Puzzle Potluck 3 Wrap-Up

Warning! The wrap-up contains spoilers for the Puzzle Potluck 3 story and puzzles.

Congratulations to the 266 teams that helped defeat Thanos! Special congratulations to Aviation Laws for being the first team to finish, on Saturday 6/20 at 1:12pm PDT, a little over 4 hours after the hunt started. 39 other teams finished within the first day (24 hours) of the hunt!

Thank you so much to everyone who participated in Puzzle Potluck 3. This was a way bigger turnout than we expected and we’re glad you were able to join for our most ambitious $puzzlehunt crossover!

We were really happy to see that there were 1048 teams that signed up, 582 teams that solved at least one puzzle, 438 teams that finished Trivial Pursuit, and 266 teams that finished the entire hunt (and 189 teams that found the post-credits scene!). Those numbers beat out Puzzle Potluck 2 across the board, so thank you everyone :)

The next few sections of the page will cover our decisions while writing the hunt and some reflection now that it’s over. You can also find some interesting commentary in the author’s notes in the solutions. Feel free to jump straight to the fun stats and stories at the bottom if you’d like!

Goals

Puzzle Potluck continues to have the everlasting goal of providing a hunt that would be fun for all puzzlers, regardless of their experience level, but this year we had a few other goals in mind.

This was the first hunt we wrote together as a team, and we saw the opportunity to create a hunt with a much stronger story, theme and cohesion between puzzles. We think we really succeeded on this front!

We also wanted to keep the length and puzzle difficulty of the hunt roughly the same as Puzzle Potluck 2. As we developed the structure of the hunt it became clear that we would have significantly more puzzles, so we aimed for keeping the length the same with a lower average difficulty of the puzzles.

Another goal for this year was presenting a broader variety of puzzles, including interactive puzzles. While we missed out on creating a logic-based puzzle this year, we succeeded in branching out our puzzle types and were also glad that the interactive puzzles made online collaboration that much more enjoyable.

Finally, we made a big push towards the end to make sure our puzzles were cross-browser, printer, and copy-paste friendly. Hopefully you felt the difference!

Format

In our last hunt we said that Puzzle Potluck was a one round hunt and we planned to keep it that way. Well, clearly that changed. The idea for having an Avengers theme and snap twist came at the end of Puzzle Potluck 2, but wasn’t considered seriously until we started working on Puzzle Potluck 3 in August. There was some debate over deviating from the one round structure but the idea was too good to pass up. We decided that the snap mechanic was consistent with this vision, since the same “round” of puzzles is involved in both rounds. Of course, we then added the Avengers puzzles and the time machine mini-puzzles, but the single-round soul of Puzzle Potluck remains. We want Puzzle Potluck to remain a hunt that can be completed over a weekend, and we definitely hit the upper limit of content for such a format.

We kept the same track and hint structure from Puzzle Potluck 2, but there was some feedback from last year that our hints were often straight giveaways. We tried pulling back on that this year but we probably went a little too far back, as several teams found this year’s hints too cryptic to make progress. We also added partial answer confirmations, which we were happy to see were helpful to many teams.

This was also the first year we introduced a (much requested) live leaderboard, as many other online hunts have. One thing we explicitly did not want to show in doing this was how many teams had solved each puzzle, as we didn’t want teams to be scared away from attempting a low solve-rate puzzle.

Process

Since this was the first Puzzle Potluck that we wrote as a full team with the experience under our belt from running the first two, we introduced a much more rigorous process around puzzle writing to make sure that every puzzle was solid.

Each puzzle idea was first vetted by Rajeev to make sure that it was interesting enough and didn’t overlap with other puzzle ideas. Once a puzzle draft was completed, it was test-solved multiple times within the Puzzle Potluck Crew. In April and May we had two teams do full test-solves of the hunt on the website. We used the feedback from each test-solve to make puzzle revisions and improve the full experience on the website.

It was fun for us to look back on some of the early puzzle drafts and funny moments over the past year, so we compiled some of them in this short presentation:


Story

Summary

The hunt opened with a man “in a rather large suit,” seeking help with Trivial Pursuit. Someone replaced all of his trivia questions with puzzles, and teams had to help him solve the puzzles, collect all six wedges, and answer the final question. Trivial enough, right?

But after collecting the wedges and answering the final question, the man revealed his true intentions. The wedges were in fact the Infinity Stones, and with one swift motion, Thanos snapped away half of the universe!

Teams then received a transmission from Tony Stark. As it turned out, the puzzles were set up by the Avengers to prevent Thanos from achieving his goal. Luckily, the six original Avengers survived the snap, and they then recruited the teams to retrieve the Infinity Stones and reverse the snap.

