Cute Mage's Tower

Crow Facts

Let me tell you a story about a MIT Mystery Hunt puzzle. The year is 2014. The round is Humpty Dumpty. A new puzzle unlocks with instructions to text a phone number. I take charge and start texting, not knowing that I was about to create a lifelong meme. This is the story of Crow Facts.

For those who are not familiar with the puzzle, here is the brief rundown of the solution. You started texting a phone number who would respond with weird facts about crows, each of which were numbered. The phone number required you to do various tasks to get the facts, such as searching up a specific crow video on YouTube and looking up trivia about different types of crows. Once you had accomplished all of these tasks, you were left with 10 weird facts about crows. Each of the facts was actually from Game of Thrones - you had to identify the speaker of each quote and index by the fact number.

But I’m not here to talk about the puzzle. I’m here to talk about what happened afterwards.


My :Twitter account at the time was still relatively new, so I didn’t have a lot on it. I had the Twitter account since 2011, but I only really started using it in 2013. I decided that I was going to try to use it a lot more during this Hunt, and boy did I tweet. There were some pre-Hunt tweets showcasing my excitement:

Make sure you’re following! Otherwise how will you know if the #MysteryHunt has started yet?

Finished my pregame puzzle! Ready for the #MysteryHunt :#Engagelejeuquelegange!

Then, during the Hunt, I had some vague updates:

I’ve been :running around MIT all day. Fun hunt so far! #mysteryhunt

It’s 11:00 and I feel like I’m on a roll. Let’s see how well this holds up into the night. #mysteryhunt

I promised myself I would actually get some sleep at the #mysteryhunt this year. Time to hold myself to that…

Good god this meta event is hard… #mysteryhunt

I also decided to add things to the traditional Mystery Hunt :Have You Tried List?

Have you tried: Looking in the box of girl scout cookies you bought earlier in the weekend for the answer? #mysteryhunt

Have you tried: dissecting the fake lobster you got at an event? #MysteryHunt

Have you tried: :brute forcing the coin all over MIT’s campus? #MysteryHunt

Have you tried: raising your fist in the air, yelling at the puzzle answers Stephen Colbert-style? #mysteryhunt

Have you tried: getting your entire team to solve one puzzle? #mysteryhunt

That last one is actually pretty interesting, so let’s stop and talk about it now.

Saturday Night

On Saturday night, there were some serious frustrations. At this point, we had solved every round except for the Red and White Knights, but we had a problem. We had just a few puzzles open, and we were stuck. It wasn’t that we didn’t know what to do on the meta – we understood what the meta was doing with only a couple answers and we were just waiting for more so we could figure out what was going on. The problem is that we knew that there were going to be 16 answers in the round, but we only had about half of them available. Most of the puzzles we had available were solved, whereas about three of them were open. We were super bottlenecked.

This brings forward one subtle difference in this Hunt that hadn’t been true since 2009, and this talks about the difference between :throttling and bottlenecking. The issue was that solving puzzles in a round only unlocked puzzles in that round. If you already had all the puzzles in, say, the Humpty Dumpty round unlocked, solving more puzzles wouldn’t unlock more stuff for you to do. The :train tickets did help with this, but they only unlocked new rounds for you, not later puzzles in those rounds. The problem is, solving metapuzzles also did that. If you were solving metapuzzles at a reasonable pace, then the train tickets didn’t do a ton, and solving puzzles in a round only affected that round. Herein lies the problem. The puzzles that we had in Red and White Knights no one really wanted to tackle, so we solved other problems. When we had solved those other rounds, those puzzles were staring back at us. Thus started a significant period where our large team only had two or three puzzles open at a time. We were super :bottlenecked.

:Somehow we got past that block, but at that point we would be in a loop of “unlock 2 puzzles, solve those 2 puzzle, unlock 2 new puzzles, solve those 2 new puzzles”. Because the unlocks were in batches of 2, there would be times where everyone who was still solving would be working on the same puzzle because that’s all we had. One by one people starting going to sleep, until we had a little less than 20ish people in the room. We were in the throes of the night shift.

