(This is a recap/review of the 2018 MIT Mystery Hunt, which happened this month. Puzzles and solutions can currently be found here. This recap may contain spoilers, but I’ll try to avoid them where possible, especially for puzzles I recommend.)
Going into this weekend as a member of Setec Astronomy, I did not expect my team to win the Mystery Hunt. Some of my teammates were taking a more relaxed approach, although I pretty much only have one intensity setting at Hunt; for me, the most fun thing about a puzzlehunt of the size and complexity of Mystery Hunt is figuring out the most efficient way to get from Point A to Point C (for coin) through strategic puzzle-solving, meta-cracking, backsolving, and time management. Navigating the Hunt itself is the most interesting puzzle of all.
On Sunday, after grappling with the final Sci-Fi Island metapuzzle for about five or six hours (for those on the construction team who asked, we were NOT stalling; we were missing a key insight for more than half that time, and then once we implemented the right thing, it gave us garbage output due to a technical error we didn’t find for quite a while), it became apparent we were about to win the Hunt, and we had a conversation about that. My philosophy, which I acknowledged is not every participant’s philosophy, is that if you choose to be on a team strong enough to win the Hunt, you owe it to the community to write the Hunt if you’re the fastest team to reach a point where you can finish. Members of our team had lots of different perspectives and conclusions, but the vast majority agreed, for one reason or another, that we should finish what we started, and we surprisingly found ourselves in charge of next year’s Hunt. Again.
We learned a lot writing in 2017, and we have yet another data point in seeing this year’s Hunt and how it worked from our perspective (and I hope to hear how it went for other teams). There are some things we’ll try to do again, there are some things we’ll try to do differently, and there are some things that will be much easier having done them recently (I know TK is super-psyched that most of the Hunt website code already exists in a form he understands). We’ll probably lose some members, we might gain some members, and our Hunt may be very different than the one many of you just solved. But like every construction team, our goal will be to produce an event that everyone enjoys, and we’ll do our best to achieve that goal.
But enough about 2019, what about 2018? My favorite Hunt of all time, the 2011 Hunt (Video Games) felt in many ways like an improved form of the 2009 Hunt (Zyzzlvaria) taking a lot of the bolder innovations of that Hunt and presenting them in a more effective way. Similarly, I thought that this year’s Hunt felt very much like an improved version of the 2013 Hunt (coin heist); the rounds had complex structures, often with multiple tiers and/or multiple metapuzzles, and the metapuzzles themselves often involved elaborate procedures as opposed to a simple aha about the feeder answers. I was thinking this long before we opened the Sci-Fi Island (which was our fourth), so I already made a connection to the 2013 Hunt before another six-sided-cube-with-overlapping-metapuzzles appeared!
There were at least three big improvements over 2013, however: (1) Puzzle quality across the board was a lot higher. Puzzle concepts were bold and creative, and execution seemed clean. 2013’s Hunt only ended on Monday because the organizers handed out puzzle answers like candy; this year the first completion happened later than Death & Mayhem (Life & Order) expected, but it happened organically (by the end we could afford to “buy” two free answers, and we just bought one to make our last meta solve easier). (2) While the metapuzzles were often very difficult, they were reasonable. All four of the top-level metas were complex but none of them took hours of processing; if we had all of the answers and knew what to do, they would not be a slog. (3) I found the theme (at least the main theme) very engrossing; Inside Out is fantastic source material, and the production values in the kickoff and final runaround were ridiculously high.
If you can’t tell, I liked this Hunt a lot… there were a couple of two-hour periods where I felt useless and a little frustrated, but the Hunt never felt unfair, and most of the time it was a lot of fun. My biggest complaint was that the island themes felt pretty random and a bit disappointing. The Hacking round, which we opened first, seemed like a labor of love clearly designed by people familiar with MIT hacking culture, and the tier filled with physical puzzles was very cool. We opened Pokemon next, and the structure of that round was neat, but “Pokemon” and “hacking” don’t really seem like items in the same category. We could tell from the “preview text” that our third island was going to be Catan-themed, and ultimately I didn’t really buy it as “Games Island,” since it wasn’t really about general games. And as for the Sci-Fi Island, the cube structure’s been done, and a space fiction theme that unites lots of different existing properties is a great Hunt theme… which is why we used it in 2009. The point made at wrap-up that these were different chronological stages of Terry’s life made them feel more related, but I didn’t grasp that at all during the Hunt when it matters most. I was definitely expecting the islands to be either closer to the Inside Out plot or themed around other Pixar films, and I think either of these would have resulted in a more cohesive Hunt. This was a great Hunt, but the stitches in the overall structure were showing.
One intriguing decision by the organizers was to allow solving teams to choose the order of the round unlocks, which hasn’t been done since 2004. (For anybody who heard James mention the “Vatican Effect” at the wrapup and didn’t know what that means, in 2004 you could choose which round you’d unlock next, and the organizers thought solvers would likely open rounds in numerical order… but the highest-numbered and hardest round was visually closest to the opening round on the visual map, and so many teams chose that thinking they were supposed to.) This “choose-your-own-adventure” approach has been brought up in multiple years when I wrote, and I’ve always opposed it out of fear that solvers who choose the “wrong” order might get bottlenecked and get screwed out of the opportunity to win by random bad luck. It also gives the organizers less control over when solvers encounter key concepts… When we were dealing with the scavenger hunt, we wondered how frustrating it would be to open said scavenger hunt in one of the last unlocked rounds, since we had to send team members home to gather stuff. From talking to other teams at wrap-up, I can confirm it would be very frustrating.
This got discussed a little bit at wrap-up, but I was hoping for a lot more information. What order did you open the rounds and why? And do you think it helped or hindered your progress? I’m glad someone had the guts to try this in a year where I wouldn’t be responsible for the result, and I’d love to hear how it affected your experiences.
Coming soon: More posts on my personal Hunt experience with the metas and sleep (or lack thereof) and some notes about puzzles I particularly liked or wanted to throttle for one reason or another. If there are things you particularly liked or disliked this year that you hope next year’s team will preserve or change, chime in via the comments; I can’t guarantee any/all requests will be honored, but I’m pretty confident that next year’s constructors will at least hear about your feedback.
Puzzlvaria 
we opened sci-fi, hacking, games, pokemon (but the last two quite close together in time, iirc). i can confirm that we were a little dismayed that the scavenger hunt was in the last island we opened (on saturday afternoon), and we certainly had a team debate about whether to spend the buzzy bucks on it. but we were able to *very* quickly scrape up enough stuff to get it done by sunday night. (and hey, the mental break i took while going back home to retrieve scav hunt items might have been just what the doctor ordered to think of the key idea on the sci-fi meta-meta, which i’d been working on for pretty much all of saturday without success.) the plan had definitely been to spend the buzzy bucks on the *evolved* scav hunt, but then that turned into an arts & crafts/BSing activity that was actually super-enjoyable for everyone involved, rather than another scav hunt. so it all worked out.
i’m very glad, in retrospect, that we opened sci-fi first, because we needed every single answer to crack it; it took well over 24 hours from unlock to solve, even though for a big chunk of that, it was the only thing we were working on. (i still haven’t given deep analysis to how fatal it would have been to have one missing piece of information, but my gut says the answer is “very fatal”.)
LikeLike
Just one datapoint, but: I’m rather impressed that my team managed to solve the sci-fi metameta without one of the *metas* (and something like 2 other puzzles)
LikeLike
It turns out that the circuit was somehow pretty stable to even fairly large changes in the resistor values. We (on teammate) were able to toggle the resistor values for the two we didn’t know a little bit and managed to establish the currents to within plus or minus a couple amps. From there, we constructed a nutrimatic query which spit out the answer
LikeLike
The only free answer we used was on “Feeling Cross” in order to not have missing data on the Sci-Fi meta-meta. We had quite a bit of data on the answer from the two submetas it fed into, but our backsolving attempts were not fruitful, and I agreed that trying to do what we thought we needed to do with the meta-meta (which was correct) would have been massively difficult. Though I guess you could have narrowed down the remaining resistor based on possible flags and tried a finite number of possibilities; since we had a data entry error (one of our power sources was input backward) none of those possibilities would have given us an answer until we found our mistake.
LikeLike
FWIW, our testsolvers were able to solve this with 5/6 metas, 16/18 puzzles and had backsolved a few puzzles along the way. We thought that solving with 4/6 metas was actually possible although I’m not sure this happened during Hunt. I’m sorry if any of you found the experience too frustrating. It is certainly possible our testsolves led us astray.
LikeLike
Feeling Cross was my favorite puzzle in the entire hunt (that I solved). Y’all missed out.
LikeLike
One team did solve the Sci-Fi meta-meta during Hunt with only four of the sub-metas, though they had all eighteen of the individual puzzles.
LikeLike
interesting! for what it’s worth, i think it’s totally fine to have a meta-meta that requires 100% completion of the metas that feed into it, in the same way that it’s fine to require a team to complete every round to finish the hunt. i’m less sure about requiring every puzzle answer in a round to get the meta, but i can certainly understand that, or something close to it, as a deliberate design philosophy (“every puzzle matters”). it certainly requires the writing team to double down on their commitment to clean, solvable, not-broken puzzles, but i do feel that this hunt achieved that overall. and the sci-fi round was, ultimately, very satisfying.
