2023 Mystery Hunt, Part 1: Less is More

(This is a recap/review of the 2023 MIT Mystery Hunt, which happened this month. Puzzles, solutions,and solving stats can currently be found here. This recap will contain spoilers.)

Erratum on Tue 1/17/23, 4:36 PM EST: The narrator of this story may occasionally be unreliable.

Mystery Hunt was back on campus this year, and I applaud Teammate for the Herculean organizational effort that that must have been involved. I know from working at a university myself that over the course of the pandemic, a lot of administrative procedures have been interrupted and complicated by staff turnover and reluctance to reinstitute stuff that wasn’t necessary when classes were remote. All things considered, MIT is very supportive of Hunt (if you don’t think they are, name another college that yields this much classroom space to an event they don’t directly supervise that doesn’t directly give them money), but getting everything you want when running Hunt is still a delicate process in the best of years. This may have been the worst of years, but apart from clearing out of HQ overnight, things felt shockingly back to normal.

Jackie and I decided that Simon is still too little for on-campus cameos (particularly with XBB raging in New England; I still seem to be healthy thanks in part to Setec’s requiring everyone to test every morning, knock on wood). I essentially asked for Mystery Hunt for Christmas, so effectively I got a weekend pass to stay in Cambridge for the entire Hunt, and Jackie showed up on Saturday when we could get babysitting suport during the day. I’m hoping that next year we can find a way to have us both there longer, because I especially missed solving with her during the quiet morning shifts (though I also missed large chunks of those shifts anyway, for reasons I’ll describe).

I parked at the same Red Line garage I’d discovered as my go-to Hunt Weekend access point in 2019 or 2020, got to campus bright and early, and got lucky twice in that my hotel room at the Marriott was available early, and I was able to find a kiosk quickly to print a physical alum ID. Unfortunately, neither my mobile ID or physical alum ID was able to open a single door all weekend. I’ve got to get that looked at before next year.

Setec has used a non-classroom Stata HQ for years, but since 2020, our connection has been severed, so we were in the classroom HQ pool for the first time in ages. Even if we’d still had our previous home base, maintenance work in Stata would have displaced us anyway. I had some trepidation about how much the change in location would faze us, and the answer was surprisingly not much. There was a classroom next to us reserved for The Team… To Be Named Later, which they apparently decided not to use, and which they even more apparently (by now) didn’t need. If our classrooms had been too crowded, I assume we would have hijacked their room once it became clear they weren’t using it, but since we didn’t, I guess we were ultimately fine. We also had an overnight HQ booked at the Marriott, which was very helpful (we had arranged for a suite, which flooded shortly before Hunt(!), and the staff at the Marriott was gracious enough to find us some meeting room space and waive some of the rules that had caused us to choose the suite in the first place). That worked out nicely and definitely soothed some of the sting of having to relocate a couple of times per day.

When Hunt runs smoothly, I tend to focus on metas whenever possible. Maybe that’s just an internal yearning for glory, but I also just like getting big ahas and solving around things. When we were solving 2021 and 2022 remotely mostly in Zoom breakout rooms, it was often hard to find the people who were thinking about metas. One of the best things about being on campus again was being able to hover near the front of the room and get involved with the meta brainstorming easily. We only solved two metapuzzles on Friday, but I ended up doing a lot of work for both of them on paper. Our first meta fell after someone found the restaurant connections for Atrium, and then I solved the Waiterlink (which probably could have been done in a spreadsheet, but it was faster just to draw lines).

And I also ended up being the pencil guy for a small group collaboratively labeling molecules on the Science meta. The latter is in a category of metas I often don’t appreciate, since it’s much more of a self-contained puzzle that happens to have the feeder answers as givens than a final stage bringing the answers you’ve gathered to life. But among metas of that ilk, I thought the Science meta was really nice; the examples given did a good job gradually revealing and confirming what color meant, what order meant, what interactions were happening, and so forth. Visually it reminded me of the Feynman meta from 2013 (maybe not the best meta collection to emulate), but I found this solve path much smoother. I usually try to be one of the team members that goes to sleep early and wakes up for the morning shift. We started making Science progress when I was about to leave, and it kept me up for at least another hour and a half.