Each of the original puzzles was also snapped away, leaving behind a portion of the original puzzle. Teams needed to figure out each of the Avengers’ special abilities to solve the “snapped” puzzles. Solving a “snapped” puzzle rewarded the team with a clue for retrieving the corresponding Infinity Stone.

After solving all of the “snapped” puzzles, and with the help of some “expert teams,” they were able to assemble the time machine and go back in time to retrieve each Infinity Stone. Teams traveled to six different points of time in the past with each of the Avengers, and after solving the puzzle they found there, were rewarded with the Infinity Stone. Once teams gathered all of the stones, they placed them in their respective places in the gauntlet, found a way to “snap” the gauntlet, and successfully defeated Thanos!

Lindsey's Notes

Discerning devotees of Puzzle Potluck may realize that Potluck 2 opened with a story about Layton that turned into a board game. On the other hand, Potluck 3 opened with a story about a board game. At this rate, Potluck 4 is going to start with a game of Monopoly and each of the railroads will be a separate game of Ticket to Ride. And Electric Company is Power Grid.

Development of Puzzle Potluck 3 started right after we wrapped up Potluck 2 in summer 2019, with the observation that the six Trivial Pursuit wedge colors (in certain iterations) roughly match up with the Infinity Stone colors. From there, the group considered the question: can we possibly structure our hunt so that each of our second-round puzzles is embedded in first-round puzzles? It took a lot of work, but the short answer is yes. See infra, “The Snap.”

Aside from the plot twist, my favorite part of writing the story was playing with characters’ voices, and I certainly had a lot of characters to work with. I hope you enjoyed reading my glorified fanfiction, and I also hope you enjoyed the shawarma :)

The Snap

The centerpiece of this hunt was the “snap” mechanic. We hadn’t seen it done before (and we are not planning to do it again :P) so we came up with some guidelines that would ensure that we did it justice. (There have been mechanics in the past that were similar in nature, for example The White Queen round from 2014 MIT Mystery Hunt or the excellent standalone puzzle Middle of the Road, which served as inspiration that something like this could be done.)

  • The mechanic of the “snapped” puzzle should be different than the mechanic of the full puzzle so that it doesn’t feel repetitive
  • Cover a variety of puzzle types (word, grid, interactive; we also wanted an image puzzle but we kind of covered that with Julz’s awesome homepage artwork)
  • Vary the amount of each puzzle that disappears
  • Embed the titles of the “snapped” puzzles in the titles of the Trivial Pursuit puzzles
  • Additionally theme the Trivial Pursuit puzzles on the corresponding Infinity Stone (e.g. History + Mind was a puzzle about philosophers)

It was pretty tricky to write these pairs of puzzles. First, the author had to come up with a “snapped” puzzle idea as well as the corresponding Avenger ability. The Trivial Pursuit puzzle could only be written after the “snapped” puzzle was test-solved and finalized.

Writing the Trivial Pursuit puzzles was especially challenging because they were so constrained, and they also needed to be simple because they were in the intro round. We tried our best to make them solvable despite the extra information in them, and we also greatly simplified the first metapuzzle to allow for more flexibility in the round 1 puzzle answers.

Originally the “snapped” puzzle pairs were intended to be the only puzzles in the hunt, but we decided to add the more free-form Avenger puzzles since we weren’t sure how good these puzzles would turn out and we wanted to ensure that we at least had some good-quality puzzles. Based on the feedback we got on individual puzzles, it does seem like people enjoyed the Avenger puzzles!

We intended for the “snapped” puzzles to be impossible if not very difficult to solve without the Avenger power. Despite this, at least 14 teams solved Space, at least 10 teams solved Soul, at least 5 teams solved Power, and at least 1 team solved Mind before obtaining the corresponding Avenger power and without hints!

The final big thematic piece to the snap was the end of the hunt. We wanted teams to get the experience of a final victory lap, collecting the stones, placing the stones in the gauntlet, and “snapping” the gauntlet to defeat Thanos, just as it happened in the movies. We were very impressed with teams that were able to snap the gauntlet with just 5, or even 4, of the stones!

Future

Wow! How do we top this one?

There’s still a lot we can do in terms of making the hunt more accessible to less experienced solvers, as well as tightening up our puzzle construction and mechanics, making more interesting metapuzzles, and getting a little more creative at the individual puzzle level. This time we really elevated our story, theme, and structure. Next time around we want to get closer to that sweet spot of having a creative overall structure while still allowing us to write better individual puzzles without too many constraints.

We’re so glad you joined us for Puzzle Potluck 3 and hope to see you all again for Puzzle Potluck 4!