LikeLike
Just one data point. I was on a medium non-competitive team. We opened SciFi round first, and it’s the only Island we recovered. We were able to solve it Saturday night with only 4/6 metas.
LikeLike
To add onto what joon said – we had a very specific reason why we went in that order. We went for Sci-Fi first because even though Palindrome split we were still a big team and the round mentioned that it was sprawling (or something like that). Hacking came next because we wanted to do the MIT specific round as soon as possible, but not first because we didn’t want to accidentally lock out remote people out from helping us by having no MIT-specific puzzles unlocked. Catan and Pokemon were then chosen to alternate MIT-specific stuff so that we kept up the interest for remotes. The only regret I had about this order was that the scavenger hunt came last, but there was no way to know about that. We still did it that night, mostly because I reminded people how badly we stalled out on the Red and White Knights meta in 2014 with only 2 puzzles open in that round.
LikeLike
teammate unlocked them in the order of Sci-Fi, Pokemon, Hacking (Saturday morning), Games (Sunday morning). We solved the metametas in this order as well, although I doubt we would’ve been able to finish Games had the puzzles not all unlocked Sunday afternoon.
This was democratically decided, but mostly motivated by perceived onsite requirements – we started with Sci-Fi and Pokemon for a good mix of onsite/offsite, and we wanted to time the “significant onsite presence round” for Saturday morning (which we thought was Hacking).
Regarding “choose your island”, I think it’s essentially fine, although I guess there’s a good point to be made about certain fixtures like scavenger hunts placed late and whatnot. (For fairness, put one scavenger hunt every round! /sarcasm) What puzzles you choose to focus solving efforts is a choice anyway and affects which future puzzles you unlock in certain round structures (especially map-based ones). So there is a certain bit of chance, although I suppose that on a puzzle-level (as opposed to round-level) randomness ends up being a bit more averaged out.
I personally felt that this year’s hunt was a bit too narrow – often one puzzle unlocked one puzzle, which is good for not making a hunt go as fast as 2017, but not exactly the most pleasant solving experience. (The metametas were hard enough that it was probably fine to be more open with puzzle unlocking, although maybe they wanted to reduce backsolving? I feel like we backsolved plenty anyway.) Probably most notable in the Hacking and Games rounds. It’s a bit ironic that in Games two of the three bottlenecks were due to onsite puzzles, since we believed that it wouldn’t require nearly as much on-site presence. From the flavor at least, I had hoped that the round structure of Games would be pretty open.
Also I suppose as a team who isn’t the biggest powerhouse (we are, effectively, a smaller offshoot from Galactic) brainpower was a somewhat pressing concern. And its acquisition can probably be best described as capricious, sometimes getting 70 for a solve, then 25 for the next. It was a little dismaying.
With that said unlock structure is probably my only even slightly substantial criticism of this hunt. The production value of this hunt was very high, and it was challenging and fulfilling to solve to completion.
LikeLike
Based on the announcement about team size, I figured there would be a smaller number of puzzles (there actually was, but I barely noticed it) and my wife suggested it might just be a low number of puzzles available at a time. Of those options, I definitely preferred the latter, and I think that’s the best way to allow low-to-midsize teams to have a shot over giant teams, although it appears that a lot of the giant-est teams split up in response.
We similarly had two times that I can think of where we felt very bottlenecked with only a few puzzles to work on, and they happened to line up with times we were short-staffed anyway, which was frustrating at the time but convenient on a big-picture basis.
LikeLike
Hello! Chiming in here from Super Team Awesome. We opened Hacking, then Pokemon, then Games, then Sci-Fi. We solved Pokemon first on Saturday afternoon (around the same time as Setec, according to the graphs shown during wrap-up), then were extremely stalled on metameta’s until late Sunday afternoon, when we solved the remaining 3 in a ~5 hour period. Note that we weren’t stuck on the metameta’s themselves, but just the island puzzles leading to them. Each of the metameta’s fell within ~5 hours of looking at it with all/almost all of the previous puzzles answers in hand, and hacking metameta was solved less than an hour after unlocking it.
Some context on our team – we’re a relatively new team (started in ~2007 from my understanding, though I’ve only been hunting since 2014, always with STA), and a relatively young team – almost everyone is still in their 20’s, with just a couple in their very early 30’s. We only have 3 or 4 members that have done 10 or more hunts. We’ve been steadily growing, gaining experience, and moving up the ranks – 16th, 15th, 14th, 10th, and now 7th this year for the years since I joined as a freshman. Last year was the first year we had the experience of finishing the hunt, which was incredibly momentous and satisfying, and we were happy to repeat and improve on that this year.
Like you, I generally enjoyed the metameta’s level of difficulty – not just waiting on an aha moment from the feeder answers, as you said, but also not being impossible. We got the scavenger early on, but with an official team size of 85, decided that it wouldn’t be worth it to try to coordinate collecting things for it – we really only have a core of ~40 people, with another ~30-45 showing up throughout the weekend, plus ~10 remote solvers. This year’s scavenger in particular was difficult for a younger team and for a team with only ~15 of the core members even living in Boston any more. In previous years, we’ve been able to exercise some creativity on various parts of the scavenger hunt to fill in the gaps, but this year’s specifications were less amenable to creative interpretations. We weren’t a huge fan of that, we always got some amusement out of the creativity involved with that. Tasks where team members had to perform various tricks or acts were particularly fun – in the 2014 hunt, for example, there was a call for someone that could solve a Rubik’s cube in under 60 seconds, which was one of my only significant contributions to the team that year.
Though we understood D&M’s desires in requiring every, or very nearly every puzzle to be solved in order to crack the metameta’s, that was one thing that got pretty frustrating when we got stuck on certain puzzles that were required to proceed. In hacking island, for example, we were stuck on the Deploy meta for probably 16 hours before eventually accumulating enough Buzzy Bucks to buy it (we also had to wait until after they opened meta’s for hints, we’d tried calling in hints for that before). It’s always frustrating to get stuck behind a wall of one or two puzzles.
In that same vein, the somewhat more limited release cycle of this year’s hunt led to some annoyances. At any given time, we had roughly 8-12 puzzles unlocked, but there were times when those puzzles were ~5 metas and ~5 puzzles which were required for the metas. Because this year’s hunt required every puzzle for most of the meta’s, that meant our actual attack surface was only ~5 puzzles. If we were stuck on 2 or 3 of them, that made things pretty unfun. However, that only really happened late at night when there were only ~10 Super Tired Asians Still Trying Answers late at night. On the other hand, limiting the attack surface let us concentrate efforts a bit more, since it was usually more clear what the puzzles in the critical path were. I just felt like the attack surface was just a bit too small – 12-15 puzzles at a time, with a similarly significant portion of them being meta’s would probably have been a good balance.
I personally liked the choose-your-own-adventure format, because it meant you couldn’t tell where you were relative to other teams. If errata were being called in on puzzles you hadn’t unlocked, it didn’t mean that you were falling behind, it just meant that some other team had chosen a different path. Seeing people around campus doing walkarounds also made relative ranking very difficult to guess. I’ll note that I think the choose-your-own-adventure also only worked so well because the 4 islands were very well balanced in terms of difficulty
I also loved the large number of interactions and general production value. The final walkaround and the core memory orbs were particularly stunning.
Interactive/game puzzles like TPMH and Under Control were similarly fun. We projected our own playing of TPMH in our room and most people dropped the puzzles they were working on to watch.
Overall, I thought the difficulty and writing of puzzles was excellent, possibly the best of the 5 hunts that I’ve experienced. I especially enjoyed the “evolving” puzzle mechanic of Pokemon, where we learned that D&M were “not fucking around.” There was definitely a sense of awe when we finished Shoal Patrol and saw Submarine Patrol, and a collective gasp when we finished 33 RPM and saw Mix Tape. Follow-ups that could be most efficiently tackled by people that had just solved the preliminary evolution were particularly satisfying.
Looking forward to next year’s hunt!
PS: thanks for the explanation of the “Vatican Effect” – I hadn’t heard that bit of hunt lore and didn’t experience it for myself.
LikeLike
Back when we were writing the first Setec hunts, there was an unofficial “80% rule,” stating that an ideal meta should be solvable from any eighty percent of the input answers, and not less. This is an impossible ideal, and in the era where Hunts often ran too long, it was better to err on the easier side. In 2017 we erred way too far on the easy side, with strong teams solving with less than 50% of the answers and then backsolving everything else, so it’s natural that Death would compensate by leaning on the side of requiring more puzzles. We did still do our share of short circuiting and backsolving, as I’ll go into later today when I post about metas.
LikeLike
We (Up Late) solved the Advertiser meta with one answer out of five. It was pretty obvious from the shell what was going on, and our second guess based on the two letters we had was right.
