16 January 2018

Monday was Martin Luther King Jr. Day, and as such it's time for my annual Mystery Hunt review post. For those who don't know, Mystery Hunt is an event that occurs each January on MIT campus consisting of a large number of puzzles for teams of various sizes to solve over the span of a weekend. Most of the top teams range from around 40 to 100 people. The team I'm on, Hunches in Bunches, weighed in at around 45 people this year.

Although we've thinned out a bit since our formation in 2016, I don't feel as though our solving power has decreased much, if at all. Our team started as the half of Random Hall's team that was most involved in the writing of the 2015 hunt, so our team members already tended to be strong solvers, but there's still a noticeable set of core solvers, and that core has remained intact.

Mystery Hunt has a tradition where the winning team writes the hunt for next year. Writing a hunt is a huge amount of work, and so whether it is a blessing or a curse depends on the team. Thankfully, it is essentially impossible to win the hunt by accident, so the winning team will be one that is interested in writing. Last year, Death and Mayhem was the winning team, so they wrote and ran this year's hunt (under the name Life and Order).

For this year's review, I'm going to split it into two separate posts. This post will talk about the part of hunt outside of the individual puzzles, such as the theme, production, structure, pacing and so on. While there will be some spoilers here, they should mostly be minor and not completely spell out any solutions.

The next post will be a short review of each individual puzzle that I did significant work on. That post will have significant spoilers.


This year's theme was the movie Inside Out. As a "bait and switch", the kickoff invitation told us that the hunt would be the Heath and Safety Hunt, and the kickoff skit transitioned from the Health and Safety Hunt kickoff to inside the mind of one of the audience members, where we saw the five emotions Anger, Sadness, Joy, Disgust, and Fear react to the Heath and Safety Hunt's kickoff in a way that resulted in the "memory cores" (glowing orbs) being spilled all over and so we'd be tasked with assisting in recovering them.

When we got to our team's solving room, I was told that this skit was a reference to Inside Out. I had never heard of the movie before, and the last time that I missed an opening skit reference like that it was the Producers hunt, for which theming was a big reason that I didn't enjoy it as much as other hunts. So I was apprehensive. But they actually did a great job of taking that theme and using it for art direction but not for puzzle direction. I never felt as though I was missing out on anything for not knowing the movie, and the theme didn't negatively impact the diversity of puzzle types.

All that said, I consider this year's kickoff to be the best kickoff I have ever seen, and that brings me to the next point.

Production Value

As I said earlier, the kickoff skit started out as the kickoff for the Health and Safety Hunt before transitioning to the mind of an audience member. When the transition happened, the lights turned down so we couldn't really see as they changed the props on stage. Then, after a bit, a bunch of orbs lit up in various colors, eliciting an "oooh" from the crowd.

In addition to that, as part of the skit, Life and Order had a prerecorded video of the Heath and Safety Hunt kickoff running, which was carefully timed so that everything would match with the pacing of the skit. To be honest, when I was watching I didn't know whether the video was timed or whether they had some sort of controller offstage.

Of course there's more parts of hunt than just kickoff. One of the most important pieces of production value is the hunt website. Life and Order put a lot of thought into their website design, and mentioned some of that during wrapup. Among their guiding principles was that every time your hunt state changes, for example by unlocking or solving a puzzle, the appearance of the website would reflect that. For example, in the emotion round, the circle representing a puzzle would be colored in when you unlock it, and lit up when you solve it.

Additionally, the website was quite stable. We had no issues with it for almost the entire hunt. There was one short period where the website became unavailable on Sunday evening, which happened to interfere with calling in one of our answers, but we weren't too upset since the winning team had already finished.

The one other complaint that I had about the website was that the Sci-Fi round page was very jumpy and hard to control on my laptop. The result was that I avoided going to that round page if I could avoid it. Unfortuantely, the round page was relevant for some of the metas, so it would have taken me a long time to find those things. Luckily for me, the one meta I did a significant amount of work on didn't use the round page, but I wish that particular page were more performant.

These sorts of things might sound minor, but they actually contribute a lot to the overall experience of the hunt. The 2016 Hunt was an example of unforseen technical issues, and you can read about how it affected my experience in my post from two years ago.