Now for those of you who :haven’t experienced the Palindrome night shift in the 2010s, allow me to explain it to you. The average age of the participants on Palindrome is… slightly older than many of your standard MIT Mystery Hunt teams. This is, in general, not a bad thing, but it did mean that there were less people to do the overnight shift than you might find on other teams. This is the reason that for the longest time, Palindrome be competitive for the lead at the beginning of the Hunt, but then once the night section would hit we would lose a lot of people and :drop down in the standings. This was a particularly bad instance of that. Most of the veteran solvers had gone to sleep or had left for whatever reason, and it was mostly the 20-year olds and the 30-year olds left. Also, we were incredibly sleep deprived, and we were feeling rather punchy. The barrier between our brains and our mouths was lessened, we were joking a lot, and we reduced :how sure we were of an answer before we called it in. This led to quite a lot of phonecalls with HQ, who seemed to also appreciate the late night entertainment, so we kept doing it. For me, I decided that this was a great time to mess around on Twitter. I joked with the organizers about how they had clearly cyberstalked me to figure out what team I was on, I complained about the difficulty of the meta, and I eventually tweeted this:

Crow Fact #21: Crows were originally used for duck konundrums, but they would always kill the puzzlers… #CrowFacts #MysteryHunt

By the way, this hashtag has now been signed up for #CrowFacts. Congratulations! #MysteryHunt

Crow Fact #47: Crows aren’t afraid of scarecrows - they just pretend to be because they find them hilarious. #CrowFacts #MysteryHunt

At this point Ben and I started tweeting Crow Facts regularly throughout that Hunt. We pretended that a :real account called Crow Facts was tweeting them and that we were just retweeting them, and we posted them throughout the rest of the weekend. At one point, when Palindrome was going through the endgame of the Hunt, I even stepped out of the room and started doing Crow Facts on my tablet to avoid a :Too Many Cooks problem on one puzzle.

Now you might be asking, “Hey Cute Mage. Shouldn’t you not be spoiling a late round puzzle on Twitter?” Well, I have a couple answers to that question. First – I would argue that I wasn’t spoiling anything except the existence of the puzzle – none of our facts were Game of Thrones quotes, so the main aha was still intact. Second – at the time we were less concerned about spoilers on social media, and the team who was running it didn’t tell us to stop despite clearly having seen them. Third – I liked causing chaos on Twitter.

All of this made Crow Facts one of my favorite puzzles ever. It was just silly, good fun. But the story doesn’t end there.

The Continuation of Crow Facts

Making up fake facts about crows was funny, and there was no way that I was going to confine it to just that weekend. Every so often, I would come up with a new batch of Crow Facts and post them on Facebook. This would happen at most once a year, soon after my brain started going :HUNT HUNT HUNT HUNT for that year. It got to the point where other Hunt-related people would send me the occasional article about Crows, or funny facts about Crows, or art that depicted them as horrible killers. It was fun.

In addition, throughout this time I was talking a lot with Ben Smith. Occasionally we would pass puzzle ideas to each other that we were saving for when Palindrome would win Mystery Hunt. One that we would joke about a lot was a sequel to Crow Facts. We’d do it when Palindrome won, and that would happen :eventually, right?

The 2022 Hunt

In the January/February phase there was lots of excitement about writing Hunt, and there were definitely people who were excited about writing puzzles, but we couldn’t actually write puzzles yet. For those who haven’t written a Mystery Hunt, first you need to come up with the theme and story, then you need to come up with the metapuzzles, THEN you have the answers that you can write the puzzles for. Sure, there are sometimes puzzles that can solve to any answer, but the vast majority of puzzles are constructed around a specific answer. However, we were encouraged to still brainstorm individual puzzles that we wanted to write. After all, it wasn’t like there wouldn’t be answers at all – we just didn’t know them yet.

Ben contacted me during this time about the Crow Facts sequel. See, :he had an idea. Ben is a huge fan of Mystery Science Theater 3000. He was really excited about the new series that had just recently been made and had made a cosplay of the main non-puppet character. In MST3K, one of the puppets is named Crow T. Robot. What if, instead of the Crow Facts sequel being about the bird crows, it was about Crow T. Robot? Ben even had the perfect hook into the show that we should use – in the very first episode, Jonah performs a rap about different mythical monsters and what countries they come from. Since the original puzzle made crows seem like scary monsters, this seemed appropriate. However, we still weren’t done. After all, we didn’t just want to remake Crow Facts with a different endgame – we wanted to put our own spin on it.

After thinking for a bit, we realized that the way we would really make it our own was to add something involving Twitter. After all, we didn’t just enjoy the puzzle, we also enjoyed the Twitter shenanigans afterwards. For a little while, I was experimenting with a puzzle that was posted to #MysteryHunt. The idea would be that if you solved the puzzle, you would be encouraged to post something that also fit the category to Twitter, which would give you the answer, and also make the puzzle easier for teams that came after you. However, after much deliberation, we came up with the “sort the tweets” structure that worked great. It was one of our favorite puzzles, but remade with a touch of us in it.

Once we figured out what answer we were getting, that isn’t to say that the puzzle came together perfectly right away. :It failed testsolving spectacularly the first time, :Rex Parker accidentally retweeted one of our tweets, :it was an annoying puzzle to edit, :we had to change the avatar, and :we had to ditch one whole section and rewrite it. It turns out that having your puzzle connect to an exterior site makes it harder to write. But that being said, I loved the excuse to write a ton of Crow Facts. For me, it was the right mix of humor and puzzle that I knew it would be a good time for solvers. :Shame it was placed last.