(Didn’t get much backsolving out of it though. When we were stuck on the final extraction of one of the Pokemon puzzles, I posted “It *might* be one of these four patterns”, which confirmed someone’s theory about what was really going on. Our most interesting backsolve was one of the Emotion puzzles that fed into two metas, I think Roadside America. We hadn’t solved either of its metas at that point, but seeing that it fed two highly-constrained metas let us get it from the constraints as soon as it unlocked.)
LikeLike
Hello Dan,
Just to be clear, there was no design goal that metas had to require more than 80% of the answers. I would have to double check, but I believe all metas made it through testing at roughly that threshold. Some were solved in testing with much lower numbers.
LikeLike
IIRC, there was a team that solved the Sadness meta with zero feeder puzzles. Punny minds think alike!
LikeLike
Absolutely, I think meta’s solvable with only 20% of the answers lead teams to skip a lot of awesome puzzles. 80% seems like a good metric. Our team was just particularly frustrated that we couldn’t even approach the hacking metameta until we’d solved Deploy. We’d already worked out the likely mechanic, and would have preferred attempting the solution with perhaps only a partial list of transitions, or a partial floor plan, rather than being forced to solve Deploy to get the final floor plan at all. After eventually buying the answer to Deploy, we solved the metameta pretty quickly, having already worked out the mechanic.
LikeLike
The 2012 Producers hunt aimed for “# puzzles minus two” I think for our metas (Andrew can correct me if I’m wrong) but I think our test solvers were not as expert in solving-metas-with-incomplete-info as many of the actual hunting teams (a known weakness of Codex) and so we inadvertently found our metas solved much quicker than we expected, with many fewer puzzles solved, leading to a much faster-than-expected hunt. That was partially by design, since we were paranoid about stuck puzzles ruining the hunt, not entirely confident in our first-time-writing team’s ability to write 100% not-stuck puzzles, and the “buzzy bucks” mechanism was relatively new to hunt construction and not fully appreciated.
In a hunt constructed in 2018, I think folks are pretty confident using mechanisms like “buzzy bucks” to solve the stuck-or-unsolvable-puzzle problem, and metas can be safely written to the “# of puzzles minus one” standard — especially if you’re a team like Setec which is confident in their puzzle editors ability to ensure that every puzzle is solvable.
LikeLike
For non-pure metas there’s an obvious, 100% guaranteed solution: test-solve your meta with N (~80%, or whatever your goal is) answers, and then leave it locked until N puzzles have been solved. This also turns N into a knob that can be tweaked in real time if the Hunt is running long. Doing this would be breaking with Mystery Hunt tradition, though. I think part of the reason is that teams LIKE the chance to “break” the Hunt in a small way with some clever backsolving (as you said, it’s part of finding the most efficient way to get from A to C), and closing that avenue off with an artificial wall feels like an admission of defeat on the part of the constructors.
With such high stakes, though (great puzzles people have been working on all year not getting seen), it’s probably still worth considering, pride be damned. 🙂
LikeLike
We unlocked the rounds in the order of Games, SF, Hacking, Pokemon. I agree with you that it’s not a great structure because of the loss of control. And I’ll add something else: in this hunt, it led to a bottleneck. In the Hacking round, we got the Scout meta at 9 in the morning my time on Sunday, which is 3 in the morning Boston time. This led us to the Build subround, with 3 puzzles all requiring physical construction, and exactly 2 awake on-site solvers.
Bottom line: 40 hours into the hunt I was confident we’d finish, but at the 50-hour mark we’d barely made any additional progress and by the 60-hour mark nearly everyone had just given up (we’d gotten one more meta, Build). At the 20-hour mark I thought it was better than 2015; after the end, one of my teammates sent an email rating it a B or B+ compared with 2015 and 2017’s A and 2016’s C-. I tweeted that I still think it’s the second best hunt ever but I’m no longer sure it’s better than the two of yours that I’ve played. I had more fun but I think I just got more useful this time, maybe because of the exact nature of the puzzles, or maybe because of which hours I was awake, or maybe because I finally made significant contribution to a meta (both Scout and Build).
Last year you talked a lot about optimizing the hunt differently for teams of various sizes and power, but this experience is making me think about optimizing for on-site versus remote teams. Runarounds and physical puzzles optimize for on-site and scavenger hunts can go either way. A few puzzles optimize for remote, like the one in 2012 asking for photographs from places with distinct integer-mile distances from MIT. But mostly, what optimizes for remote solving is an unlock structure that reduces HQ’s control. If you’re hitting the hunters with a bottleneck and you don’t have control over when, then remote teams with good timezone coverage have an advantage. Unfortunately, giving us choice of what order to do things in and then adding on-site puzzles made this optimization really weird, and we got burned with when the Build round got unlocked.
Unlock rates are important. I know we fought over this last year, over my claim that 1:1 unlock rates are bad, but this really weakened the Hacking round this year. In contrast, the first round with the emotions was very good, and the SF round had a very good unlock structure as well – we didn’t get frustrated as much. In Hacking, we were frustrated. Then when you guys found the coin all the puzzles suddenly got unlocked, so at the end we got hit with the worst of both worlds: the frustration of low unlock rates, and the confusion of seeing too many puzzles at once.
But the individual puzzles were really good (unlike in 2013). I gather everyone who was on-site loved Twitch Plays. From far away I had a lot of fun doing Self-Referential Mania (which is a lot like 2015’s Coral Reef meta except not meta-size) and Murder at the Asylum (where the solution made me feel stupid but I still liked the process).
One question: did you get It’s Not Normal? The crossword answers to solution extraction seems impossibly difficult to me and I’m not sure I understand even after reading it. Curious if you guys managed to forward-solve it.
LikeLike
I was wandering by the Setec solvers as they finished It’s Not Normal. It was definitely forward-solved, in much the way that is described in the Solution page.
LikeLike
I’ve mentioned this before, but I see no reason that remote teams should be a thing in Mystery Hunt; Hunt is difficult enough to run for the people who show up on-site, and if you create an expectation that people/teams all over the word will be able to fully participate, you’re spreading the volunteers running the Hunt way too thin. If a team wants to have some remote solvers, it’s up to that team to figure out how to utilize them most effectively and make sure they have a good time; I don’t think the Hunt itself should be written with that in mind.
We did have the experience in 2015 of opening the treasure chest (more physical puzzles) when everyone was asleep, which was a giant bottleneck; based on the description and that memory, we chose to open Hacking first specifically to avoid that bottleneck. If you only had two people awake in HQ at a particular time, I recommend staggering sleep schedules. For years now, I’ve been intentionally going to sleep early (before midnight) so that I can return to HQ as numbers are waning.
I didn’t work on It’s Not Normal, but I know it was solved forward by our team, because I heard the solvers reading out the two clue phrases. (It was also after we’d solved the meta, because I’d typed out that we expected an answer of the form ???MB?????, though we didn’t guess the answer from that.) I’m not sure who on the team recognized the Stata Center structure, but our HQ is in Stata and at least one team member works there.
LikeLike
I’m from Alon’s team (Hi Alon!)
(as is human nature, I’m realizing that a lot of stuff here is rather negative. I loved the hunt! It was great. Don’t let anything I say below make you think I disliked it. I’m very grateful to everybody who puts in an insane amount of time to make this happen)
Just want to make sure that two distinct things aren’t mushed together — unlock rates and offsite solvers.
I’m fine with hunts not being optimized for offsite solvers. It contributed to bottlenecks in our particular case, but from my (onsite) perspective this wasn’t a huge problem.
What I, and a lot of us onsite, were frustrated by was the unlock rate. We rarely had more than a small handful of puzzles open and when we reached a point where roadblocked hard on a few puzzles, we were just completely stuck. The biggest example was Catan island, which we opened first, but didn’t make any meaningful progress on until after we completely finished SciFi island (opened 2nd). This was due to us roadblocking hard on Sports Radio (we had another significant frustration with this one, see below) and the really hard cryptic. We solved three puzzles on the left side of the board, but there was no way to progress into the middle of the island without solving one of those two puzzles, and we were just completely stumped. Somebody offsite solved the cryptic eventually, but well into Saturday, and we had to buy another answer just to make progress through the island. We solved that metameta something like 42 hours after opening the island.
We were also not fans of the rates that the islands unlocked, and especially the cryptic clues about the islands. We chose Catan and SciFi first because we opened them both Friday evening and the other two islands clued physical presence being necessary, and we didn’t want to be doing puzzles that might involve finding physical locations overnight. We figured we’d get the next island Saturday and we could do onsites then. Unfortunately, it was pretty late Saturday by the time Hacking (3rd) opened and we would have run into the same problem anyway. But the most frustrating thing was that one of the initial puzzles (Sports Radio) in Catan island was a runaround! It actually took us around 24 hours to send somebody to the physical location we needed to go to, because we assumed that we wouldn’t need to do that on an island that didn’t have physical presence being necessary called out in its description.
We unlocked Pokemon Sunday afternoon, and I personally was really disappointed because it looked like there were a lot of really fun puzzles that we didn’t really get to approach as much as I’d like. We did TPMH with ~10 people and it was fun, but if we’d seen it anytime Friday or Saturday we would have done it with the whole team and it would have been the highlight of the hunt.
