I am very honored and privileged to have had the opportunity to run my first hunt with such a formidable audience. I did not expect such a warm reception! I am grateful to everyone who participated and left feedback on the puzzles and the website!
I want to thank everyone who worked on the hunt with me and made this all possible:
- Puzzle construction: Timwi, TasThiluna, Roger Wrightshoe, CloudGaming, EpicToast and GhostSalt
- Additional puzzle ideas: Koala, Deusovi and cmɢʟee
- Test solvers: GhostSalt, Deusovi, Roger Wrightshoe, CloudGaming, cmɢʟee, the teams Enigmatics and Skycrab Holders, and many more individuals
- Puzzleton art: EpicToast
The overall resonance to the hunt was much more positive than I had anticipated, which is very nice. On this page, I am going to try to summarize some of my experience and answer some questions we received from participants.
- 287 teams solved at least one puzzle.
- 81 teams solved all 27 puzzles within the 10 days.
- At time of writing, a further 8 teams finished the hunt afterwards, and this number can rise indefinitely (as anyone can still register teams and still play the whole hunt).
- 9995 total guesses were submitted, of which:
- 4578 were correct answers and
- 566 were intermediate answers that spawn a “keep going” message.
- 1704 hint requests were submitted (and all were responded to).
- 296 hint requests were refunded. Next hunt we’ll have a “follow-up” feature.
- 208 teams submitted at least one hint request.
- 106 teams submitted at least one hint request that was refunded.
- 60 teams submitted at least ten hint requests.
- 5 teams submitted at least ten hint requests that were refunded.
- 2 teams made complete use of all of their 21 hint requests.
- For puzzles feeding into Inside Out, there were 38 backsolves*:
- For puzzles feeding into The 47, there were 11 backsolves*:
- For puzzles feeding into The Nuke, there were 146 backsolves*:
- 0 teams needed a hint request to solve Borders. (We did receive a single hint request for it during the hunt, but that team didn’t end up solving it.)
- 29 teams solved The Nuke without any hints (the puzzle that statistically required the most hints per solve).
- 33 teams solved Face to Face without any hints (the puzzle that statistically required the second-most hints per solve).
* We assume solving a puzzle after solving its associated meta counts as a “backsolve”. This metric isn’t perfect.
We knew going in that automated e-mails were going to be a problem, as I already struggled to find an online service that would allow us to send notification e-mails to 1000 addresses when the hunt starts, when it ends, and when there’s errata. As a result, I resigned to not having mass e-mail notifications at all and was going to send out notifications only for hint responses.
Unfortunately, I couldn’t get this to work either. I decided on using GMX for this simply because I’d used it before and knew that it supports SMTP, but early into the hunt, they locked down the account and required me to change the password, and this kept happening over and over, so we gave up. I asked their support service about it, and 5 days later I got a response saying that this is an automated anti-abuse measure and cannot be turned off.
For the next hunt, I plan to allow users to specify a Discord webhook address so they can get hint and errata notifications within their own Discord servers.
I overestimated solvers’ internet connection speeds when deciding to embed images directly into the puzzle pages as data URIs. I thought my internet was below average and for me even the chunkiest puzzle page (Postcards, 4 MB) only takes 2 seconds to load. However, several people complained about this, especially because you needed to load all of the images in order to see whether a hint response had come in. In the next hunt, we will put the hints on a separate page.
I was also surprised at the volume of complaints about a supposed lack of clipboard copyability. I wish to go on the defensive about this one, for several varied reasons:
- First off, Postcards has a copy to clipboard button and is the only puzzle in which the clues couldn’t just be selected and copied because they’re images. Comebacks has a provided spreadsheet that you can just take a copy of.
- In No, U, the clues are copyable in every browser I’ve tried. In Chrome they come out perfectly fine. In Firefox, they come out smushed into a single line, but since every clue ends with a close-parenthesis, it can be un-smushed within less than a minute. (Despite this, I added a copy to clipboard button after the hunt.) In Special Names, the grid also copies just fine.