Wrapping it Up

Crow Facts is still one of my favorite puzzles, despite the fact that it is a bit easier than a traditional Mystery Hunt puzzle (even at the time). It’s a nice example of how your goal for puzzle writing should be to make sure that whoever solves your puzzle is having fun.

Oh, and you didn’t think I was going to end this without writing some more Crow Facts, did you?

Whoops, we have to make a new puzzle for a different round in the 2022 MIT Mystery Hunt. You go dash up your crows and I’ll get Crow. After all, he is doting on me. But what’s special about my crow?

Answer Checker

Merry Christmas!

– Cute Mage

:x Named Twitter

Yes, I am calling it Twitter in this story. I have since :deleted this account since Twitter became X, so the Twitter account was never an X account. Also if Elon Musk isn’t going to respect trans folk, I’m not going to respect Elon Musk.

:x Deleted Account

You may be wondering, “Hey Cute Mage, if your account is deleted, how did you get all these Tweets?” I obviously saved all my data first. I may have deleted my account, but I didn’t want to get rid of the jokes!”

:x Engage Le Jeu

I was on Palindrome in 2014, and Palindrome changes their name to a new palindrome every year. This year we decided on a foreign language palindrome because we hadn’t done one of those before. This means roughly “Start the game so I can win!” While I am no longer on Palindrome, I know that we haven’t had a foreign language palindrome since then, and there are some good reasons for that. I know I spent the entire weekend butchering French.

:x Bad Decision

This turned out to be a bad decision, but this is a story for another blog post.

:x Have You Tried

If you are not familiar with this, here is a link to the list and here is a link to the 2018 puzzle about that list.

:x Again Bad Decision

Again, this was a bad decision. Story to come later.

:x Throttling Difference

Hey all you folks who read my old blog from like 2014 and were wondering when this would ever come up! This is for you!

:x Train Tickets

Okay, quick explanation: Every time you solved a Wonderland puzzle, you got a piece of a train ticket. Once you had unlocked the whole train ticket, you opened a further round.

:x Bottlenecked

So, what is the difference between Bottlenecking and Throttling? I define “Throttling” as limiting a team to only having a certain number of puzzles open at a time. I define “Bottlenecking” as forcing a team to solve specific puzzles in order to progress. I consider throttling to be necessary for a Hunt to function at all – part of the issue with the speed of the 2017 Hunt was the fact that the mechanisms meant to throttle the hunt didn’t work as well as they had hoped. Bottlenecking is a different beast. At some places, it feels appropriate. In the end, teams who are trying to finish will get bottlenecked by one puzzle – whichever meta or whatever else is the last thing between them and the endgame. However, bottlenecks in the middle of the Hunt can be real feel-bad times if they are not handled properly. They do have their uses though, and I think that they should be a rare tool in a puzzle hunt structure creator’s pocket.

I do think that Alice Shrugged’s decision to have these firewalls between the rounds was not great, but 1) from talking with members of Alice afterwards it was pretty clear that this wasn’t intentional choice, but a side effect of other things in the Hunt, and 2) I don’t think is necessarily always the wrong choice. This is definitely a place where reasonable people can disagree. Also, it’s one of the very few complaints I have about the Alice hunt at all. Overall, they knocked this one out of the park.

I have lots more comments about this part of structure, but that’s not part of the story here, and I’ve already digressed :far enough.

:x Or Have I

:Or have I?

:x Yes

Yes, I have.

:x Unblocking

I mean, I remember something, but I’m not like super sure about the timeline at this point. It’s been almost :10 years! If I was more sure, I’d post the exact timeline here, but I don’t want to misspeak.

:x Ten Years

Oh god, it’s been that long? Oh no. Time is happening.

:x Haven’t Experienced

Probably most of you.

:x Win 2021

I also think that it’s no mistake that Palindrome finally won in 2021, when we were all just a stone’s throw away from our beds, so we were all able to take naps without committing to travel times to get to them.

:x Answer Graph

This led to this wonderful graph in the wrapup, showing that Palindrome had 100 more wrong answers than anyone else. I am so proud of that. I’m not sure that feeling was felt by every single member of the team though.

Also, for one puzzle, we completed a task, Alice accepted it, called us to give us the answer we needed to put in the website, and then we forgot the answer between receiving the phone call and putting it in the website and we had to call Alice to ask for them to give us the answer again. It was a wild night. Somehow no alcohol was involved!