I would have really, really liked a system where solving a metameta automatically opened a new island, and solving a submeta (maybe every other meta in SciFi) automatically opened some puzzle in another meta that wasn’t opened yet (like a Catan hex distant from the ones you’d already unlocked).
Unlock rates are tough. I appreciate that they have widely varying effects on teams of different strengths and sizes, and I appreciate that this structure probably did a lot of good for the top teams, and making sure that the hunt didn’t end Saturday morning by bottlenecking top teams a bit. Unfortunately, this meant that we were also very severely bottlenecked, and we didn’t have the tools to become un-bottlenecked the way a lot of much stronger teams can, leading to a somewhat frustrating experience.
I’d be really interested to see a writing team experiment with a system where teams’ unlock rates are dynamically changed so that trailing teams unlock puzzles faster than leading teams — those are often the teams that have weaker solvers and having a wide variety of puzzles to work on does a ton to keep them involved.
LikeLike
DanielS: Playing devil’s advocate for a moment:
“I’d be really interested to see a writing team experiment with a system where teams’ unlock rates are dynamically changed so that trailing teams unlock puzzles faster than leading teams — those are often the teams that have weaker solvers and having a wide variety of puzzles to work on does a ton to keep them involved.”
I’m wary of any system that “rewards” teams for being behind. In a sense, that penalizes teams for submitting correct answers, and if a team knows about that in advance, they might even feel incentivized to jot down an answer to a puzzle they’ve solved without actually calling it in.
Let’s call the trait of wideness vs. narrowness (essentially the max puzzles available at one time) diameter. What I’d argue one wants is for the diameter and the unlock rate (those are similar but different; for example, a 1-to-1 release where all the puzzles are supereasy has very high unlock rate and very low diameter) to increase more quickly over time for earlier puzzles than later puzzles. That way teams who are still on those early rounds get naturally caught up without the teams ahead being pushed up at the same rate. In 2009 (Zyzzlvaria) we tried to accomplish this by opening rounds based on points and making both the puzzle rewards and open costs much higher further down the line. That way, we could give everyone, say, 200 dollarbucks, and that might be equivalent to either an entire round in Inner Zyzzlvaria or a puzzle or two in Outer Zyzzlvaria. There are ways to make a globally consistent aid be more useful to teams in one situation than to teams in another.
Last year, we totally blew it on diameter, because we didn’t realize how easy our metapuzzles were, and our cap on diameter was organically based on the assumption that teams would get stuck on metapuzzles (which happens in virtually every Hunt). In 2017, they didn’t, and strong teams got a firehose of puzzles, which was not our intention.
LikeLiked by 1 person
Yeah, I appreciate that a dynamic unlock rate can be a scary thing. I hadn’t considered the possibility of a team solving a puzzle but not calling it in.
That said, from my PoV on a mid-strength team, unlock rate has caused a good share of frustrations in the past and especially this year. I don’t think it needs to be a huge tweak. Even if we’d had a single extra puzzle unlock in Catan, that would have been a huge boon for us. It could also just be time based rather than dynamic, eg every 6 hours after opening an island, you get another puzzle in that island irrespective of how many puzzles are solved/unsolved on that island. Or, every 18 hours you get a new island unlock, irrespective of how many you’ve opened or how close you are to opening another.
LikeLike
As the person who works in Stata whom Dan referenced, I thought of Stata the moment I saw “Metal” and “Brick.” And the fact that there were never more than nine across clues and Stata has nine floors was also suggestive. I eagerly went to the Slack channel for that puzzle to report my discovery, but found people who had already made the connection were working on it. I’m not sure what tipped them off.
It took us a while after that to actually fit the grids on the building. Part of the reason why is that it was raining heavily at the time and people didn’t want to go outside!
LikeLiked by 1 person
I actually like the remote aspect of the hunt — the idea that on this particular weekend in January we have puzzlers all over the world holing up in rooms and working on this in-some-ways-very-MIT-and-very-Boston Hunt. MIT alumni spread out all over the world after graduation, and although it’s nice to bring them all back home for a weekend, it’s also great to have an “MIT Hunt” that you can participate in wherever you are. So I don’t see “remote solvers” and “an MIT hunt w/ strong physical-presence components” as being mutually exclusive. And, since shutting down HQ between midnight and 8am isn’t really going to happen, having a strong remote component lets the local folks get a bit of sleep from time to time w/o guilt.
I think this year’s hunt had a nice balance, actually. There were lots of physical puzzles and high production values, a lots of MIT-specific content even for non-physical puzzles, but there was excellent streaming of events for the benefits of remotes (and those in HQ) and even many of the physical puzzles could be scanned or digitized once some initial work was done, so the solving could be shared with remote solvers. I’m not very sympathic when people complain “I worked on this remotely for hours without realizing you had to go to the physical location” — that’s part of the puzzle, and part of the essential “MIT hunt” quality. But I’m also happy about puzzles like the “places around the world” puzzle that acknowledge that hunters are spread throughout the world, and even those all-in-person teams @ MIT have friends and family members all over the world.
LikeLike
Making unlock costs proportional to the unlock progress in each round (similar to Zyzzlvaria) seems fair. It is also similar to how D&M awarded more Buzzy Bucks solving later puzzles in each round. This could extend the Choose Your Own Adventure concept by allowing teams to decide exactly which puzzle to spend to unlock next. This could introduce an element of team strategy – Should we hoard points to progress deeper in a round or have more open puzzles across rounds? Should we spend big to unlock that last puzzle in the round or try to hack out the meta without that answer? This might be too much decision-making though, or overly reward teams who solved a meta early.
I would prefer the scavenger hunt to be an event rather than a puzzle. Since the goal is to maximize the fun and puzzle experience for teams, one possibility is to release the scavenger hunt like an event at a specific and convenient time instead (say Sat 7am). That way, there is certainty teams do not need to wait for members to wake up or run back home to fetch items, and could spend more time solving instead. It could potentially cause a bottleneck in the verification process for Hunt Control, but having a queue timing for a “solved” event might be better than waiting to solve the scavenger hunt puzzle and being potentially bottlenecked on the unlock.
LikeLike
I solved It’s Not Normal on Left Out. After solving the grids we pretty much were sure it was the Stata Center, but since those who solved it were remote, we didn’t know if the puzzle required someone actually going to Stata Center and looking around. A couple of teammates braved the cold and didn’t see anything. The puzzle languished overnight, until we finally decided to pay for our very first Y/N hint of the hunt: “Is there any data at the actual Stata building, that we cannot get through on-line searches, that will help us?” Surprisingly, instead of Yes or No, we got the response “WE CANNOT CLEARLY ANSWER THIS QUESTION. HINT COST REFUNDED.”
We surmised from this answer that this probably meant that they didn’t deliberately plant any information at Stata Center, but they weren’t sure if we were able to see all the windows from online photographs. It turns out, yes, you can, especially with the help of Google Maps, and we eventually solved it in the expected way.
The irony is that none of us who worked on “It’s Not Normal” had solved “It’s Not Easy”, and knowing how that one worked would’ve helped a LOT — and our “It’s Not Easy” slack channel log, which had got nuked when “It’s Not Normal” showed up, as both of them hashed to “puz-pok-mon-it-s-not”.
LikeLike
A core part of Central Services hunting philosophy is having near-zero remote solvers. So the integer-mile requirement was tricky for us and didn’t feel particularly fun. I’d imagine our average team age is one of the older ones (not counting members incapable of small talk i.e. 4 month olds) so we sometimes suffer from not being active students, but that feels like a more than reasonable constraint. We’re damn lucky MIT allows the MIT community to participate, let alone give us classroom space.
We’re never going to win (God willing) and I’m disappointed we didn’t finish this year, but I had fun. I spent a large chunk of my Hunt on the scavenger hunt(s) and don’t feel I have all that much to show for it, which is sad, but it left me unaffected by the low number of puzzles available at a time. We’re a medium to large team (50-60 members, all on campus as noted) so having fewer puzzles open was a bit of a new phenomenon to us. Much of the Hunt felt like Friday afternnoon usually does – 8-10 people per puzzle having a good time.
Each puzzle still seemed to hold a constant amount of fun – but twice as many people working on it halved the fun. I mentioned this to someone on the solving team and he said that was what much of his hunting experience was like on a larger team. Maybe this was just our taste of being a bigly team.
LikeLike
…These thoughts are those of the commenter alone and do not represent the opinions of Central Services. To obtain the opinions of Central Services, please go through the proper channels and submit a form 27b-6 “Turning On The Faucet Subsection Opinions, Editorials, and Rants” to Central Services and be ready to be transported to Information Services for an answer promptly within the next two to six weeks. Thank you for your interest in Central Services, this comment may be monitored for quality assurance.
LikeLike
I’d like to talk to you about ducts.
LikeLiked by 1 person
Mer, I infer from Dan wanting to talk to you about ducts that he has some sort of duct conundrum.
LikeLiked by 2 people
Small Llama Malls member here. It’s Not Normal was definitely forward solved on our end, but was our second-to-last solve of the Hunt because no one really wanted to do what the puzzle was asking for us to do (and by the time two of use decided to actually do it, it was already getting dark.)