Puzzle Quality

While there's a ton of important aspects to a good hunt, the single most important one in my opinion is the quality of the puzzles themselves. But in addition to having good puzzles, the puzzles need to be consistently good. When I solve good puzzle after good puzzle after good puzzle, I notice an effect in that I start to trust the author. I'm not sure I can really describe what changes when I trust the author, but it makes the experience much more enjoyable.

In my opinion, Life and Order did an amazing job, and over the course of hunt I did grow to trust their writing skills. Post-hunt, while reading the solutions, I mostly felt that the puzzles that we got stuck on were fair and we simply had a blind spot. For the Hacking metameta in particular, I think the place we got stuck was simply a result of sleep deprivation making us tunnel in on a wrong idea and miss the right one. I did find a couple of extraction methods that I'm a bit unhappy about, but one of those has been used before in other puzzles, and it's just not my style.

Hunt Structure

This year's hunt structure hit a lot of my personal favorite things to see in a hunt:

  • Puzzle answers used in multiple metas (Emotions round, Sci-fi round)
  • Intended backsolving (Pokemon round)
  • Geometric structure (Sci-fi round)
  • Real Metametas (All the islands!)

On the point of metametas, I think it's particularly interesting to compare this year's hunt to the 2011 hunt in that regard. At a high level, this year's hunt and the 2011 hunt have a lot in common. There are five super-rounds total, with one serving as the introduction and the remaining four forming the bulk of the solving experience.

Of the four non-introduction rounds in the 2011 hunt, Mega Man and Katamari stand out in my mind as some of my favorite rounds of all time. They're both composed of puzzles that are divided into groups, and the groups combine together in an interesting way to form the full round. This year's hunt captured those properties again in the Pokemon and Sci-fi rounds.

Before I get into the specifics of this year's unlock structure, I want to give a nod to the fact that teams this year were allowed to unlock the four islands in any order, with a little blurb about each round to inform the decision. There's a lot that could be said about decision making in unlocks, and I don't believe that it's definitely better nor definitely worse than the alternative, but I want to call out some of its effects.

The first and most obvious is the issue of fairness. In my opinion, the Pokemon round was the easiest. In particular, the metameta was the one that we had the least trouble with (of course, the second place team solved Pokemon significantly later than the other three, so this was certainly not universal). So I felt that a team that happened to choose to unlock Pokemon last would have an advantage, as they'd be able to keep the hard round open in the background while working on other things.

In addition, the Pokemon and Sci-fi rounds had the widest unlock structure (more on that in a bit), and so they are the ones where it hurts least to have that round as the only remaining puzzles at the end of hunt. That said, I don't think this fairness question is a dealbreaker by any means.

If/when this decision is repeated in future hunts, I think it would be cool to explore different mechanisms for revealing information about the unlockable rounds. We got thrown off a bit by the fact that the Hacking round's description mentioned physical presence on campus, which we interpreted as runaround-type-puzzles, and as such we avoided using unlocks on it in the middle of the night. It turns out that it meant physical puzzles, which we could do in the comfort of our own room, so really we could have done it at any time.

The second major effect of this decision is that teams will now be doing the HQ interactions and campus-based puzzles more spaced out through the hunt. I think for the solving teams this is a huge win, because each individual interaction will be less crowded. On the other hand, as pointed out at wrapup, the writing team now needs to be equipped to handle any interaction at any time. As a solver I'm glad that Life and Order made this decision, and I hope that future teams can find a way to mitigate the logistical problems.

In addition to reducing crowding, this also plugged some common information leaks about how your team is doing during the hunt. For example, there was errata for the puzzle Zelma and Frank release before we had unlocked it. But, because of the choose-your-own-adventure unlocks, this didn't necessarily mean that we were way behind. We figured that it was probably at the end of the Hacking round, which was the last round we unlocked, and so it could have been that the errata was from a team that chose to unlock it first.

The third major effect was pointed out by the writing team at wrapup, and mostly affects the writing team rather than the solving teams. Because teams can choose their unlocks, each puzzle will have some teams looking at it early on in the hunt. This is quite beneficial because it means that any needed errata is likely to be found more quickly and affect fewer teams. Furthermore, it means that the writing team will have confirmation that all of their puzzles are solvable long before the coin is found, and as such they can have confidence to not try to push teams in the right direction.

