MIT Mystery Hunt 2023 Recap

[Note: I will avoid specific puzzle spoilers, but will talk about structural things and how rounds worked. I am on team NES but my opinions are my own]

I’ll start out by saying I thoroughly enjoyed hunt this year. Being back on campus was great. I thoroughly enjoy Hunt every year—even the past two year of fully-remoteness—which is why I didn’t notice how much I missed the excitement of being back in-person. Near the end of 2022, I was getting worried that I was starting to burn out a bit on hunts, but this Mystery Hunt truly reinvigorated my love for puzzles, and I think being with a lot of friends having a great time is big part of that.

Besides being on campus, the story and theming was excellent. This hunt had a good steady flow of story beats, from the early discovery of the factory, to the visual-novel-interactions, to the post-Reactivation interaction, among other exciting moments. It was also really fun to actually be on campus (rather than in a far-away dorm) where HQ was more likely to visit us regularly, rather than once on Sunday afternoon.

Obviously the elephant in the room is the unexpected length of the hunt, and the necessity of handing out dozens of free answers. I’ll first say that I think teammate handled this well. They clearly acted swiftly and effectively and made some tough decisions that ultimately kept hunt running smoothly enough. It was pretty worrying to see all the hints-as-errata and the flow of free answers, but it certainly could have turned out worse.

I really don’t want this recap to just be a laundry list of complaints—especially because I had such a good time overall and there really were a lot of things I liked—but I’ll try to laser down and be specific about the things that did affect my experience:

(1) I thought the initial round felt very hard, and we expanding quickly into different rounds rather than finishing one. The difficulty was in part plain-old-difficulty, but there were also some deep cuts in the early puzzles. Puzzles that require coding, or reference pretty specific pop culture bits are pretty unfortunate for being that early. I also didn’t like continuously getting new rounds rather than finishing off the rounds we had. There was a point where we had solved all the puzzles we had open in the Science round, and didn’t unlock more in the round (even though there were more to come). This felt pretty bad. This also led to feeling like we were doing terrible. I honestly was pretty doom and gloom Friday night about how poorly we were doing in number of solves, especially since I had such high hopes for our team. This was exaggerated by having a ~2 hour period in the evening (at about 10 solves) where we were stuck on all our open puzzles and I was genuinely worried we just would be stuck for the whole night. I thought once we got past this (team-specific) bottleneck, the puzzles actually felt easier.

(2) The free answers really are hard to use because there’s no way to make everyone happy. Some people just want to do puzzles and some people want to try and get to the metas and see how far we can get. Our compromise was to basically decide if puzzles seemed interesting and if there were people who wanted to work on it. By the time of the AI rounds, this decision was made pretty quickly after unlocking, usually. There still were problems with this, however, because you can still effectively kill a puzzle if you kill all the ones around it. If someone is enthusiastically working on a puzzle they like, and you get free answers for all the other ones in the meta, and you get the meta, then you’ve done a synthetic kill before they might have even had a chance to do that puzzle. The inverse situation also happened, where I stopped people from killing a puzzle I wanted to do (Invisible, in the Admiral Bootes round), and it was clear I was blocking meta progress. What made this worse was the length of the puzzles in this round—it’s been beaten to death elsewhere, but the length is particularly bad here because rather than just working on this puzzle in parallel in the meta, I was blocking things for all (checks timestamps) 11 hours I worked on this puzzle to it’s completion.

(3) The Ascent round was a neat meta and it’s a cool round gimmick, but this meant I was basically uninterested in every puzzle in this round (besides Mosaics). I think a large reason why this was the least-solved round is not because it’s the hardest, but because it limits what the puzzles can be in a not-great way. For what it’s worth, I felt the same way about the Microscope round from Galactic Puzzle Hunt 2020, even though I think that was much more popular among solvers.

I thought the AI rounds were creative and very cool, both story-wise and gimmick-wise. I mostly wish it was earlier in the hunt (i.e. the stuff before it was shorter). I can’t speak too much to the length of individual puzzles in the AI rounds, since I worked on very few of them (though the one I did felt too long/hard), but they at least SEEMED long at a glance. Opening up a puzzle that is so big is not an encouraging thing to see when it’s near the end of hunt and you want to get as much done as you can. Strangely enough, I got the impression that the Conjuri puzzles were noticeably easier than the rest of the AI rounds. We forward-solved (with much time and effort) 4 of the Admiral Bootes puzzles, and 0 in Ascent and Wyrm, but managed to have a good time with several of the Conjuri puzzles.

I really hoped that this would be the year Hunt pulled back the power creep and really took a stand in resetting the standard for how long hunt is. I have also hoped this would be true the past couple of years as well, but to no avail, so I feel a little guilty but also vindicated about not really believing teammate’s reduction in team size recommendation. I am however very glad to see that teammate had that as a goal even if it didn’t work out, and I sincerely hope TTBNL and all future running teams consciously aim smaller as well.