I also typed STOP at Mate earlier in the day, but not until we reached the point where puzzles weren’t loading. I had noticed much earlier that there were weird square sequences but assumed it would come up in the Science meta (which wasn’t open at that point), and when this opened a new level, I was very pissed at myself for not trying anything an hour before I did.

Hunt is not all metapuzzles, so for the record, some of the early puzzles I enjoyed working on near the beginning of things included Inscription, Museum Rules, Natural Transformation, Scicabulary (someone mentioned “Leakfast,” and I said, “Hey, I think I wrote that Shinteki Puzzle of the Month years ago!”), and One of the Puzzles of All Time, which was morbing great.

I wasn’t as consistently involved with metas on Saturday; Art was solved entirely while I stepped out of the room for five minutes, and I never really got what was going on with World History. I did help get the aha for Natural History, but I was mostly pleased I could guess the appropriate wrestler who’d held the KOD 6-Man Title without looking him up. Fun non-metas I worked on (some of these may not have been on Saturday) included Baking Bread, Redacted Recipes, and This Puzzle Is Just Another Regular Cryptic. And Think Fast was neat, though we got incredibly stuck on the final puzzle, even after winning the game step and the in-person testing step. Sometimes things that have three parts should only have two. Foreshadowing.

And one more time I did not get to sleep as intended, becasue around dinner time someone figured out that the factory gizmos altered Innovation puzzles. This was a fantastic mechanism, and once we opened the meta, I got very sucked into it. Unfortunately, we made some spreadsheet errors when solving the wordplay bit and placed old and new letters in inconsistent columns, and so we didn’t see the first phrase we should have extracted. This means that instead of the likely solve path (submit that phrase, be told your gizmos are wrong, fix the gizmos) we followed the not-recommended Setec solve path (don’t see a phrase, figure out all the gizmos, spend a long time trying to extract an answer from the gizmo positions, eventually fix the spreadsheet and submit, never seeing a message telling you your gizmos are wrong). The bright side of this is that while analyzing the gizmos, I drew a graph almost exactly like the one we ended up being given in the Factory Floor meta (same orientation and everything!). We’d also already written down the Factory answer-gizmo sequence, so we had everything in place to solve the Factory meta; that kept me up a solid three hours later than desired, but as a result, we did have the fastest unlock-to-solve speed of any team in the Hunt for the Factory Floor meta.

Sunday morning, after a decent but probably insufficient amount of rest, I came back to HQ and soon we opened the Basement meta, despite having opened the Office round first. We caught onto the recycling and plastics code theme quickly, and after a few false starts at how to approach things, we managed to solve the meta and discover the drives in the basement. After that, we focused on the Office, used a few free answer chits and solves to open the meta, and solved it in a big group, following a correct hypothesis that got much better once Josh Oratz proposed that SOLID YELLOW was a pool ball rather than a naval flag. Now having gathered all the key elements of the story beat (Mate’s having a rough time! There are mysterious hard drives in the basement! There are other weird AIs about!), we solved the four reactivation puzzles and initiated a final runaround in which we used the new AIs to improve the factory and brought our Hunt in for a gentle landing on Sunday afternoon.


…except the end of that last paragraph isn’t true. What actually happened is I woke up Sunday to be told that there was an entire third level of the Hunt we hadn’t unlocked yet. We did solve both the Basement and Office roughly as described, although we were given Reactivation to solve before we finished the Office (apparently because fewer metas were now being required for that step), which made finishing the Office very anticlimactic from a story perspective, and possibly unnecessary for advancement. But more importantly, completing Reactivation didn’t trigger the endgame. Instead it opened FOUR MORE ROUNDS of puzzles, many of which felt long for the sake of being long, arcane for the sake of being arcane, and I’m sorry to say, poorly edited (though probably not for the sake of being poorly edited).