Extraction-wise, once we started doing what the puzzle was asking of us, the extraction wasn’t difficult. By then though, our team had dispersed and I was working alone in my hotel room at The Kendall.
LikeLike
Hi there, thanks so much for your recap. I’m chiming in as someone from a casual team of ~25 solvers.
We opened up two islands: Games, then Hacking (both earned). We went with whichever one seemed like it’d be the most fun to us. So “Games” was an attractive name to us, but we were dismayed to find the three opening puzzles there too difficult for us to solve. So then we had to solve more puzzles from the Control Room to earn more Buzzy Bucks to unlock another island. We chose “Hacking” next, because the flavor text suggested we’d be going out and about campus and that sounded like fun. Though we were able to solve a few puzzles on Hacking Island, we never did encounter a puzzle from that island that led us to go out and about.
So the Buzzy Bucks system felt opaque and expensive – it wasn’t clear that we’d be purchasing something that’d help us unlock puzzles we wanted to work on. Later on, when islands were opened up to all teams, we bought the answer to one of Pokemon puzzles in hopes of unlocking something more our speed. We chose the scavenger hunt one because “Hey, they wouldn’t have more than one scavenger hunt, right?” Oops. Now we’re out of points.
This is the first hunt (out of six) in which I disengaged for several hours due to a lack of puzzles on which I felt I could make any headway. I know other team members felt listless as well for a time. I felt myself longing for a School of Fishes Island/”kiddie pool” type area to wade around in. Or at least an easier and/or clearer way of unlocking puzzles (e.g., I think the D&D year was successful at this).
That said, I don’t mean to sound ungrateful for all the hard work L&O put into this event. I think L&O executed their vision and principles brilliantly, and I could feel it from Kick-Off to Wrap-Up. Even if there wasn’t some perfect number of easier puzzles for me, all of their puzzles were clean and fair, spanned a diverse set of genres, and I thought it was nice that many of them were conducive to teamwork due to the gradual a-ha moments needed. Every year I am in awe of the generosity and creativity of the hunt organizers, and so thanks to Life and Order, and thanks in advance to Setec Astronomy.
LikeLike
Angel,
I’m sorry you didn’t find our hunt as enjoyable as past years. I’m even more sorry that we apparently missed your team on visits we made to try ensure teams were having fun. Thank you for posting your feedback here so that future teams can learn from what we have done.
Henry, L&O
LikeLike
I’m really, really happy you liked the hunt! Co-author of Fowlty Towers here — just wanted to issue an apology about the inclusion of ducks in the DK. We found your blog post where you asked DK writers to stay away from ducks… but by then we had already ordered ducks that we hoped to include in the kits, and it was too late to take it back. They didn’t arrive in time for hunt, so if you’re ever annoyed about this specific thing, please remember that our uppance is coming, in the form of 800 tiny resin ducks from China.
LikeLike
Honestly, I think the “don’t use ducks” part of my konundrum guide is outdated… at this point the theme comes up a lot and I consider it an homage (particularly within the Mystery Hunt), so I’m not offended as long as it’s not called “The Duck Konundrum.” (So perhaps I’ve given up my territorialism about “ducks” and am doubling down on my claim to “the”… which is consistent with my MIT username, I suppose.) So no apology necessary.
I actually backsolved Fowlty Towers… except I didn’t, but we were told our incorrect answer (BRAZING FOIL) was correct, and they couldn’t take back the brainpower, so HQ awkwardly decided to just give us credit and show us the right answer. So I apologize for trying to skip your konundrum, and for succeeding despite failing. I looked at the solution, and the configuration at the end is nifty.
Also, sorry I didn’t say hi at wrap-up; I saw you on stage and was 99% sure you were the same Anisa who was in Sweeney, but I have a paralyzing fear of misidentifying people. Thanks for the great Hunt!
LikeLike
Oh good! The ducks are definitely an homage. And yes, Sweeney Todd! Good times.
LikeLike
Wow, they really had their wires crossed on that puzzle solution. We (Leftout) were told that BRAZENLY EVIL was incorrect, and we then spent 40 more minutes rechecking our duck towers and calling in other answers (BRAZED EVIL? BRAZEN ELVIS?) before GC called to sheepishly (duckishly?) tell us we had been right all along.
LikeLike
I would help fall on that duck bomb.
LikeLike
Central Services: we unlocked Hacking and Sci-Fi together at 6.20 p.m. Friday, almost at the same time as completing the Brainstorm; the first few hours had felt really, really productive. We picked explicitly things that commented about being physically present. And then it got harder :). There was a down patch – there always is – but through at least 3 p.m. Sat everything still seemed to be ticking along; I especially enjoyed the Asylum Murder logic puzzle as one of the very few non-easy-round puzzles that I led a solve on and feel confident I could have solved solo in only slightly more time :). But it wasn’t until 2.15 on Sat that we unlocked the third island (Pokemon); don’t know why we picked that one, but somewhere between about 9 and 3 on Sat my sense went from ‘we might actually finish’ to ‘holy cow we may not even unlock the 4th island.’ As it was, that 4th got unlocked at 6 a.m. Sunday. We (really ‘they’ as I was gone by then) recovered the 2nd and 3rd core memories at 7 and 9 a.m. Monday.
My only real gripe on organisation was lack of clarity/information about when HQ would close (which was stated several times to be 6 p.m. Sunday, hence in part my arranging to leave back to NY Sunday afternoon) and the late announcement about being able to experience the finale (which I would definitely have planned to stay in town for); similarly, I would have loved to buy a shirt and bid on a souvenir, but all of that was announced too late to be possible. And as other have noted: these are criticisms in the context of a really well-written hunt that clearly represents a massive effort for which I and the team are hugely grateful.
On the actual Q posed: I don’t think that island order affected us much, but I was really really surprised to have the island flavourtext not then reflected in the first few puzzles (e.g. initial Hacking Island puzzles could have been solved remotely). For something that was driven by the very little information given, it seemed an odd lack of follow-through. A a team of 30ish-at-any-time solvers, there were definitely bottlenecks, but none truly hideous. And the *many* points of humour were very welcome – I thought that worked really well.
LikeLike
Belated addendum from another Central Services solver:
I was on the night shifts, and overnight Friday to Saturday morning did feel like a truly hideous bottleneck to me. We unlocked our last Sci-Fi puzzle at 4:30am (Bloodroots, which went straight in the “huh?” pile until much later when Francis happened to remember having noticed a magazine lying around campus with triangles on the cover) and our last Scout puzzle at 7:30am (the aforementioned Murder at the Asylum logic puzzle that Ewan worked on for 5 hours). This meant that between 7:30am and 2:15pm we had zero unlocks (aside from the events at 9 and 2). When I went home at 10am I remember thinking that it felt like an “eat your vegetables” hunt: stingy unlocks and few short-circuit/backsolve opportunities, so you really just had to sit and solve almost every damn puzzle. By 2pm our only open puzzles were: two stragglers from Emotions (Nobody Likes Sad Songs and Jeopardy!); 10,000 Puzzle Tesseract and the Scout meta; and Bloodroots, Worldwide Contacts, and four of the Sci-Fi metas. Fortunately at 2:15pm we solved both Scout and Novaldex, which unlocked the Build round and gave us enough Brainpower to open Pokemon Island, and we were off to the races again. (Sad Songs and Tesseract were eventually backsolved, and Jeopardy! finally fell at 5am Monday.)
In retrospect, the worst island order would have been Hacking and Games first, since they both had 1:1 unlocks. We were somewhat fortunate to get Games last, because when the coin was found on Sunday afternoon, we got a flood of Games puzzles that we hadn’t unlocked yet, so we had plenty to work on Sunday night/Monday morning. And given that it took us 60 hours to trudge through all of Hacking, if it hadn’t been one of the first islands then we wouldn’t have had a prayer of finishing it.
LikeLike
It seems to me that Hunt could benefit from taking a page from the NPL’s annual extravaganzas and having running and walking teams. It seems to me that there is not going to be any way to offer puzzles to teams in such a way that small casual teams will always be able to have a reasonable number of fun and interesting puzzles to work on, while at the same time guaranteeing that large competitive teams will not be flying through puzzles.
A system whereby the walking teams get to unlock puzzles far faster than running teams at the expense of never unlocking the meta puzzles would give casual teams lots of puzzles to pick and choose from but would guarantee that none of them were actually in the race for the coin.
LikeLike
Given that I already think the runners vs. strollers dynamic at the NPL convention is horrendously broken (partially because solvers are not great at self-identifying their solving style), there’s no way I would ever support implementing it for Mystery Hunt. It works well for afternoon events like BAPHL and DASH where teams are small and self-selected, and the organizers have time to create additional material for the casual teams… Given the size, scope, and interactivity of Mystery Hunt, it’s much more difficult to create two standards of unlocking, and it’s an unreasonable demand on the constructors.
(Also, there’s previous feedback that indicates that some small teams don’t always *want* “lots of puzzles to pick and choose from”… Small teams have complained in some Hunts that they were inundated by a puzzlelanche and couldn’t process what was going on.)
