squeezes its way into my all-time top ten. As one of my teammates said, “I hope they don’t use all their best ideas by the time they win Mystery Hunt.” I thought there was so much creativity demonstrated in difficult but fair puzzles with artful and clean execution. If this were the first four rounds of a Mystery Hunt (which is my gold standard), I would consider it on pace to be one of the better Mystery Hunts in history.

For the second year in a row, Jackie and I solved with our adopted Aussie-hunt teammates on Killer Chicken Bones, a group of seasoned solvers that mostly (all?) solve the Mystery Hunt with Left Out. Near the beginning we were mostly dueling with plugh for first place (not counting Cookie Monster, who “won” the hunt by cheating and thus held the top of the leaderboard by a wide margin after a while), but after we reached the hard puzzles, we gradually started to lose ground on the other teams. When I went to sleep, I think we were in seventh or eighth place. In the morning, I nervously loaded the standings to see how far we’d fallen… and we were back in first. My west coast colleagues apparently had a lot more oomph left, and as I understand it, Ian Tullis (a power-solver that didn’t know if he was going to be able to contribute but whom we added to the team in case he had time) showed up briefly and magically turned puzzles into answers). I did not witness his presence, so he may be a unicorn, though I have seen him in person before as a human, so he would have to have been unicornified some time in the last decade.

After that we made steady progress and hung on to the lead for a while, but we then got bottlenecked Saturday night, and in time two teams passed us: have you tried random anagramming (members of [full text of Atlas Shrugged], whose team name is presumably advice for solvers in their 2004 Mystery Hunt) and Tasty Samoas, Please Ingest, which I assume is the cookie version of Reddit-based hunters Test Solution, Please Ignore.

Our six-puzzle logjam consisted of:

1) Unusual and Strange Puzzle Collection, which I’d immediately identified as a Best of the USPC puzzle. We’d solved enough of the logic puzzles to extract a clue phrase showing us where to look for something obviously significant, but we had a whole lot of data with what seemed like no obvious extraction. Jackie and I went back and solved the remaining puzzles, but between the new puzzles, the original USPC puzzles, and the puzzles referenced in the extra grid, we were in giant-spreadsheet-with-no-idea-what-to-index-into mode.

2) Cryptic Command, in which we’d solved some fairly clunky cryptic clues (the “modifications” were loose enough that it was easy to get an answer and not know if it was right) and hadn’t gotten much farther.

3) Adventure, in which I’d explored enough to determine what the puzzle was asking you to do (wander around a virtual keyboard and somehow “type” the same sequence of letters that you were actually typing to do this), but which none of the non-unicorn members present felt they were capable of doing. I read a lot about quines, but couldn’t grok how to make the blackboards process input.

4) Destructive Interference, an excellent example of an “oh you f***ers” puzzle, where shortly after encountering it you realize what you’re expected to do and find yourself glaring at the constructor (in a playful way, honestly!). Rich Bragg had been working steadily with Audacity, and after hours of work he was fairly sure he had the first two letters of the answer. It had thirteen. [This is a good time to applaud the organizers on their solutions and author’s notes, which were posted immediately after the event ended and are incredibly well written. The writeup of both ways to solve this puzzle is fascinating.]

5) Third Rail, which we (correctly) believed was a mashup between two specific board games. We had few conclusions from there on, and none of the ones we did have were particularly accurate.

6) Sequencing: Actually, we had a 5-puzzle bottleneck, since solving this one would earn us chocolate chip cookies, which while delicious were now a useless delicacy for unlocking purposes.

Late Sunday morning, I found Todd working on Cryptic Command, and between us we decided that the answers might be cluing unique casting costs (which seemed unlikely, but these were indeed some cards with very strange casting costs). Between the two of us we managed to work out most of the steps of the puzzle (except the very last one, in which we didn’t work out the correspondences of letter-to-item-in-final-list and just arbitrarily assigned them to make a word) and solve it. Then Todd had the aha for USPC (we weren’t using the hex grid given with the puzzle correctly), and once we knew what to do, our giant spreadsheet thankfully had all the information we needed. As noted earlier, the metas this opened were not hard, and we were then able to backsolve everything remaining besides Sequencing.