I am speaking purely for me, and not for Setec or for the Hunt community at large. But for me personally, the AI rounds took what could have been a really great tightly constructed Hunt, and ruined it. While I have considered trying to retroactively brainwash myself into believing we finished Sunday afternoon, I’m not going to do that, both because I don’t have access to the right medications, and because I think there are some vitally important lessons to be learned by the Hunt community from this year’s Hunt. I apologize in advance that this will probably involve some harsh criticism. I again want to emphasize that I really liked most of what we encountered in the Museum and Factory. And also Dispel the Bees. But I value quality over quantity (most of the time), so in the next post, we need to talk about why more can sometimes be less. After I calm down a bit.


23 thoughts on “2023 Mystery Hunt, Part 1: Less is More

  1. A few thoughts:

    1. You’re right that this was overly long.

    2. That said, I think the later part was better than the earlier part. The Conjuri and Eye metas were hot messes, but the Boötes and Wyrm ones were not, and a lot of object-level puzzles were cool (Bees, yeah…). Plot-wise, I think it would have been better to shorten the museum part – it was a decoy theme, like the workshop in 2015 or dog breeds in 2016. Having a double decoy is an interesting experiment but I don’t think it worked; the Hunt rebooting and transition to the AIs should have been at least 24 hours earlier.

    3. Because the Hunt was overly long, the tiredness factor detracted from a lot of puzzles that at a better time would have been great. Personally I’m annoyed that the Wyrm meta involved the Mandelbrot set and it was 4:30 in the morning my time and I had been up since 8 and eventually I just couldn’t stay up and the rest of the team had to figure out how it did relate to the Mandelbrot…

    Liked by 2 people

    • As somebody on a team that is in a place to potentially win and write hunt in the next few years, I think it’s important to understand just why the hunt was as long as it was and what lessons we can learn from that. In retrospect, it seems obvious that the hunt as planned was doomed to go on until Tuesday (or even Wednesday taking into account fatigue). Double the intended time! How could one get this so wrong and what can a writing team do better in the future? It cannot just be a matter of first time writers being unprepared, because Left Out and Death and Mayhem put on spectacular first-time productions in recent years. Is it simply a matter of totally forgetting to add in a fudge factor of mid-hunt tiredness to estimates of puzzle solving times? Where did it all go wrong? Some very important lessons can be learned from this experience as we learn more about what happened behind the scenes – it is hard to tell just from the finished product what kind of flawed reasoning we should try and avoid ourselves in the future. I am taking this year personally as a message that my own team may not be as ready to win hunt as we think we are!

      Liked by 3 people

      • One thing is simply that estimated hunt difficult is really difficult! Even the most knowledgeable and experienced writing teams will frequently miss by a factor of 1.5 or more.

        Back when Plant wrote for the first time all the way back in 2006 it was shortly after the Time Bandits hunt which was interesting creative hunt that lacked goid judgement and appropriate difficulty and as a result went way too long. There was a lot of fear that we’d repeat that. One thing we really took to heart from the advice Setec gave us was that this was our first time writing, but if we did things right it wouldn’t be our last time writing, and so you should err on the cautious side your first hunt. I think this is great advice for a writing team. In our case we did get to write again and go for bigger ideas, but even for teams that haven’t written a second hunt having a more cautious clean first hunt I think is the right approach .

        Liked by 3 people

      • @Noah Snyder

        Sage advice from Setec Astronomy, thank you for passing it on. I am glad you got to write again, 2011 was one of my first hunts and I have many great memories from it.

        Liked by 3 people

      • Speaking from LeftOut, the people in LeftOut had a lot of experience writing different kinds of puzzle events like weekend-long driving puzzlehunts and the googol hunt. It was our first Mystery Hunt, and the first time us we all worked together, but for most of us it wasn’t really our first puzzle writing experience.


    • 100% agreed that the museum round was obviously fake from the kickoff and especially from the factory opening. Why did we have to solve five rounds of it before getting to the good stuff? The highlight of our Hunt was the reactivation interaction, but that happened late Sunday evening for us. People had left already, and people were too tired to do much afterward.