LikeLike
I think by this point it’s well-established that the Mystery Hunt analog to the runners-vs-strollers is the round 1 mini-hunt. That is, round 1 is a self-contained hunt which is solvable over the course of the weekend by a small team (and by Friday night for a competitive team), but has as much “good stuff” in it as possible, including a final runaround and a sort of coin. That was originally an experimental feature of the hunt, but has been repeated and refined enough to be canonical now.
LikeLiked by 2 people
The Microsoft Puzzlehunt (which is similarly structured on a smaller scale, with a hard cap of 12 solvers per team and runs in ~30 hours) tried this several years ago: teams could self-select as “competitive” and show up on the leaderboard, or “recreational” and have access to any hints/assistance they wanted on anything.
While I wasn’t there, what I’ve heard from people who were was that it worked terribly, especially for mid-tier teams who weren’t strong enough to get through the event on “competitive”, but didn’t want to “give up” by switching divisions. (It didn’t help that this was before people had realized absolutely gating most of the hunt behind solving the first-stage meta was a bad idea.)
LikeLike
Holiday Hallmark Classics unlocked islands in left-to-right order (Pokemon, Games, Sci-Fi, Hacking). One thing I think worked out in our favor is that we unlocked Hacking last, and we unlocked the wave of physical puzzles in Build at 3 AM Sunday. We only had a dozen or so people in HQ, so we didn’t have too many people trying to crowd around a single physical puzzle.
We were hypothesizing that if we had opened the Hacking Island first, on Friday afternoon, that we would have had a lot less fun. We didn’t solve the Scout submeta until we had solved most of the Scout puzzles, which exacerbated the bottleneck effect. Not sure if other teams experienced that.
LikeLike
Codec Vorbis opened sci-fi island first, then hacking next. We really wanted to open hacking while we had lots of physical presence in HQ, but compromised on sci-fi first to ensure that our remote hunters had plenty of stuff to keep them busy, especially over night.
We were a little surprised that hacking started with a full tier of non-physical puzzles; we unlocked the physical puzzles overnight Friday-Saturday, and by that time the physical presence in HQ had thinned out a lot. But on the bright side, that meant that I could solve Marked Deck 90% by myself. 😉
We unlocked Games last, and after seeing how narrow the diameter was (only three puzzles open at once), I was extremely concerned that Vorbis was going to get stuck on a puzzle, leaving us with only two puzzles for our entire team to stare at. I strongly advocated for hoarding Buzzy Bucks to use against this scenario.
It would have been more pleasant if the islands were unlocked on a regular cadence, instead of effectively unlocking two at once, and then 12 hours later unlocking another two at once. I suspect that the original pre-modification brainpower release schedule was better.
In retrospect. we would probably have benefited from unlocking Games and Sci-Fi as our first two, to parallel a wide-fan-out with a narrow-fan-out round. That way the risk of getting stuck puzzles in Games would have been mitigated. But (as I described above) I don’t think anyone on our team feels this was “unfair”; we feel like we may have made the wrong choice, but that responsibility for the decision was on us.
But the Islands were pretty balanced overall, and pretty much always had at least two open at once, so the disadvantage of choosing “poorly” was mitigated. If we had only one island open, and it was Games, we would probably be hurting.
LikeLike
Up Late unlocked in the same order with pretty much the same reasoning.
The one time when our remote crew (which I’m on) was particularly frustrated about unlocked count was Saturday evening, where we’d pretty much finished Sci-Fi, was in the middle of Build, and just starting Pokemon. It worked out that there was exactly one non-meta non-MIT puzzle open for a few hours, and it was Submarine Patrol. But that ended when our Under Control slot opened, which very much improved mood. (Especially since the person doing it was from Seattle and flew out to MIT, so most of the remote solvers knew him, including his wife who’d brought their kids 🙂
LikeLike
II&F opened Hack, then Games. As those were the two with very constrained unlock rates, we had a dearth of puzzles on Friday. Overall we liked the hunt a lot, but that choiice both negatively affected our overall performance and our fun-levels for that portion of the weekend. Once we got Sci Fi (and Pokemon) open we had a lot of puzzles open that were more fun for folks to work on.
LikeLike
(tried to post this before, but somehow messed it up. apologies if it comes through twice.)
I hunted with SW:TNG-InfinityWar, which is a couple of former Setec members & one from Elusion Nights (I never spell that team correctly), *and* our teenage and younger kids. Our aim is to have fun, solve a few puzzles, do some runarounds, and indoctrinate the next generation of hunters. So my comments are very much from the point of view of a small team with mostly inexperienced hunters (and those of us that are experienced are not top solvers).
This hunt was okay for those purposes. It definitely could have been better, I think.
Several of the first round puzzles had more than one ah-ha or were too obscure for a small team to make progress on (we made only the basic progress on Warm and Fuzzy, AKA, & Freak-Out), or just too difficult for a team of mostly beginners (Clueless seemed way too much as a first round puzzle for inexperienced teams). We spent a lot of Friday being stuck, or with not enough to keep even our small team busy and not stepping on each others’ toes. If we’d been an undergrad team staying up all night, I feel this would have been very frustrating. As it was, we had a cutoff of around 8pm and so went to sleep and when we woke up there were more puzzles available.
We ended up solving Let’s Get Ready to Jumble entirely based on the visual clues (small team = no one knows WWE wrestlers, so couldn’t even start on the jumbles). Yay us, but the addition of “now go identify this wrestler from a freeze frame image” (after we’d decoded the clue phrase and submitted our video) just seemed a bit annoying, vs. just giving us the answer in exchange for the video.
Our team got to try one of the runarounds, but only because Setec loaned us their props & instructions (with HQ’s blessing). The younger crew are big fans of runaround (and did this one on their own with a non-helping adult chaperoning). If we hadn’t been right next to Setec’s HQ and friends with them, there would have been a big hole in the that contingent’s Hunt experience.
On the plus side, two of the teens solved The Lurking Horror all on their own, and were rightly proud, and another couple did most of the work on Bark Ode. Yeah But It Didn’t Work and In Memoriam were great for the whole team to contribute early (even down to the 7 year old on our team, who did the connect the dots).
We saw a lot of puzzles after they all opened that we would have loved to work on, but by that point we were flooded and didn’t get a chance. I worked on 33 RPM because it was such a nice idea (and so few lines of python), but ended up giving up because the clue phrase for the second part was too obscure even with the clue phrase (and a hint about keeping “revolution”) and the hunt ending. I forget the other puzzles we worked on, but there were more (and several that people enjoyed, despite my grumpishness)..
tl;dr – It seems like there were a bunch of puzzles that would have made great first-round puzzles, particularly for small & inexperienced teams, but they were hidden behind other parts of the hunt and that makes me a bit sad. We only got to do a runaround because Setec loaned us theirs.
(other tl;dr – this was one of my favorite drums to beat when I constructed with Setec. I am a grump, but a consistent one.)
LikeLike
I was one of the puzzle-on-call folks at Cranium Command, assigned to the morning shift. One of our tasks was identifying teams that were not having fun and speaking with them over the phone or visiting their HQ; evidently we failed in your case and I apologize for that.
I am sorry that your experience with our hunt was not up to your standards. Thank you for sharing your experience so that these mistakes can be avoided in the future.
Kevin
LikeLike
It’s not that we didn’t have fun. Like I said, this year was fine for our purposes. We solved several puzzles, and made progress on several others in a way that was rewarding. But looking at the puzzles that eventually got opened, we could have solved more, and there are a bunch I would have moved into the first round and some I would have moved to later (if possible).
Saturday was great, from our perspective. We consistently worked on and solved puzzles throughout the day, though about half our of our work ended up hitting some sort of blocker, usually related to not knowing the final step. So we never really had a time when we might complain to HQ.
It’s not so much about my standards (I’m a noted curmudgeon), more about lost opportunities. The Hunt is a chance for cool people to show other cool people cool tricks with words/images/math/trivia/etc. (for some definition of “cool”). When solvers can do the first cool part (image to sound, neat!) but can’t figure out the last step (“505.570147997”?) it’s sad, to both the solver and the constructor.
It’s also sort of a bummer when a constructor builds a really nice, elegant, fun puzzle that most people don’t get to see, but most could solve. Those should go into the fish round as much as possible.
LikeLike
Rage unlocked the rounds in what I think is the _reverse_ of the canonical order: Hacking, Sci-Fi, Gaming, Pokemon. We chose Hacking first because it specifically alluded to MIT presence, and we wanted to get it out of the way ASAP, as the temperature was predicted to drop below freezing on Saturday. After that I know we intentionally unlocked Sci-Fi next, but I forget why; the last two were arbitrary.
I think we reached the level of being slightly annoyed by the number of puzzles we had open at time, but I don’t think it was ever got to be a serious problem. (We have a registered size of 60; it seemed like our typical number of puzzles open was ~10 but some of those were metas.)
LikeLike
Solved from Seattle on Codex Vorbis this year. I had no involvement in the decision of what order to unlock the islands in (I had no idea that there was a choice until this post), but I’m really happy with how it turned out for me as a remote solver. It felt like there were always just enough puzzles for the Seattle crew to work on, but not so many that we were overwhelmed with choices. By Saturday night we started bottlenecking and I felt comfortable doing stuff like spending a few hours learning about triangle centers, which I’m not sure I’d have been willing to do otherwise.