This brings us to the very creative prelude to the final puzzle, which I’ll spoil since it likely can’t be recreated in a meaningful way on the website. The text on the unlocked puzzle (which had no answer submission option) strongly suggested security cookies, and we had been informed when we logged in that the site uses them, which I thought was just a thematic pun. In order to access the puzzle, we had to edit our own cookie, which is something I’ve never done or would have expected to do to solve a puzzle. In a hunt where the puzzles were mostly self-contained, it was cool to come across one that was more environmental in nature, almost like playing an ARG.

According to the solution, our cookie should have shown our answers to the security questions from when we registered. Ours actually just said “[our username]_COMPOUNDWORD_MYBACKYARD_ACTOR.” I’m not sure if this means we said our favorite compound word was COMPOUNDWORD, or if we left it blank and this was the default. I won’t say which of our team members did the registration, but now you know they approach their security questions in one of two ways, so hack accordingly. (They were unfortunately unavailable for most of the period when we were trying to solve this, or they might have remembered the terms from registration. Also, I’m sure they take their real security questions much more seriously.)

We actually did almost the right thing hours before we did the actual right thing, as we correctly changed the second through fourth items, but never changed the first, as we didn’t realize we should be trying to impersonate someone. We were looking through the Story page when we realized that mewantcookie was actually a username. This caused me to check out the “forgot my password” link, and I was about to point out that the three strings indicated security questions when Rich tried changing the username and got us to the final challenge.

Said challenge was very cool, but in certain ways it seemed designed to maximize tension and grumpiness between team members. I would have enjoyed it much more if we’d had the ability to independently play with the system (Rich was actually coding one before Kenny Young stumbled upon a number close enough to a reducible one that we could apply our intended algorithm), and most of all, if THE OVEN HAD HAD AN “ARE YOU SURE” BUTTON. An undo button would have been even nicer, but having an irrevocable reset button right next to a button we had to click frequently turned a fun algorithm challenge into a massively stressful experience. By the time we got to this point, Atlas and Reddit had already finished; we had no idea plugh was lurking right behind us, and we ended up beating them by only eight minutes.

I’ve touched on most of the puzzles I spent the most time thinking about above, but a few others worth noting my experience with:

Make Your Own Fillomino, Cookie Clicker, The Answer To This Puzzle Is: I only worked on the second and third of these, but I wanted to express my appreciation for the varying interaction mechanisms that were employed in this hunt. Puzzles weren’t necessarily self-contained, but rather things you had to interact with (well, I guess Cookie Clicker was self-contained in that sense). Another mild example is:

Pride and Accomplishment: Somebody fixed Fifty-Fifty! But seriously, I really liked the initial step (well, five steps) of this puzzle, as well as the assembly… But from there on, the actual answer extraction felt tacked on. In the last stage of solving, it sort of seemed like we were solving a puzzle that was separate from the puzzle we initially set out to solve (apart from a theme-appropriate answer). For me, that detracted from the cohesiveness of the puzzle.

Lips Are Movin’: When Jackie and I woke up on Saturday, a lot of the low-hanging fruit had been devoured. Thankfully this was inexplicably unsolved, and we solved it together sitting on the couch in maybe fifteen minutes. Not a complicated puzzle compared to a lot of the rest of the hunt, but fun and clean.

In conclusion, I loved this hunt, and I encourage Galactic to keep producing it… just don’t make it so good that at some point you burn yourself out trying to meet your own expectations. Mildly fantastic would still be appreciated, especially if you can make it last. 🙂



6 thoughts on “Recap: Galactic Puzzle Hunt 2018

  1. Agreed that it was a really good hunt. I hadn’t actually planned on solving it that seriously, but I got sucked in. I also didn’t expect it to be as meaty as it was, because half of our team blew through the opening round while the west coast contingent was at work. Good thing I didn’t have any plans that weekend…

    I really liked Make Your Own Fillomino. It’s a cool twist to make solvers construct a puzzle. And then you sort of know what’s going to happen next, but it’s still a delightful surprise to see it actually happen.

    I have seen Ian Tullis looking like a person fairly recently. That doesn’t prove that he’s *never* a unicorn, though.