:x Too Many Cooks

This is referring to when the issue when there are too many people that we’re all stepping over each other making it harder for all of us. In this case, we were manually playing a record and we needed to be quiet so that we could hear it. I figured that the rest of the team had it and I chilled in the hallway so that I wasn’t making any noise. I had basically understood what was going on and didn’t need to get involved with the solving anyway.

:x Real Account

Just to be clear, @CrowFacts was a real private Twitter account, and I don’t know who controlled it, if they were active, or if they even knew that they were getting pinged a million times that weekend. It was easy back then to pretend to retweet someone though because instead of putting the Tweet from their feed into yours, it would repost the tweet starting with “RT @” the person’s username. So we just started all of our tweets with “RT @CrowFacts” to continue the joke.

This may have backfired on us since I can’t find the RT ones in my Twitter archive, and I think that something may have happened to the @CrowFacts account or they complained or something and all of those came down.


Okay, so there’s a point in the year where my brain just switches all of its free time towards thinking about the MIT Mystery Hunt. Usually it’s like the middle of November. It’s the reason why you’re getting a lot of blog posts :in quick succession right now.

:x Train Rides

Also, I’ve been traveling by train to visit my family for the holidays, and writing blog posts is something I can do with limited internet.

:x Spongebob

Insert Spongebob title card saying “Many :Second Places Later…”

:x Second Place

Okay, they weren’t all actually second place, but we came close every year from 2013 to 2020.


Actually, I think Wil had the initial idea and told Ben and Ben was like “YES” and fleshed it out more.

:x First Fail

The first time we testsolved this, the key words in the facts (the parts of a pocketknife, the units of length, the words that need accents) weren’t always at the end of the fact. This caused way too much confusion and was just too finicky and too hard. We realized pretty quickly that we needed to rewrite things in order to put the relevant word(s) at the end of each fact. This is why you testsolve!

:x Rex Parker

We learned that Rex Parker searches for the word “crossword” on Twitter, and so he retweeted the one that said something like “Crow’s least favorite crossword fill is ICE AX.” We were super worried about people finding the account ahead of time, and ended up nuking the account and filling it again privately.

:x Editing Tweets

As you probably know, pre-Elon Musk you couldn’t edit tweets, which meant that if we found an error in one, we would have to delete it and post it again. However, because you couldn’t manually reorder your own feed and we wanted things to be in alphabetic order to show that the order of the tweets was irrelevant, this meant that we would have to delete everything after it and repost all of that. Basically every time we did a new testsolve Ben had to nuke all the tweets and repost them.

:x Crow Avatar

Since we knew that we were going to do this puzzle very early in the Hunt, Ben picked up the twitter account @CrowFacts3000 in February. He then put up a profile picture of Crow T. Robot up and just left the Twitter account alone for a little while. The problem was that the profile picture was hexagonal, and in the middle of 2021 Twitter started experimenting with hexagonal profile pics for NFTs. Since we didn’t want to be associated with stupid scams, we changed the profile picture.

:x Werewolf

One of our original categories was “Roles in One Night Ultimate Werewolf”. The problem was that we needed literally every role to make the W happen, which included Werewolf. Unfortunately, it turns out that Werewolf was also the name of a MST3K episode. That broke those two, and so Roles in One Night Ultimate Werewolf were removed. We were able to save a couple of the facts (For example, the Hunter Pence one was originally for the Hunter role before it was used for Stream), and then we moved some stuff around to add in Bodies of Water.

:x Placed Last

For the record, it was not placed last to save on costs, although I was a little concerned about that. Testsolving showed that each team took about $3 worth of Twilio credit to solve, which means that if each team that had registered had solved it, :it would cost us a little under $1000. Fortunately, it was in the last round, so not many teams got to it during the Hunt itself, and it ended up being a negligible amount.

No, it was placed last for two reasons:

  1. It was not a crucial answer for the Sci-Fi Meta Round
  2. I didn’t want people accusing me of using my position to put all my puzzles early in the round so that they would be guaranteed to get forward-solved.

In particular, the last puzzle of the Hunt is the least likely to get forward-solved, especially with ours where we knew that the Sci-Fi meta was easier to solve with less than the full complement of answers. I figure that if I’m going to put anyone in a bad position, it might as well be something I wrote. Many things in a hunt come down to what is best for HQ or for the solvers, but some things are politics (or trying to avoid politics).

:x Number of Teams

I think one thing that many teams who have not run a hunt since about 2020 are not aware of is just how much bigger the Hunt is than before. Small interactions with teams or even small physical puzzles can cost a LOT of money due to the scale. I am glad that this many people get to experience the Mystery Hunt, and I’m glad that we’re back in person which will help mitigate the scale issue, but it’s just a thing you have to think about and prepare for if you’re the running team. We no longer can run a puzzle for one team at a time in a random building hallway anymore.