  2. Looking at some of the puzzles in the rounds my team didn’t unlock … A (just one of those things is enough for a puzzle even in Mystery Hunt) … Author’s notes that say “This puzzle being solvable at all was honestly a huge surprise to me” … A puzzle with almost 200 jumbled and intentionally maliciously obfuscated trivia clues … A puzzle with 147 fables with unnecessarily long padded-out cluephrases … A video game with an unremovable item whose only purpose is to make the game harder as a mean trick to solvers … It all speaks to writers with bad judgment or blatant disregard and disrespect for our time or both. Normally I would be disappointed with my team’s performance but in this case I am glad we did not get this far.

    Liked by 4 people

    • Not to mention a puzzle involving obscure slang and word pronunciations in a foreign language … that is also a cryptic crossword … that is also a diagramless. “long for the sake of being long” is right!

      Liked by 1 person

  3. As someone who is on a mid-tier team, I’m actually quite glad of the general shift towards a split point, such that the teams not in contention for the “win” get what feels like a satisfying finish point at about the point where the full Hunt itself will be over, whilst also accepting that there are going to be whole rounds of puzzles that we were (a) unlikely to see and (b) even more unlikely to solve.
    (FWIW I’m really proud that I spotted STOP very early, but I actually think that might have distracted us more than it helped us. The Innovation round was excellently done though with just the right level of enraged shouting when someone who hadn’t seen all the warning notes changed a gizno.)

    Liked by 1 person

  4. Leading my team in solving the expert level of Think Fast was the highest point of this year’s Hunt for me… and then yeah, we had no clue what to do with that final bit and some of my teammates spent several hours on way too much complexity. (Did you know that the expert-level anagram only uses 9 [I think – some very small number] words out of the >100 possible? That has to mean something, right? Oops.). We got so used to everything having at least one more level than we expected/thought reasonable that at least twice we didn’t both calling in an answer because we assumed we hadn’t gotten there yet. Oops again. (Weaver! Is four unlock rounds/re-solves to get to a really cool and apposite answer phrase not enough?)

    The other irritation for the weekend is ALL us, not teammate: at least twice we had puzzles where we did exactly the right thing, first time, but because of multiple people working together we had data entry errors and hence the readout did not make sense… until someone carefully checked everything a day later. Grrr. Need to remember that for next year.

    We (Central Services) are working through Hunt still. Once we finally spent all of our unlocks, progress through the Office and Basement metas was swift (like, <1 hour, and we solved reactivation before finding the drives which was pretty whack). To their great credit, teammate are still responding and then unlocked the drive rounds (I assume that there was an interaction that would have done this during Hunt), which is where we sit currently. Despite the really quite high disappointment at wrap-up (wait, you mean we solved all of those puzzles and only made it like 30% of the way into Hunt?!), the individual puzzles were to me universally clean and often fun – we had a super time all weekend and never hit the wall of frustration that sometime occurs. Still, geez, if we had realised that we could (I think?) have shunted our efforts from the Museum (where we solved basically *every* puzzle) to the Factory rounds, earlier, maybe we would have felt more satisfied. Overall I think that I have fewer irritations than some here – just wish (as almost every year) that the whole thing could get a little more achieveable. We're an experienced, if not top-level, team: to not even have reached the 'midpoint goal' made us feel bad :).

    Liked by 2 people

  5. I hope that future teams do a significant rethink of puzzle unlocking on Friday. I’m on a team (Metaphysical Plant) that for a decade usually finished hunts, but since the size made an enormous leap after 2017, we now know going into hunt we have zero chance of every finishing a hunt and the leading teams go literally twice as fast as we do. And we’re far from the only team in that situation, the teams at about half the pace of the leaders are lots of old strong teams that traditionally finished hunts (codex, hunches, IIF, etc.) And the way Friday works right now is *unfun by design* out of some misguided attempt to slow down the leading teams (which doesn’t even work anyway!).