  2. have you tried random anagramming (members of [full text of Atlas Shrugged], whose team name is presumably advice for solvers in their 2004 Mystery Hunt)

    I would have thought 2014 would have negated all of that by now. Also, only Brandy and I were both involved with 2004 and on have you tried random anagramming, so I think we’re the only ones who are supposed to be offended by that. :-p


  3. Now that I’ve read the whole entry and not just the part where you subtly insulted me… 😉

    Our cookie experience sent us down a big rabbit hole. Sometime Saturday, well before we had even opened one of the final three metas, someone (Ankur? Kevin?) noticed that our cookie was “anagram_numberwang_a_body_pranjal”… and Pranjal, who registered us, was asleep. We looked for a lot of ways to apply this cryptic cluephrase — at times, not even being sure if “anagram” was in reference to our username (because “pranjal” identified us also, maybe) or an instruction. Aaron Bader watched a few of the numberwang episodes on YouTube at one point, and catalogued a bunch of data from them. (We were stuck on most of our open puzzles, so this seemed a reasonable thing to spend time doing, I guess.) Then Pranjal woke up and eventually remembered that he had provided all the information in that cookie and we… felt a little foolish.

    On the other hand, when the final meta came along, we breezed through the part where you had to change your cookie in like 5 minutes, which might have been one of our biggest time advantages over the Reddit team. Although that led to one of my other favorite exchanges… apparently Kevin didn’t grow up watching Sesame Street.

    Bader: Oh, we want to use the C word from each of the meta answers to get Cookie Monster’s security.
    Kevin: Why?
    Bader: Because C is for cookie.
    Me: That’s good enough for me!
    Kevin: Wait, why is that good enough?
    Bader: Because C is for cookie.
    Kevin: So?
    Me: That’s good enough for me!
    Kevin: What?!
    (repeat about a half a dozen times)

    Also, it’s worth noting that this means that Pranjal entered himself as his own favorite actor. Are we really surprised?


  4. Thanks for this Dan. I don’t know what motivates Galactic to keep making these, but I now see the Galactic Hunt as a puzzle-hunt highlight of the year. The puzzles are generally fun and clean, they take comments very seriously, the hint mechanism is great, and there’s no need to pay for travel!


  5. I solved on a team of 3 people last year and didn’t have a great time–the hunt didn’t seem to scale well to that team size. This year we tried it again and had a blast. Make Your Own Fillomino is the best idea I’ve seen in a while, Cryptic Command was super clever, and the intro puzzles were all clean and nice. Probably going to try to scrape together a full team next year.


  6. This hunt was great! I am totally rooting for them to win Mystery Hunt.

    It was definitely a little challenging for our four-person-two-toddlers-mostly-remote team (ok, maybe Nikki is a bit of a Nikoli unicorn)… we only unlocked and solved Animal Crackers in the last few hours of the hunt, and unlocked Chessmen with literally minutes to go. Because we weren’t solving in lockstep between the rounds, we didn’t unlock Animal Crackers until we had solved all 8 puzzles, and we didn’t unlock Chessmen until we had solved 7/8. (And with our luck, Chessmen appears to have been the most difficult meta to solve without all the answers.)

    For folks who missed it who are looking for great puzzles, I’d highly recommend “o ea / Wrd srch” as a great combination word/logic puzzle which was challenging for us but still clearly approachable at all points.

    I honestly enjoyed a lot of the education in this hunt. Exceptional Expedition and Sequencing taught me legitimately interesting things about the world and science, and Destructive Interference pushed me a bit out of my comfort zone to learn how audio editing (via programming and manually) worked. (Though I didn’t realize a fact about sample lengths which meant that while I got really close to finishing, my mistake introduced enough artifacts that I thought there were two more very difficult instruments to remove after I should basically have been done. Oh well. I agree that the solution writeup for this puzzle especially was amazing.)

    I’m curious what folks thought about the unlock structure. The parallel unlock structure (where puzzles in one round explicitly helped open puzzles in other rounds but at a reduced rate) isn’t something I’ve seen much before: I’ve certainly seen “solving any puzzle helps open any other puzzle” but not like this, unless I’m forgetting something. It was nice that when blocked on a round we could still unlock puzzles in it by solving puzzles in other rounds, but it was also somewhat weird when the best way to unlock in a given round was to solve a puzzle in a different round! The stinginess of meta unlocks was frustrating for us. We certainly had a different experience from an early finishing team, as the use of hints was a major part of our hunt, and we did get a few of the puzzles time-unlocked. Because many of the main hunt puzzles were very difficult, we expected that the metas might be as well, so we were very cautious about blowing our last hints on normal puzzles rather than saving them for metas. If we had been less careful we probably would have actually made it to all the metas (which weren’t that hard!) rather than ending the hunt with a few unused hints and a few puzzles where a hint would have helped get us to the finish line.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s