LikeLike
As a high school Geometry teacher, I’m embarrassed to admit how few of those points of concurrency beyond the major four I actually knew…but I’m glad to have learned something this weekend, and did help in identification quite a bit (Geogebra helped too…).
LikeLike
A few things that I really loved about this year’s hunt:
1) The high production values were truly superb. It’s one of those situations where you don’t realize you’re missing something until you have it, but I really felt that the production values elevated the experience to something a bit more magical than usual, especially with such an adorable theme.
2) The acting was really great. The people playing the different emotion characters did a fantastic job of staying in character the whole time, and everyone from HQ was so encouraging when they were interacting with us. It was really nice and genuine.
3) I think it was one of your hunts that first put a runaround in the middle of the hunt so that more teams could experience it, and I really appreciated that. The thing from this year that I felt was really innovative was the decision to run the finale at the end for multiple teams *and* the decision to live-stream it so that everybody could see it. More than anything else, I truly hope that becomes an annual tradition and that it’s announced ahead of time. That way some people can choose to stick around until Monday so they can participate in it, and those who have to leave early can do so without worrying that they won’t get to experience the endgame.
4) I really appreciated that there were several puzzles that involved running around campus (which are often my favorite puzzles), as well as several puzzles that had physical components. These aren’t too accessible for remote solvers, but the hunt isn’t built for remote solvers. I actually won’t be able to come to campus for the hunt for the next two years (for various reasons), but I still want to encourage the constructing teams to include these sorts of puzzles.
And this probably goes without saying, but there were some really fantastic puzzles.
On the other side of things, I’d like to add myself to the list of people who felt like the unlock rate made the hunt less fun than it could have been. The Providence Transplantations isn’t one of the largest or strongest teams, and I felt as though there were times when we had very few puzzles open, we were stuck on half of them, and the other half didn’t fall into the genres of puzzles that I’m particularly good at. If the unlock rate is going to be close to 1-1, then it would be nice if the diameter is a bit larger.
Also, I don’t think that many of the events this year actually had different teams interacting with each other. This was not optimal.
LikeLike
Care and Feeding paired up two teams to a group, and the final event had everybody in the room collaborating. I can’t speak to the others.
LikeLike
At the risk of seeming like a grump, let me make a pitch *against* “high production values”. Now, perhaps I’m spoiled by both having worked on a lot of professional theater and watching a lot of professional theater but — it was still very clear these were puzzle authors, not professional actors. And that’s fine! Although the props were beautiful and appreciated, there were a number of moments where HQ visits couldn’t be timely because there were only two copies of the “actual character” and their elaborate costume, and that even caused puzzle confusion when they tried to have a single character leave multiple items at once.
It would have been far better to just have HQ not-in-character deliver the appropriate items and explain that the character visit would be later. We’re puzzlers, we understand. We’re not here to watch the show, really, we’re here to solve puzzles — although we’ll certainly appreciate the scene when it arrives. We’re adults, we can watch the story out-of-order even and figure out how it fits together.
So, sure: high production values. But remember that the puzzle event is actually the important thing, when push comes to shove. If the desire for high production values is causing hour-long puzzling delays, something has gone wrong.
(For example, we waited in line 40 minutes just to get into the room for Disgust, then spent an hour in the room with the puzzle, because the physical object was defective. There’s basically no excuse for forcing teams *ever* to wait in line instead of puzzling. If there’s a line, collect their team name and call them back on the phone to return at a specific time.)
LikeLiked by 2 people
The ‘wait in line for Disgust’ thing was very odd. We eventually just left two people there while we did the rest of the runaround – and they reported that it took them well under 5 min once in the room, they had no idea what other teams were doing :).
LikeLike
Hello, I’m sorry you had to wait for 40 minutes during the fun-around. We certainly tried to correct things as soon as possible, but sometimes that took a while. If I remember correctly, in this case we did not realize that the Disgust artifacts would lengthen from handling and make the challenge in the room impossible. I’m sorry for the trouble that this caused teams.
I would also like to clarify one thing – we did send people out of character to deliver items several times. In addition to sending out characters to deliver multiple times. Most of these happened during Friday afternoon. We sent whoever we had available to deliver as much as possible as soon as possible. We simply could not keep up with the demand created by 80+ teams and continue staffing everything else. Hopefully it did not ruin your weekend.
LikeLike
Left as an Exercise here. Our island order was Hacking, Games, Pokemon, Sci-Fi. We picked Hacking first because it said physical presence was required and we didn’t want to get burned on that later. We wound up getting burned on that up front, but we did eventually finish it. I think we picked Games next because we were able to guess at Catan from the flavortext. At some point, Henry or James from L&O came by our HQ, asked what islands we were working on, winced, and suggested Pokemon for our next unlock. Our completion order wound up being Games, Pokeon, Hacking, and then about half of Sci-Fi (solving 2 submetas, being close on 2 more, and missing about 4 puzzle answers).
I think the CYOA aspect really throws a monkey wrench into fairness. It’s impossible to predict what the actual effects of the ordering are — taking the hard round first could be an advantage because you don’t get slammed on it late, or a disadvantage because you have a harder time progressing to subsequent islands — but given that the margin of victory was an hour, I can easily imagine scenarios in which Left Out or even Palindrome might have won with a different unlock order. It’s definitely a cool idea, though.
We had some issues with the puzzle unlocking structure as well. A 1:1 unlock rate is brutal unless the puzzles are impeccably edited. The puzzles were well edited, but it only takes a couple of too-hard ones to grind everything to a halt. And having Games be not only 1:1 rate but 1:1 structure, where puzzle Y is ONLY unlocked by solving puzzle X, was even more of a challenge.
That being said, there was an awful lot I liked about this hunt. I understand the production values occasionally backfired (interaction delays, technical difficulty with the radios) and negative experiences weigh more heavily than positive experiences, but there were a lot of positive experiences. Many thanks to L&O from Left as an Exercise!
LikeLike
(thought I posted this last night, hopefully it doesn’t show up double.)
When we realized that the announcement meant low branching factor rather than easy or short, I said “Well, I guess this Hunt is optimized for smaller more experienced teams. Anybody know if Setec exists this year?”
(I will admit that I had kind of psyched myself up for it meaning easier/shorter like last year, mostly because for the first time in 16 years I made a non-Hunt plan during Hunt (nephew’s birthday party on Sunday). And well, because as a parent, longer times leaving my wife doing sole child care is harder. This probably means I should stop doing Hunt eventually 😦 )
Metaphysical Plant opened the islands in the order Sci-Fi, Games, Hacking, Pokemon. Our theory was that getting non-physical-presence ones first would get us the most puzzles the fastest, since physical presence puzzles often were slower to solve (and were less accessible to remote folks, though the team is mostly on-site).
Plant made very very little progress on Games Island. It took us 7 hours to solve Sports Radio, nearly 24 hours for This Year’s Hardest Crossword, and (embarrassingly) 25 hours for Flattery Will Get You Nowhere. I’d say we found the round’s puzzles to be the hardest (though I think we found the metas to be the easiest — I wasn’t involved with Robber though).
The idea to open all puzzles as soon as the coin was found was an interesting one. I definitely like the idea of giving slow teams the ability to try any puzzle in Hunt, instead of having late puzzles only be accessible to strong teams, and I think future teams should consider doing this again if it doesn’t break some element of their structure. On the other hand, it would have been nice if this was opt-in rather than forced. We had had Pokemon open for “only” 12 mostly wee hours when coin was found, and were mostly focused on other rounds during this period, so at least for me getting access to evolved puzzles without first solving the unevolved ones made the concept of puzzle evolution less obvious and took away from the fun moments of “finally solved puzzle X — ha, the next one works how?!?!?”.
By the way, I thought it was super nice how some of the first puzzles (“Yeah, But It Didn’t Work”, “A Learning Path”, and to some degree “In Memoriam”) explicitly taught solvers useful facts about how Hunt puzzles work that I imagine could have been very helpful for beginning teams throughout the rest of Hunt. If every Hunt started with an easy puzzle that reminds you about the top 15 extraction methods, I wouldn’t mind.
LikeLike
“On the other hand, it would have been nice if this was opt-in rather than forced.” Agree with this 100%. If I had been a round away from finishing when the coin was first found, I would have been very frustrated if all of the unlock structure suddenly vanished. Even when my team doesn’t win, I like the pride of completing the Hunt legitimately. Giving teams a button at that point that would break down all the walls seems like a good compromise.
LikeLike
Hiya, I’m Wei-Hwa Huang from Team Left Out, which finished second by 10 minutes to you guys (Dan knows who I am, so the intro is for the rest of you commenters). We had 33 members playing on-site, 21 members playing in Los Altos CA, and 4 members elsewhere.