- If you are OK with puzzles that rely on tools like OneLook or Nutrimatic, you should be OK with puzzles that assume you can use other tools, especially tools that come with every browser. In particular, I do not feel bad about the grayscale ordering in Throw — any color picker, or the browser’s Inspect feature, works here — or the point coordinates in Circles, which aren’t required to solve the puzzle at all (I expected most teams would use Inkscape or take a screenshot of it).
- In puzzles like Objectionable Ranking, Totally Ordinary Crossword and Swerve, providing a premade spreadsheet comes with a number of issues. Firstly, when I’m a solver, I much prefer to create my own spreadsheet for these, or solve them in Paint.NET or Inkscape (graphics) or Qxw (crosswords) instead, and I consider the choice of software or method to be part of the puzzle. For this reason, I decided to put everyone on the same footing and not give an advantage to competitive teams who happen to prefer spreadsheets. Secondly, as a setter, I would have had to make the difficult choice between trying my best to replicate all of the puzzle-relevant graphics/quirks on the page exactly (which is very hard) or leave some out, which is just as likely to provoke the same amount of ire. Not providing a spreadsheet was therefore the only way to sidestep this issue.
Only one respondent in the feedback form said this was their first hunt, so if we scale this up to the number of participating teams, there would likely be about 6 newcomers to the world of hunting. Welcome!
On the questions “How do you feel about this hunt’s length and difficulty?”, the overwhelming majority (86.7% and 71.1%, respectively) responded “just right”. A minority of comments we received highlighted the somewhat wide range of difficulties presented: the hardest were rather brutal in comparison to the easiest puzzles. As one respondent put it, this “produces some whiplash”. This did show up in test solving but we didn’t have enough experience making puzzle hunts to know that it was an unusually wide range for a hunt this size.
To the questions “Which puzzle(s) in this hunt did you like the most?”, Board State is a clear winner, followed by Postcards and Objectionable Ranking. This is quite different from the average ratings given on the website (see Stats), where the clear winners are Totally Ordinary Crossword and Rectangle Mangle. This likely shows a statistical bias: the full feedback form was probably only filled in by completionists, who tend to be more experienced puzzlers, while the star-based rating was likely given by more newcomers.
However, the biggest surprise to me here is that the first two puzzles written for the hunt — The Nuke and Board State — were generally rated among the worst and best puzzles in the hunt, respectively.
Another surprise is that the subjective difficulty rating given in the feedback form is wildly off from the statistics on number of guesses and hint requests. People only agree on Face to Face. But according to the statistical metrics, The Nuke is the hardest puzzle, but in the feedback form it’s only in 8th place! The situation is reversed with Circles (feedback rated it difficult, but statistics show much fewer guesses and hint requests).
The puzzle that I had the most fun constructing, Signal Link, was also the most backsolved. Sadface!
When asked about the meta structure, respondents tended to describe it as “a weird choice” and “a strange concept”. Only one respondent called it “clever and innovative”. It’s really hard for me to tell which novelties will be received as off-putting and which ones are seen as an interesting surprise. Most hunts’ meta structures form a balanced tree (each level is either all metas or all leaf puzzles) so I decided to shake it up and have a somewhat lopsided tree where two metas are mixed with leaf puzzles on the same level (which also played well into the mechanics of The Nuke). What I didn’t consider — and I’m grateful for one respondent to point this out — is that “some answers are effectively worth more than others” because one of the answers to an 8-feeder meta “effectively only gave 1/8 of an answer for the final meta. [...] All answers should be treated equally!” That’s an interesting way of looking at it that I hadn’t considered.
Lots of disappointment was raised about the lack of a full story and the lack of an epilog when solving the final puzzle, resulting in a bit of an underwhelming or anticlimactic ending for some. I am not a writer and I don’t come up with stories, and when I solve puzzles I tend to find story/theming more distracting than engaging. The little story that we did have — there’s a city and 13 locations in the city have bombs planted in them — already caused frustration: solvers complained that the location information was mostly unused, even when they recognized that it did play into the final meta. This indicates to me that any amount of story I add on top of the puzzles only serves to add red herrings. So... this is not going to change unless somebody volunteers to write a story for the next hunt and also volunteers to be the scapegoat when solvers complain about the red herrings this produces.
