(This is a recap/review of the 2021 MIT Mystery Hunt, which happened this month. Puzzles and solutions can currently be found here. This recap may contain spoilers.)
I’ve gotten some sleep now. Disclaimer: I want to get some Mystery Hunt posts on the record (and give people a forum to comment on the Hunt if this is where they want to do that), but after that, Puzzlvaria will likely be mostly dark for a while. Brown’s semester begins today, so I’m kicking off another period of online teaching, which is challenging in itself, but we’re also expecting a baby boy in March, which will make things exponentially harder.
How parenting will ultimately affect my puzzlehunt bandwidth is certainly not the biggest concern when it comes to raising a child… but for me, it’s not the smallest one either. I’m sure that balance will come up as a topic moving forward on this blog. But for the immediate future, my sporadic posting schedule will probably become even more sporadic. The upcoming puzzlehunt calendar is a lot leaner than in recent months anyway, so there may not be as much to cover.
This year’s Mystery Hunt was a monumental achievement on multiple levels. Even without the constraints of the pandemic, there were some excellent puzzles and metapuzzles, and the “projection device” MMO is the kind of big idea that I would have vetoed out of the gate for being completely unfeasible. Certainly, when we solved the opening meta and watched the video explaining the projection device, my instinctive reaction was, “This is going to crash constantly.” But to Galactic’s credit, apart from some hiccups when the device was time unlocked and thousands of people gained access at once, the thing worked pretty consistently, and certainly more reliably than Cyberpunk 2077. It was also not just a collection of bells and whistles spread on top of a “traditional” Hunt, with unlocks/events/runarounds happening within the virtual world in a variety of creative ways (although I didn’t become aware of a lot of these until wrap-up, which is an issue I’ll talk about). And even though Galactic’s MMO idea pre-dated the need to put the Hunt entirely online, ultimately it allowed for the best possible remote Hunt experience, especially with all the loving MIT-specific detail invested into the virtual campus.
Back in the day, there was a rule of thumb that teams should consider not totally reinventing the wheel in their first Mystery Hunt, since Hunts are hard to write without introducing unpredictable variables, and then experiment if/when they win again. In a certain sense, Old School Setec (2000/2002), New School Setec (2017/2019), Random/Metaphysical Plant (2006/2011), and Evil Midnight (2007/2009) all did this, with the level of second-hunt innovation relative to the era. Left Out made bold choices out of the gate last year (at least in terms of round structure), and Galactic made even bolder ones, and both were astonishingly successful.
I plan to talk about some puzzles and metapuzzles I participated in solving (and have opinions about) in a later post, but while I’m in big-picture mode, I want to take a deep dive on two concerns I had about this Hunt (and perhaps past and future Hunts), despite being so impressed by it; one, which Galactic had minimal control over, is isolation, and the other, which they did, is Hunt size.
Setec usually has a handful of remote solvers, but not enough to warrant full audio/video chat capabilities; in recent years for both solving and writing, we’ve used Slack for most discussion, and an organizational platform wired into both Google Sheets and Slack channels. In a typical year, most of this is intended to make sure information doesn’t get lost between solving sessions, and to allow people to look at and edit spreadsheets on their own screens, although in a lot of cases those people are across the table from each other and can actually speak in meatspace.
For this year, we used Slack plus Zoom, with most of our co-solving interactions happening in breakout rooms. Shifting puzzle-solving from in-person to breakout rooms has a lot of the same drawbacks as shifting classroom math collaboration from in-person to breakout rooms. When you’re in your portion of a physical room, you hear a lot of ambient noise from other groups; that noise can be distracting, but it also makes you aware that other groups are doing things, and as an instructor, I rely on that general sound of the room to know whom I should probably check on. But in a breakout room, you have zero A/V from other groups until you leave the room, so it’s like erecting a ton of completely impenetrable walls in the middle of your HQ. There were a few times when Jackie and I jumped into a breakout room to work on a logic puzzle, and after an hour I’d realize that the entire team could have given up and gone home and we wouldn’t notice, because they were totally invisible to us from inside the breakout room. It was hard to feel part of a team at those times, and I’d sometimes park myself in our Zoom “lobby” just for the human contact.
There was also a self-inflicted element to this isolation; early in the period that the projection device was available, I puttered around the Green Building for a few minutes, unlocked a couple of puzzles, but then swooped on some puzzles and didn’t really look back much. Then throughout the rest of Hunt I heard people saying there were puzzles available to unlock or “field goals” that needed to be done, but I always figured these would be more easily carried about by people who had been paying attention to the projection world, and I should stick to what I’m good at, puzzles. Bu this means that just like I couldn’t hear what was going on in the other breakout rooms, I couldn’t hear what was happening inside the projection device. For me it was mostly a black box… other people went in, puzzles came out. And as someone who usually thrives on having a good mental picture of how the Hunt fits together, I frequently did not.