Again, I want to reiterate that I really enjoyed this hunt as a whole, and I’m very thankful for everyone on teammate for putting in the monstrous effort required to pull everything off that they did. I don’t want to spill paragraphs about every puzzle I worked on, so I’ll try to semi-speed through some of things I have something to say about:

(may contain some theme spoilers for puzzles)

• Dropypasta appeals to me in particular as a viewer of Melee. I was happy to start pasting in potential answers to the dropquotes before even trying to solve any of them. The bracket was extremely error-prone, so I might have hated this puzzle if it wasn’t for util.in’s ability to squint the answer out of the gibberish we extracted
• Inscription was a questionable opening-puzzle, but I loved it. The presentation could have been much more boring, so I appreciated it for that
• Showcase was a large puzzle with a lot of steps, some of which were quite the slog. But I thought it tied itself together pretty nicely and was very cool. I enjoyed the aha when watching the video, since all the strange words started popping out at once
• Scicabulary: fantastic puzzle; should have been much earlier in the hunt
• Formula Deluxe: This was pretty cool, but the final step was so painful and hard to do. I tried every possible way I could think of to measure things and nothing worked cleanly (even the method proposed in the solution). Looking at our sheet, I can now kind of see the final cluephrase in the letters we settled on, but we never really got close to solving it.
• G|R|E|A|T W|H|A|L|E S|O|N|G: This was hilarious and it was very satisfying to slowly figure out how the grid worked. It was tricky and we started down several wrong paths, but it was pretty fun in the end. However we never solved this without a hint, because we stopped after four letters of the correct extraction. The extraction here is ambiguous because there’s no obvious way to extract—I agree that the true method is the most reasonable of all the things you could try which is why we tried it first, but there’s a million ways you could think of extracting from this. I will use this in the future as a perfect example of why final clue-phrases might want to start with ANS or ANSWER or ANSWER IS
• Broken Wheel/This Puzzle is Just Another Regular Cryptic: These were both fun, breezy word puzzles with pretty cool ideas/twists behind them. I would always like puzzles like this, but they were particularly refreshing in the context of this hunt.
• Subterranean Secrets: I tried to come in and rescue this from extraction hell. I just want to point out that it’s a huge red herring that the one riddle answer from the second set that isn’t 5 letters long is one of the weakest to me (both as a riddle answer and as a pairing with the other words). We eventually got it but I was using this as evidence we had it wrong for a while. I was also convinced the differing arrangements of the puzzle material on the papers was going to be relevant (another red herring IMO)
• Book Spade: This was really fun. The final extraction step is insanely cool—I’m wildly impressed that it worked.
• Parsley Garden: Much like G|R|E|A|T W|H|A|L|E S|O|N|G I think this is a puzzle that either needed some extra confirmation of the final cluephrase, or a firmer path towards it. We had the final cluephrase written down much earlier than when we sent in a hint to ask what to do. The final step was hilarious, but it turned into more of a salty moment because it was too convincingly not real.
• Zambonis: We didn’t finish this during the hunt, but I think the idea behind it is excellent. I love the games and this makes me want to grind to the harder difficulty.
• Invisible: The dropquotes were hard. Figuring out what to do was fun, but man I must just be terrible at dropquotes. Once I finally got the first of the outer dropquotes (by typing basically the whole thing into util.in), the rest fell from there. But the inner dropquotes were still killer. I was slowly chipping at them but found them near impossible that I figured I was missing something. I asked a hint explicitly to see if I was missing some way to get the word boundaries, and they just sent back the puzzle-specific search website linked in the solution. Frankly, I should have just known this existed, but in my defense there’s no reason why this should exist except for this puzzle in particular. It made the dropquotes so much easier that I could quickly get the ones where it took me forever on the ones I had already done. The extraction also felt incredibly wishy-washy on exactly what to do. I genuinely don’t believe I could have solved this puzzle without having all of (1) the word-length specifying on util.in for the break-in, (2) the puzzle-specific search engine, and (3) backsolve info from the meta on enough of the letters to narrow down what we were doing on extraction.

As for my team, after looking at some of the graphs people have made, I’m incredibly proud of NES for doing way better than my wildest expectations (especially when I was feeling pretty down on our progress on Friday). We’re getting closer than ever to finishing the hunt and I can’t wait until we do—just hopefully a little early than Monday 7am. I’ve learned I really need to adjust my expectations for how good we’re doing, since after unlocking the AI rounds, I was adamant this was the halfway point and there would be a 4th area after the AIs, since it made sense we would be about halfway on Sunday afternoon based on previous hunts/our current progress. It turns out it was intended to be the halfway point in terms of total length, but with all the free answers we had saved up, we were much closer than I thought. Puzzling impostor syndrome is real.

Mystery Hunt was a blast as usual, and while there are plenty of things to complain and discourse about, I’m incredibly grateful for all the memories and fun moments I got to experience throughout the weekend. Long live the Caesar Salad-flavored Candy Canes!