We opened Hacking first on Friday afternoon, mostly because of the warning of location-specific puzzles. We then opened Catan soon after, thinking that it was going to be more puzzles than it actually was. At around 11PM Friday evening we hit a very nasty bottleneck — we had about 15 people awake at MIT and 17 people awake in California, 9 puzzles open, but only *TWO* non-meta puzzles didn’t require presence on-campus — Special Delivery and Middle of the Road from the Catan Island. Having to bunch up 8 people per puzzle isn’t fun, although we took it in stride because, hey, we know that the hunt isn’t actually designed for us.
About 10 minutes left, we got the message from HQ as to whether it was okay with us that the target brainpower for the third island be raised. We responded that it’s their game to run, and we trust their judgment, but we also explained our current bottleneck and to please keep that in consideration.
We opened Sci-Fi at 4am Saturday, and that meant that we had plentiful puzzles for most of Saturday. The final Hacking meta wasn’t solved until after 10pm on Saturday; I wasn’t involved in the solve.
The Sci-Fi meta was solved at midnight between Saturday and Sunday. Ian Tullis had exactly the correct approach at 9pm Saturday, but it took us until 2am Sunday to have the right expertise to do the calculations. And then we probably lost another hour because someone used “M” to mean “milli” and someone else thought it was “mega”. So it fell at 3am Sunday.
The Games meta really kicked our ass. Mainly, it was too easy to assume “they wouldn’t make us construct this giant 3D structure for the sole purpose of pairing up tiles, and then throw away all the rest of this data”. Someone on our team had tried to correct approach Saturday afternoon but their data had too many errors for them to notice it was working. Eventually we solved it cleanly at 5am on Sunday, after having all but one puzzle (Thanks).
This meant most of Sunday was on Pokemon Island. This was a bit disappointing, because Pokemon Island was certainly the most “interactive” of all four islands. Had we known, we would’ve probably opened it second or third, so that the fun stuff would happen on Saturday instead of Sunday.
At around 3pm Sunday, we had three puzzles left: The Pokemon meta, Hacking: Model Kit (because we apparently don’t have an organic chemistry expert on our team), and Pokemon: Vain Snowball (the not-a-Duck Konundrum puzzle) that was being worked on for about an hour. We had enough bucks to buy 1.9 puzzle answers, so once we hit 1.95 at 3:13pm we bought Model Kit. At 3:22pm we had enough to buy the puzzle, we had the angsty question of whether we buy the Konundrum or not, since the five players working on it said they were still making progress and having fun. So we figured we’d hold off unless the meta people decided they really needed that last answer. The meta was solved 14 minutes later, at 3:27pm. We called it in, only to be told that some other team had just started the runaround, and whether we’d be okay with waiting. Of course.
It’s a very, very, strange feeling to have your team’s last puzzle solved be a Konundrum.
So. Everyone on-site starts cleaning up the room. 5 players in California are working on the Konundrum. 4 players in California are shooting the breeze. And 2 players in California, myself included, figure the only puzzle-solving left is to see if we can backsolve Vain Snowball before it gets forward solved. It’s tricky because we’re throttled at only one guess every 60 seconds. And… we fail, because we didn’t think to look at people named Vinson other than Carl. The last puzzle is solved at 4:38pm, and our final runaround starts at 4:44pm.
Second place is awesome.
I wonder how other teams felt about the Y/N hint structure? We actually had what I thought was a very strange situation, where we asked for a Y/N hint three times, and all three times we failed to get a Y/N answer and had our payment refunded.
LikeLike
For what it’s worth, our Team Log lists a final meta solve (and confirm) at 2:44pm; based on the meta solve time you reported, that’s closer to a 45-minute difference (which is what Death told us after we finished) than a 10-minute one. However, we then had a serious discussion to decide if we were capable of writing next year–very few of us wanted to sandbag, but we also wanted to make sure we weren’t claiming a Hunt if there was a chance we would let down the community as a result– so after that discussion the margin was probably tighter. It looks like our last core memory retrieval (the post-meta interaction) completed at 3:33, so presumably that was in progress when you called in Pokemon.
We never used a Y/N hint, so I can’t speak to how that went in general. The questions weren’t for metas, were they? The FAQ noted that those wouldn’t be answered.
LikeLike
Ah yes, the 10-minute estimate was from some of the confusing times around the final solve. 45 minutes is in fact more accurate, and I remember it being banded about in the postmortem.
LikeLike
palindrome was similarly ass-kicked by the robber meta. a big part of this was a failure in team communication: if anybody on our team had noticed earlier in the round that the roads on the map had been used as the unlock mechanism, then we could have avoided the “we have to use these roads on the soccer ball, but how does it work?” problem that we struggled with for about 12 hours. perhaps the flavor text should have put us off this path, but it seems plausible to interpret “you’ll have to build something bigger than a few roads” as either “this does not involve building roads” or “you’ll have to build a *lot* of roads”, which are sort of opposite in meaning. it reminds me of the snakes meta from indiana jones in 2013, where the flavor says “i’m not cut out for this” and you had to guess whether that means “cut out the snakes” or “don’t cut out the snakes”.
anyway, the roads were one of three things strongly suggested by the art choices on the catan island that we felt absolutely sure had to be involved in the meta meta:
1. the robber’s outline is rich uncle pennybags, from monopoly
2. there are four ore tiles and only three sheep, instead of the standard catan distribution of 3 ore/4 sheep
3. the roads
#1 was probably relevant, in the sense that pennybags appeared in a (non-monopoly) 1940 game called dig, which is a pretty oblique way of hinting at the antipodal mechanism (which, i have to say, the flavor does just fine anyway, via “deeper” and “australian” and “across the world”). i still don’t know if #2 mattered. was this another way at *very* subtly hinting at “dig”, because you dig for ore? (… and not sheep?) neither one of these is addressed on the solution page for the robber.
LikeLike
My team is probably sick of me saying this, but my dream is for the running team to bring back the Oracle from 20,000 Puzzles, including the broken math that results in non-competitive teams like mine having approximately infinite questions. Oddly, I felt the Buzzy Bucks were less fun this year, as most of our discussion centered on when we should outright buy answers rather than engaging with the tricky puzzles enough to know what hints needed to be requested.
LikeLike
I was part of the discussion for one of Left Out’s hints, submitted for the Lycanroc 2B puzzle, early on Sunday (2B was our designation for the six blank puzzles whose answers fed into the supermeta). At that point in time the only 2B puzzle you had open was Lycanroc and so it may not yet have been clear to your team what exactly was going on with those puzzles. While they appeared like normal puzzles (albeit blank) on the map, we’d always considered those as part of the meta itself. However, we couldn’t really reveal that information without revealing the island/meta structure, which we didn’t want to do, and so we wound up refunding that hint.
(My personal belief was that once you unlocked any other 2B puzzle you’d probably understand the entire structure of the island immediately, and so we were very wary about giving extra information at that stage.)
I wasn’t part of the discussion portion of the other two hint requests you mentioned, but I did show up back at our HQ right as another one was refunded, I believe for very similar reasons.
LikeLike
For the record, Left Out’s three hint requests were (minor spoilers):
Pokemon: It’s Not Normal — “Is there any data at the actual Stata building, that we cannot get through on-line searches, that will help us?”
Pokemon: Lycanroc — “Will the presentation of this puzzle change over time?”
Pokemon: Harsh Financial Scrutiny — “Do we need to first fix the accounting errors in the financial tables?”
For the first one, the response was that they couldn’t answer the question with certainty, and they refunded our cost. We surmised (probably correctly) that this meant that no data was planted at Stata (otherwise the answer would’ve been an unequivocal “Yes”), but that HQ wasn’t sure if all the window arrangements were deducible from on-line searches. (It turns out that Google Maps’ 3D view actually is pretty accurate and gives nearly all the windows used in the puzzle, but it’s understandable that HQ might not have known this.)
For the second one, our most likely hypothesis was the correct one — that Lycanroc was probably a puzzle that was intended to be backsolvable-only — but our second most likely hypothesis was that since Lycanroc was a Pokemon that changes its form based on the time of day, we seriously thought that it was possible this puzzle would change based on time of day. It didn’t quite feel right because we knew that other teams would’ve unlocked the puzzle at different times. We figured ruling out the latter was easier because we didn’t have to go into the technicalities of defining “backsolve”. The response we got — something along the lines of “The puzzle is correct as is; money refunded” — again allowed us to rule out the unequivocal yes and made us decide to wait for more puzzle solves before asking for more hints.
For the third, there were accounting errors in the tables, but we couldn’t be sure whether those were deliberate and part of the puzzle, or whether they were inadvertent mistakes and we should be focusing elsewhere to solve the puzzle. We got a phone call, asking something like “what accounting errors are you talking about?” — turns out it was the latter and HQ wasn’t aware of the mistakes.
In retrospect, HQ could have just answered “No” to all three, taken our money, and still have been technically correct.
It sure gives me the feeling that there might be an art to crafting a yes-or-no question that cannot be cleanly answered yes-or-no and yet still gives information when it isn’t cleanly answered!
LikeLike
It’s been my firm opinion for some time that for these Oracle-style yes/no questions, a possible answer from the start should be “We cannot answer either yes or no to that question”, which ought not to be accompanied with any refund of points/questions. As you noted, this response can often be extremely valuable, far exceeding the informational value of an actual yes/no answer.
LikeLike