During the endgame, I actually did have a reason to walk around the projection device, and I found it a lot more user-friendly than I’d expected. I think it still would have been tough to be an “unlocker” without diverting a lot of “traditional” puzzle-solving time to navigating PerpIW. I’m still not sure if I would have been happier making that trade, but my experiences certainly left me feeling like I was missing big chunks of the Hunt. And I’m curious how things were for a small casual team that couldn’t devote some people to navigation and others to puzzle-solving. Though for such a team, I imagine there were bigger problems, with emphasis on “big”…
Devoted Puzzlvarians will remember that I was shell-shocked by my experience of not finishing the 2020 Hunt, and while I thought said Hunt was very well-written, I also thought it was too long. Setec did finish the 2021 Hunt, although we cut it about as close as one could, finishing our last meta (Giga/Nano) at around 9:30am ET on Monday, before the cutoff time of 10am. So did finishing this year make me feel like the size was better calibrated? Nope. In 2020, on Sunday evening when Hunt wrapped up, we were stuck on one meta where we’d solved a big chunk of the round (Cascade Bay) and one where we’d only scratched the surface (Cactus Canyon). This year, we really only finished because none of us had to travel home, and we were able to solve (in shifts, of course) through Monday morning. As for Sunday evening, we were stuck on… one meta where we’d solved a big chunk of the round (Clusters) and one where we’d only scratched the surface. As it happened, we didn’t know about the surface-scratching part until Palindrome found the coin, since we didn’t understand the Giga structure and had no idea there were additional levels below it. We were apparently not the only team to have this issue.
So in a typical year, this Hunt would have run just as “long” as last year’s, and Galactic wouldn’t have come close to their goal of 10+ teams finishing, though that goal may have been based on a hunt-until-Monday assumption.
I’ve been spending a lot of time yammering in the Puzzlers Club Discord recently, and there was an interesting brief discussion of the “epochs” that the Mystery Hunt can be separated into. There’s innovation in every Hunt, but I argued that the key turning points for me are 1998 (rounds and unlocking rather than having all the puzzles at once), 2002 (structure heavily influenced by theme), 2009 (significant variation of structure between rounds), and 2018 (Hunt size and completion time go way up). The more I think about this, the more I think there were two big influences to the Hunt blowing up from 2018 on.
One was, of course, the 2017 Hunt and teams not wanting to repeat the surprise of having a team find the coin Saturday morning. (I know many teams were totally fine with the 2017 Hunt and the fact that it allowed way more teams to reach the endgame. I personally would have liked it to run a bit longer for the top teams, and whether you agree with that or not, having the coin found way before the construction team expects is not the best experience.) Another was the fact that Galactic Puzzle Hunt debuted in 2017; I believe this was a direct result of the 2017 Hunt, since Galactic found themselves with a free Sunday to start writing a kickass puzzle hunt.
The Mystery Hunt always had something of a reputation for puzzle types and interactivity and nefariousness that didn’t show up in any other puzzle hunts (there are immersive puzzle experiences that only show up in The Game, but those feel extreme on a different axis). I believe that GPH was the first online hunt to really start pushing that envelope. It showed you could have online puzzles with crazy procedurally generated elements (man, do I love Ministry of Word Searches) and teamwide interactivity and messing with expectations like team standings and answer submission. Teammate did a great job in their hunt this fall following in those footsteps, and even hunts like the My Little Pony hunt and Puzzle Potluck did ambitious things this year with metapuzzles and structure and plot twists. (Incidentally, have I told people to solve UMD Puzzlehunt from a few weeks ago? Run, don’t walk. It was excellent and came totally out of left field.)
Anyway, I love GPH and hope it will continue to push boundaries in its own lovely way. But as it pushes boundaries, the expectation of what can only happen in Mystery Hunt gets higher, both in terms of individual puzzle complexity and scope of the Hunt itself. And between that and teams erring on the side of not ending too quickly, recent Hunts have only been solvable in a weekend if your team is huge. And then your huge team feels like it needs to cater to other huge teams, and pretty soon there are only half a dozen teams that are really in contention not just to win, but even to see the whole Hunt.
I’m very curious about the experience of small casual teams (especially student teams new to puzzling) in this Hunt. I had a great time, but I had 59 teammates handling the bits I couldn’t or didn’t have the time for. This Hunt had a very friendly opening round that I’d expect small teams to be able to handle, but by Friday evening everybody had been thrown into the deep end, and the organization of what was ahead did not become clear for quite a while. 112 teams solved a metapuzzle, which is fantastic, but did those teams feel like they really experienced this Hunt? Maybe they did, and I hope they did. But at the current trajectory of Hunt size, I’m not sure a team off the street can make a dent in the puzzle structure, and that’s not great for sustainability (both from a writing perspective, and from a perspective of serving the Hunt’s intended audience, and as much as the puzzling community would like to claim ownership, as long as MIT is hosting the event, that audience is the MIT community and especially MIT students).
Okay, I think I’ve said enough controversial stuff to get people sniping at each other in the comments. I’ll post at least once more talking about puzzles I liked (and at least one that I literally ripped into tiny pieces out of frustration). For now, tell us what you thought about Hunt and/or about my thoughts if you’d like to. And somebody generate the team progress vs. guessing graph, because I’m sure I’ll have opinions about that again! (Five minutes? You call that throttling?)