Story time! When we were running the 2015 hunt, pretty much every team got stuck on the Spotted Tower Meta. After the top two teams had been stuck on it for a few hours, a conversation started in our HQ about if we should ever try to prod the teams in the right direction, and if so, what would be a fair way to time the hint to the teams in contention. Luckily, we didn't end up having to make a hard decision, because we all agreed to leave the teams alone for at least a few more hours, and a bunch of teams broke through in that time.

Allowing teams to unlock rounds in different orders completely avoids that issue, and Life and Order mentioned that it allowed them to be much more comfortable with the puzzles that teams were stuck on near the end.

Unlock Structure

I mentioned earlier that Pokemon and Sci-fi had the widest unlock structure. Width in this case, refers to the number of unsolved puzzles that are unlocked at one time. Let me quickly go over the unlock mechanisms as I understand it.

  • In the Pokemon round, you'd always have 3 first evolutions unlocked (until running out). Solving a first evolution also unlocks the evolved form. Looking back at our team log, the first evolutions might be unlocked slightly slower than this. Update: I've been told that each first evolution counted toward half of an unlock in addition to the evolved form, and the non-blank evolved forms counted for a full unlock. So your width would increase by 0.5 for each first stage puzzle solved, until you started solving the blank puzzles, at which point it would start decreasing again.
  • In the Games round, solving a puzzle unlocks any puzzle that is across a road from it.
  • In the Sci-fi round, solving a puzzle unlocks any puzzles connected to it via corridors.
  • In the Hacking round, you have 3 puzzles unlocked within each phase (until running out). Solving a phase meta unlocks the next phase.

The Hacking round is the easiest to analyze here. The meta-structure of the round and backsolvability meant that you generally only cared about the puzzles in the latest phase, so the effective width was 3 (4 if you include metas).

In Games, most of the hexes only had 2 roads next to them, not including the desert tile, and since you use one to unlock the puzzle, solving a puzzle would unlock one new puzzle. As a result, the width is the number of puzzles you start with (which I belive was 3).

In Pokemon, you get to have 3 first evolutions, but then you also have any number of solved first evolutions with unsolved evolved forms. As a result, you can have more than 3 open puzzles, even not including the blank puzzles.

In Sci-fi, it looks like most rooms are connected to 3 other rooms, meaning that solving a puzzle unlocks two. As a result, this round unlocks quite quickly compared to the others, and the effective width is much higher.

I will say that even with a team of about 45, having only 3 puzzles open at a time is not very fun. As a result, unlocking Hacking last was rather unfortunate for us. For most of Sunday, we were done with Pokemon and Sci-fi, and had effectively only the metameta of Games remaining (there were a few unsolved puzzles, but we clearly had enough to solve the metameta). So our whole team was crowded into three puzzles, which is particularly suboptimal when all three are physical.

As such, it came as a great relief when Life and Order announced that they would unlock the whole hunt for all teams, because it meant that we went from a couple puzzles in the build phase to having those plus all of the deploy phase puzzles, and could separate out into more manageable group sizes. I don't know if there's a right answer here. Perhaps it would be interesting to allow teams to spend their hint currency on unlocks.

Backsolving Support

At wrapup, Life and Order mentioned that they tried to avoid puzzles getting backsolved, because they didn't want to have authors feel like their puzzles were being skipped. I can sympathize, since during our hunt, it felt like a lot of teams essentially skipped The 10,000 Puzzle Pyramid due to its placement in the round and the backsolvability.

What's interesting, though, is that I felt that the puzzles were just as backsolvable as usual. Throw in the fact that the Pokemon round had blank puzzles which are effectively forced backsolves, and it felt like the hunt supported backsolving more, and not less.

Now, I'm personally a huge fan of backsolving, so I was quite happy with the state of affairs. As always, the prevalence of thematic answers makes backsolving easier, so it's interesting to see how people deal with that clash of priorities.


This year's hunt continued with the Oracle system of hint giving, in which teams acquire some sort of hint points over the course of hunt, and have the ability to spend them to ask yes/no questions on non-meta puzzles. Life and Order also added the ability for teams to spend the equivalent of five hints to buy an answer instead. This seems to be a popular system, and so it makes sense to continue with it, but I wonder if there's room for more experimentation.