The limit of 5 guesses per 6 hours was too tight. We will make this more lenient in future hunts. Despite, I hope that at least a couple teams experienced a good puzzle this way that they would otherwise have backsolved.
I don’t want to draw too much attention to these, but they occurred in public, so you deserve to know. There were two unfortunate implications that surfaced during the hunt:
- In Throw, one of the darts players used in the puzzle was convicted of rape and is now in prison. After the hunt concluded, I replaced his name with the only other darts player containing a Y whose surname starts with E. I’m lucky there is one. This change is the reason why the most common wrong answer given during the hunt no longer matches what the puzzle-provided order now results in.
- The puzzle Drumheads involved a YouTube video showing a marching parade held in Northern Ireland. The author chose this video simply because he’s one of the accordionists in it, which is a perfectly reasonable impetus in my book. What I didn’t know was that, in the words of one solver, “Orange Day marches like the one depicted [...] cause great upset to the Catholic community [in Northern Ireland] and are regarded as a deliberate provocation by many.” After reading more about the Drumcree conflict on Wikipedia, the author gave me full permission and support in pulling the puzzle and replacing it with a new one (which we did after the hunt concluded).
Assorted comments from assorted solvers!
- we broke the hunt 🎉
- “PRAISE THE LORD AND PASS THIS” → GO
- — Hm, why is The Nuke in a different color than the other metas?
- You’d be surprised about how bad all of us are at math, science, AND history. We’re pretty good at unravelling the mystery tho.
- this is my teammate. but this is not my cat.
- I can’t figure out how to tie [these puzzles to KTaNE]. So there’s 3 options: A) I haven’t found the link yet B) The link is actually used in a supermeta C) Timwi made all these KTaNE links to mess with me specifically /j
- (Re Circles): Prior to us finishing the rest of the feeders, I noticed most teams getting stuck at 26 solves and assumed that one puzzle had a really weird answer that couldn't be guessed. By the time we got down to having just Circles unsolved, we still hadn't seen any weird answers, but we definitely knew The Nuke had an explicit list of words and phrases that could be used as answer lines. I looked at the list of options and confidently picked the one that made the least sense. It was right.
- (Re Home Depot): Pretty sure we didn't solve this the intended way. From the movements indicated by the arrows, I assumed the upper-left square had to be a letter extracted from either "vaccuum cleaner" or "rubber mallet". This was deducted from having "weather stripping" and "power fogger" in the bottom row, among several other conclusions. The edge squares were the least ambiguous. Used nutrimatic to extract the answer from this string of options:
[ua]. Not elegant, and probably not intended this way, but it worked.
[fbtcn] [osolo] [fwn] [ebdeu] [atsrwphvg] [atsphwvg] [raie] [wrwdm] A[emomegla] [raie] [rsr] [orrd] [npoic] [or]
- (Re Postcards, a puzzle made by Roger Wrightshoe; in reference to this image from the gallery): That’s Roger in the top left laughing at us.
- Spotted on a team’s spreadsheet for Face to Face:
- (Re Board State, an answer to the clue “Annoying person (4)”): yosh :^)
- (Re Objectionable Ranking): these are logic puzzles. there is a reason this team is called Team Illogical
- (Re Borders): u know what? i'm gonna give you 5 stars for this one because you tricked me, a person who hates logic puzzles enough to name myself team illogical, into doing a logic puzzle. and i might have even not completely hated it. well played
- (in a hint request for Objectionable Ranking): CAN WE HAVE A REFUND WE FINALLY GOT THE ANSWER YAY
- (hint exchange for Comebacks):
- (solver) En garde, Sword Master! EVERY WORD YOU SAY TO ME IS STUPID
- (hinter, not sure what this was about) I SHUDDER! I SHUDDER!
- (solver) I’ve extracted DO THE SWORDMASTER, which I thought meant throwing insults at y’all via the hint box. Did I miss a step, or do I need to check my work?
- (Re Board State, spotted in a spreadsheet mere seconds after a hint request):
OH MY GOD(same solver, in feedback): really enjoyed this despite having an absolute bruh moment (meta) wherein i stared at it for hours, wrote a hint request, and then read "sew fee ya" out loud... i am not very smart
I JUST READ OUT LOUD
- (Re Circles, submitted as a guess): IVE PRINTED THIS AND DREW EIGHTY ONE ARCS DO YOU KNOW HOW MUCH TIME THIS WAS
- (Re Totally Ordinary Crossword): Why "totally"?