    Currently the way things work is:
    1) Friday you have very few puzzles to work on and lots of people and its kind of miserable because every puzzle either has 15 people working on it, or it’s a puzzle that 15 people already got stuck on and you’re probably never going to solve. And in this hunt you can’t even get around the puzzles you got stuck on by solving a meta, because the very first meta didn’t even open until around midnight!
    2) Saturday is great. Lots of fun. Plenty of puzzles. Plenty of metas.
    3) Sunday is kind of a mess, because it’s unclear what to prioritize and there’s just so many puzzles and there’s so many hints and free answers available that it’s hard to decide whether you should even bother working on a puzzle when you could just get a free answers.

    I wish more of hunt could be like Saturday. There’s just no reason to bottleneck Friday the way that happens nowadays. Somehow in the 20 years I’ve been doing hunt the total number of puzzles has gone up by like a factor of 4, while the puzzles that are available to work on at 3pm on Friday has gone *down*. It just makes no sense! And the increase in number of puzzles solved on Friday would really help to get ahead of the flood on Sunday.

    Why are we designing puzzle unlocking mechanisms to further slow down teams that already know they have no chance of finishing hunt? Just let us solve your puzzles! You spent so long writing them, why not let us see work on your puzzles instead of sitting around on Friday trying to find stuff to work on and bashing our heads on the two puzzles we happened to get stuck on?

    Liked by 4 people

    • It’s interesting what you say about Sunday tending to have many many puzzles. Usually that is the case, but this year, I (on the same team as Noah) actually felt the opposite! My frustration at not having access to enough puzzles and just wanting to see the puzzles that the writing team had presumably spent countless hours writing, which I usually associate with Friday afternoon, reappeared on Sunday!

      By Sunday afternoon it was apparent that there was some sort of plot or meta bottleneck that was preventing us from unlocking the next phase of hunt, which turned out to be the new AI rounds. We finally solved the metameta and reactivation around 10pm on Sunday night, just as I (and most of the rest of the team) were finally packing it in. When we started to see the new rounds, all I could think was “well, there’s all those puzzles that I’d wanted to see earlier today.”

      I guess on some level I understand the desire to have a coherent sub-section of the hunt that intermediate teams can feel good about having finished. But I don’t think it’s worth it to have something like that if it means starving solvers of puzzles for almost the entire hunt.

      Liked by 1 person

    • “Why are we designing puzzle unlocking mechanisms to further slow down teams that already know they have no chance of finishing hunt? Just let us solve your puzzles! You spent so long writing them, why not let us see work on your puzzles instead of sitting around on Friday trying to find stuff to work on and bashing our heads on the two puzzles we happened to get stuck on?”

      I can speak to this a little bit… Whatever unlock mechanism you install at the beginning of the Hunt is going to apply to all teams, and having too large a width of puzzles gives an advantage to large teams. Constructing teams that are trying not to let giant teams run away with the lead might not want to have a large width right away. Maybe there’s an argument to have a large width early and then shrink it down once you get past the opening round(s) that super-teams will speed through. But shrinking width mathematically means solving some puzzles has to give you no additional puzzles, and that generally results in a low fun factor.

      There’s also an argument (made by actual small teams) that having a lot of puzzles at once is not always desirable, since it can be overwhelming or not make it clear what would be advantageous to work on. So from that perspective, having a small number of puzzles early might actually be preferable for some small teams. But if you go this direction, it’s vitally important that those few puzzles are extra-approachable, and I don’t feel like that was the case with the first batch of puzzles this year. (They were easier than the average puzzle in the Hunt, but I don’t think a student team with minimal puzzle experience would have much success on them.)

      Liked by 2 people

      • On the first point one way would be more emphasis on time release. If you haven’t already opened a round fully at some time just open the rest of it. You can keep it narrow only for the super teams.

        On the second point, you’re basically already saying this, but I think you can easily design opening rounds that release more puzzles at once while still making it easy to plan and prioritize. If you just drop a whole first round that shouldn’t cause problems, full round release is how lower tech hunts always worked! No one is overwhelmed by picking up an P and A magazine as a solo solver. You just need a simple clear structure with a well-defined goal.