I don't have a huge amount to say here, because our team is quite averse to taking hints. We didn't spend any of our Buzzy Bucks until Monday morning, despite the fact that we could have bought most of the remaining answers much earlier. Of course this is largely because we aren't interested in writing again (yet?), so our goals are to enjoy solving puzzles, and asking for hints or answers is antithetical to that. I mentioned before the possibility of using hint currency for extra unlocks, and that would probably be a good way to get a team like ours to use it more liberally.

Length of Hunt

There's a huge diversity of opinion of optimal hunt length. For me personally, this year's hunt hit my favorite timing of the coin being found on Sunday afternoon. On the other hand, I don't think it's useful to put much emphasis on hunt length during the writing process, because it's so difficult to predict. This year's hunt is a great illustration, since the writing team apparently had a pool for predictions and nobody guessed anything past Sunday at noon.

However, there's one length of hunt that the writing team has complete control over, which is how long HQ stays open after the coin is found. This year, Life and Order decided that they'd keep HQ open until 10am on Monday, which was quite the extension compared to 6pm on Sunday which had been commonly done in the past. I think this was an amazing decision and I hope that it catches on for future writers. I think a two hour gap between HQ closing and wrapup is sufficient, and it allows people to continue having fun with the hunt for longer.

Writing Team's Values

At wrapup, Life and Order showed a slide with the following values that they kept in mind while writing:

  • Assume good faith among the team
  • Make Hunt fun
  • Set clear expectations for participants
  • Don't write Hunt just for the winners
  • Hunt is more than solving puzzles
  • Make every puzzle matter
  • This is an MIT Hunt
  • Puzzles can be obscure
  • Subvert expectations

I want to call out the last two in particular, because I feel that they help produce some of the best puzzles. As a writing team, there's a particular conflict with regards to making obscure puzzles, because it can be difficult to get two clean testsolves on a puzzle that relies on something obscure, and two testsolves is the usual baseline for having enough confidence to allow a puzzle into the hunt. On the other hand, as a solver, obscure puzzles create some of the best feelings. Rather than feeling like a random person searching for things on Google, it feels as though you were an integral part of the soltion process, and that without you it might not have been done, or would have been much slower.

The other value, to subvert expectations, feels very similar in spirit to how I feel that the best puzzles are the ones that push the boundaries of what it means to be a puzzle. Most new puzzle writers (myself included), will naturally start by writing ISIS puzzles (Identify, Sort, Index, Solve). Often, these puzzles will be based on a theme that excites the author, and if they happened to get into the hands of someone else excited by that theme, it would make a solver happy.

Unfortunately, ISIS puzzles, especially those written by new authors, will tend to be straightforward enough that someone without much connection to the theme can solve it, which makes it less likely for that one excited solver to see it. I think that explicitly listing some variant of these two values allows authors to feel more free when designing their puzzles, and to feel as though they can create puzzles that not necessarily anyone will be able to solve.

Hunt Ranking

As usual, I'm going to close with my personal ranking of how much I enjoyed each hunt that I've solved onsite. I started hunting in 2009, and wrote the 2015 hunt, and so far no hunt really captured the magic of that first one. When we were listing our goals for 2015, I said we should aim to surpass Zyzzlvaria, and I hope that we did, or at least came close. Of course, since I was on the writing team, Zyzzlvaria was still the top of my ranking.

This year's hunt changes that. I loved almost every minute of the weekend. I think I got less sleep during this hunt than any other because of my excitement to get back to puzzling. For a while, I wasn't sure if I should put this hunt at #1 or #2, because it's so hard to compare things separated by nine years. But of course it has to be #1, because hunt has evolved so much in those nine years, and the fact that Life and Order was able to compete with the magic of my first hunt means that they definitely surpassed Zyzzlvaria. More than that, this hunt really inspired me, and reignited my desire to write again. Maybe in a few years the rest of Hunches will feel the same way.

Here's hoping that we see this hunt surpassed soon!

  1. Inside Out (2018)
  2. Escape from Zyzzlvaria (2009)
  3. Coin Heist (2013)
  4. Time Travel (2010)
  5. Video Games (2011)
  6. Alice in Wonderland (2014)
  7. Dungeons and Dragons (2017)
  8. Huntception (2016)
  9. The Producers (2012)