- (Re Face to Face):
- (Re Postcards): How on god's flat earth are we supposed to make a Voronoi diagram
- (Re The 47, hint exchange; note these are three separate hint requests):
- (solver) is this a logic puzzle
- (solver) say only yes or no
- (hinter) No, it’s more of a math puzzle.
- (solver) even after the first step?
- (hinter) Yes, it is still not a logic puzzle.
- (Re The 47): We accidentally made music out of the result matrix for The 47! instead of reading the product matrix as index numbers into prefecture flags, we were expecting the matrix itself to spell something out (since all the numbers were small). We saw letters A through H, a structure clearly designed to be grouped by 8s, and so... we tried translating it into sheet music, using B for B-flat and H for B-natural in the German notation! We got this discordant mess...
- (Re The 47, hint request): The pog moment for me was when we saw SATOS from The 47 and immediately guessed BITCOIN.
- (Re Special Names): I was tabbing around, saw a list of names in our Special Names sheet, thought "Oh this is an Ultimate Showdown of Ultimate Destiny puzzle but it looks like it's well under way", and went elsewhere. Much later I went back and realized the person working on it didn't know that it was an USoUD puzzle!
- (Re The Nuke): On The Nuke, we thought SEPTUM was a pun for QUANTUM because the ending words of Big Bang episodes kind of relate with it (especially Quantum Entanglement), so we thought Quantum Scattering was a cryptic clue meaning anagram of QUANTUM, and we got UMAQTUN which means "to sleep" (we laughed because it was 1 am)
Apologies to the considerable number of teams who went through Face to Face smashing every single statue, hoping for something special to happen!
- I'm mostly curious to hear about the design decisions that departed a lot from other hunts, especially the leaderboard ranking and the lockout system. Tell us how you decided on those!
- I genuinely didn’t think the leaderboard ranking was unusual?
- We went back and forth on the answer rate-limiting system, but we didn’t have much time because the question only came up shortly before the hunt (I had totally forgotten to implement it sooner). Once the hunt started, we decided that we should hold all teams and players to the same standards, so we didn’t want to change it mid-hunt. We’re definitely going to relax it in any possible future hunt.
- WO IST LINIE U DREI
- DIE LINIE U DREI IST WEGEN BAUARBEITEN GESPERRT. SÄNK JU FOR TRÄVVELING WISS BEHFAUGEH.
- More seriously though: the fact that U3 isn’t used in the indexing step is a good way to indicate that the line numbers are not used for ordering.
- When did puzzle creation/writing begin?
- The earliest work on Board State (then called “Country flags+capitals puzzle”) technically started on 2019-04-07 at 12:26. However, QOPH itself (then called “Bomb disposal hunt”) was conceived later. Work on the hunt structure started 2019-08-04 at 05:47 and then I finished Board State to fit. The next earliest puzzle to be conceived was Year In, Year Out (then called “Games We Play”).
- The fact that there’s a Meta symbol in the final meta, when Meta only was announced something like October 2021, makes this hunt construction process seem relatively short haha
- Believe it or not, in the original conception the idea was for the final meta to be pure and for solvers to notice the commonalities between puzzles and create the chain all by themselves, so there were no icons.
- Were the locations chosen for any particular reason other than the need for the train station?
- I don’t even remember which came first: the idea of showing buildings with bombs in them just as a semblance of a story, or the idea of using the placement of two bombs as a commonality in the puzzle chain. Either way, at no point did we question the idea of having the buildings as a “story”. I’d like to say “now we know better” but I can’t; the only lesson I can draw from this is to not have any story at all, not even a semblance of one; but a lot of solvers complained that there wasn’t enough story.
- How was [puzzle] constructed?
- The majority of puzzles were constructed using algorithms. Some of the puzzle solution pages have a little more details.
- How was Face to Face constructed? Would you recommend using Unity for such a puzzle (if you had to do it over)?