        Liked by 1 person

    • Agreed, Friday is when we have the most people, the most awake people, the most enthusiastic people, and the most new people trying to see if they like this. Honestly, the small number of puzzles goes back to Normalville, maybe, when the map-based unlock started us off with only a few puzzles at first. I don’t see why they can’t dump 10 or so at 1 PM and get things started quickly.

      Liked by 1 person

      • The Matrix introduced map-based unlock, but in only in one very late round. Time Bandits were the people who really made map-based unlock a dominant method. I think Time Bandits had this problem of slow unlock, but I’m not sure, certainly we had that problem in badly in SPIES (exacerbated by how we dealt with events). IIRC, Normalville had map-based release for ordinary puzzles, but dropped the whole super-rounds as single drops, and their map was pretty well connected so it wasn’t too bottlenecked. Then things mostly switched over to points-based unlocks (we definitely did that in Video Games, but I think we stole it from something earlier, maybe Zyzzlvaria?), which is less prone to bottlenecks, but for some reason people have mostly maintained a very small number of opening puzzles. I really do think it’s largely driven by the primary goal of hunt design being to slow down the superteams, despite that goal coming at the expense of the fun of 90% of the teams.

        Liked by 2 people

      • Noah, of course you are right, though I can’t seem to reply directly to you. Even as a smaller team, I think we got frustrated when “fewer opening puzzles” became a thing, because we weren’t very good and we liked having a diversity of puzzles to try. Now, as a larger and better team, it often feels like there’s just not enough to do for the number of people we have available. (If I try to eat lunch or debug some global team problem, I miss the chance to do any work on the three open puzzles.) I guess this year the opening puzzles were hard enough to counteract that a bit. But two wrongs don’t make a right.

        Liked by 1 person

  6. Interesting to read all the comments from folks about their perspectives. This was my second hunt ever, first on-campus. I have a lot of feelings about that fact but none of them are particularly relevant here, so I’ll save the bulk of that for my own blog sometime.

    A lot of my puzzle friends are on other teams. I think one of the biggest bummers of the length of the hunt was having less time to hang out and see people I only see once or twice a year. I’ve seen a lot of sentiment about that from other folks; paraphrasing at least one person, it was less about winning and more about getting to hang out and do puzzles with friends all weekend. I think between this factor and a lot of the teardown and setup required with being off campus overnight, extra fudge factor and shorter hunts would be welcomed.

    A pain point that was more on us (Providence) than on teammate was the whole having to stop mid-hunt and hash out whether or not we actually wanted to win. (This is probably a good discussion every team should have right now when it’s not in the heat of the solve, btw.) This was apparently not a consideration for our team prior and the team leadership was somewhat alarmed to get ‘the call’ from teammate. We settled on going for it this time, but either we had the wrong mix of knowledge/experience, poor free solve choice, or just got unlucky, and only managed 6th place, which feels like an arbitrary number of no particular significance given all the effort to fast forward the hunt. Oh well.


    • I have strong feelings about this that I’ve voiced countless times elsewhere, but I firmly believe that anybody who is participating in Mystery Hunt and could win is either qualified to write (or too big). If you’re organized and puzzle-savvy enough to beat the other teams, you’ll be able to create a good Hunt if you’re willing to put the time in and take advice. If you’re not willing to put in the time, hunt with a less effective team. 🙂

      Liked by 1 person

      • Valid things to consider, and still an ongoing discussion within Providence.

        One other thing is I never felt confident enough to ask to do Mystery Hunt with anyone before, and I kinda just ended up on Providence because a recently-made friend invited me to join them. Now that I’m less green about the whole experience, I wouldn’t rule out switching to another team that is willing to write MH if they won, if Providence decides that they’re not inclined to be that kinda team.


  7. The mantra of “slow down the super teams” partly comes from an era where as soon as the coin was found, everyone would go home. When we wrote 20,000 puzzles in 2015, I remember feeling pressure to stretch out the time when the first team would finish, and that lead to the poor decision to have a long runaround. The “hunt ends when coin is found” norm weakened in the couple years after that and now I think most teams barely look up from their puzzle when the coin is found. So hopefully the norm of a scarcity of puzzles on Friday will also disappear soon!


Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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