- Don’t know about recommending it but I would use it again simply because I’m familiar with it due to KTANE modding. I have definitely considered trying to get even more familiar with Unity by writing a game of my own.
- Most people would likely use 3D modeling software like Blender for this, but I found it less frustrating and more fun to write my own code for 3D modeling, which I used extensively in KTANE modding, so it was only natural to use it for Face to Face, too. However, I used free models I found online for the box, the radio, and all of the Smash Bros characters.
- Which cyan numbers to include was chosen at random. I could have chosen any 24 of them to make a system of equations with a unique solution. In an earlier version I had all of them shown.
- The crossword subpuzzle was made by first deciding at random where to place bars, then using an algorithm to fill it with words. I initially tried to use Qxw for this, but in hindsight it was much easier to use my own PuzzleSolvers library to accommodate the weird geometry.
- Was there a reason why Face to Face was done as a visual game rather than just a text adventure?
- Making it a text adventure would have been more work for me. Plus, I learned a fair bit about Unity in the process, so I’m happy. Plus, by making it a point-and-click adventure game, it created an excuse for me to make a puzzle based on a point-and-click adventure game from my childhood to create the meta-meta chain (Comebacks).
- Was there a reason for having clues in Special Names, instead of just taking the letters in the X positions?
- (TasThiluna): I remember not being able to spell the cluephrase using only letters from the character names. It’s probably possible but I didn’t think to use an algorithm for it at the time.
- (Timwi): Personally I would have considered it a lamer puzzle without it (it would have had little to no novelty).
- Was there a reason for the black and white board in Board State?
- Yes, see the solution for The Nuke.
- Did you intentionally include BRAIN BOWL because it included RAINBOW to trip up solvers?
- No, that was a genuine coincidence. We failed to account for this possibility because we gave the Inside Out test solvers only the feeders they needed. By the time we performed full-hunt test solves, it was too late to change, but we also didn’t consider it a problem because the test solve teams coped with it fine.
- Why did the roundabout only have two exits?
- The other exits were bombed.
- How did you feel after creating Face to Face?
- So many questions about Face to Face. Well... it’s hard to pin-point a time that counts as “after creating Face to Face” because it went through so many revisions, and I did its individual parts (modeling, Unity programming, planning the subpuzzles, etc.) at very different times. However, I can’t deny I was very pleased with the result when I had the first version that I could actually click through and play.
- How to know when a puzzle needs a technological vs. a by-hand solution?
- This is a highly subjective question as it depends on what you’re good at and what you enjoy using.
- As a solver, obviously you first need to know what the puzzle even asks of you. Once I have that, I can think about whether there’s something I can do with either programming or a software I know well (such as Inkscape) that would help.
- As a constructor, I will generally gravitate towards algorithms because they can serve to automate a trial-and-error process that would be frustrating to do by hand. Unfortunately, in some puzzles, there’s a certain dimension of human intuition that algorithms cannot capture, so my algorithms sometimes run throughout the night and go through millions of possibilities before finding a suitable one.
- I also suck at deciding where to start when making a logic puzzle. For example, if you want to make a Battleship puzzle, where do you place the first battleship? It would just be completely random. This gives me choice paralysis. So I get an algorithm to spit out a couple random puzzles (with unique solutions), then I solve them by hand to see which ones have reasonable solve paths.
- When constructing, I do not try to find ways to “defeat automated solving”. I consider automation a legitimate strategy for hunt puzzles — which does, necessarily, give an advantage to teams that have a programmer and someone proficient with various kinds of software. In many puzzles, the mechanic itself already defeats standard solvers: you can use an automated solver on a standard Kakuro for instance, but the Fillomino in Objectionable Ranking is not unique by itself, and there aren’t really any Nonogram solvers that can do the digit encryption mechanic.
- If you were on the USS Enterprise (you can choose which one), what would be your position/job?
- Maybe school teacher? I don’t think the USS Enterprise has any need for me. I would most definitely not be a Starfleet officer as I do not fit into the idea of a command hierarchy. I need a job with a lot more flexibility than that. Plus I would most definitely not want to go on away missions.
- Who’s your favourite Olympic mascot?
- I don’t have a favourite, though that might change if they turn them into characters in a Pixar movie.