Credits

Puzzle Crew: Bradley Wu, Curtis Liu, Darren Yin, Julz Huang, Lindsey Shi, Rajeev Nayak, Stephanie Chang
Editor Crew: Rajeev Nayak
Story Crew: Lindsey Shi
Artwork Crew: Julz Huang
Test-solving Crew: Amy Li, Cyril Lan, Dabin Choe, David Wong, Dillon Zhang, Landon Carter, Norman Cao, Stephanie Yu, Victor Hung, Yaning Zhang, Yanping Chen
Website Crew: Curtis Liu, Julz Huang, Rajeev Nayak
Thank you from the Puzzle Potluck Crew!

Fun Stuff

Leaderboard Race

Here's a graph of the top 20 teams' effective puzzle solves over time. When a team solved Trivial Pursuit, they automatically got credit in the graph for solving all first round puzzles. When a team solved Endgame, they got credit for solving all first and second round puzzles.

Highest Accuracy

We were impressed by a number of teams who finished the hunt with very, very few incorrect guesses:

  • synod (1 + 2, SPIDER for Battle of Copenhagen and 2 regarding Snapchat: DOWESERIOUSLYHAVETOINSTALLSNAPCHATTOFINISHTHISHUNT, SNAPCHAT)
  • ReD'oh! (1, HMSELEPHANT for Battle of Copenhagen)
  • potato (1, NUMBERTEN for Their Finest Hour Speech)

Guess Log

For your enjoyment and amusement, here is the full guess log for the entire hunt.

The Sacrifice

We decided to include an interaction this year where teams had to sacrifice a member to obtain the Soul Stone. It was really fun for us to see what you came up with and send you some fun responses (although sadly we had to turn on our auto-responder after the first weekend). Here’s a thorough analysis of everyone’s sacrifices:


Typing of the Dead Stats

We collected stats for A Not So Leisurely Sport!

  • Sautee Mushrooms won the game the most times (16)!
  • 453 games were won out of 22,259 games
  • 2.15M characters typed, 94.9K mistakes made, for a combined accuracy of 95.6%
  • Fastest fingers award goes to Fig Newton at 5.224 characters per second (for games won typing for longer than 60 seconds)
  • Sharpshooters award goes to clairet for typing at 100% accuracy for the longest period of time (193 seconds, at a speed of 2.749 characters per second)

Unfun Website Problems

A brief postmortem on the website issues at the start of the hunt. We use Google App Engine for our website, and the night before the hunt began, out of concern for the number of teams that joined, we removed the auto-scaling instance limit. That ended up being a big mistake, as our database allowed a limited number of connections, so as servers spun up, they could not connect to the database, which caused more servers to spin up. We reverted the change to limit the total number of instances, which mostly resolved the issue. Lesson learned for next time!

Funny Stories

We compiled our favorite stories from teams (and ourselves) from the hunt. Here they are, with occasional commentary from us in italics:

The Puzzle Potluck Crew

There were a fair amount of teams that got stuck on Time, but this email had our favorite title.

🐿integirls.org/cowpeabara🐿

We were soooo stuck on Iron Man. We had finished the crossword, found the tee, the ball, and a putter (we don't know if that was intentional)... We tried so many things like looking for a driver or another type of club to the left in the direction that the putter hit it.
This was a total throwback to our first test-solve, where the team found some... very questionable clubs.

Dogs Bound By Rules

I'm not sure if you adapted the snap effect in the second story segment from the Thanos easter egg in Google search -- from digging through your code, it seems like you might have. I was actually the one who designed the algorithm for that effect. I've been happy to see people around the world reverse engineering it since it launched!
I (Rajeev) did try to model our snap effect after the Google one, but I couldn't find any perfect reproductions online, so I had to take an imperfect one and modify it myself. I remember wishing I could just ask the person who made the original one!

hoME AT last

We decoded fragmented QR code by hand. In case you're wondering how useful each piece is, it's SPACE = POWER > TIME > MIND > REALITY = SOUL = 0.

Multiple teams

idk
Hastings
In case anyone needed help visualizing Iron Man playing golf :)

Pan-Galactic Goose Syndicate

Stoichiometry 101 - We unfortunately messed up Jeopardy question (didn't think to add 2), and the feedback we got back was giving us a mixed message (did we get it right or wrong?!?), so I ended up reverse engineering the entire puzzle. First, a teammate backsolved the answer from the first round meta, which was easy enough. From there, I deduced that it was probably spelled with atomic symbols. Then the hard part came: trying to figure out the variables from the atomic numbers. The break-in was with the second expression: we have two integers multiplied which results in 39^2, and there are only a few possibilities. We already had all the other coefficients, so this meant that we knew the ratio between H and K. So we can deduce that the ceiling of H*H is 3 and the ceiling of K is 3*13^2. From here, I was able to fiddle around with the value so that the other expressions satisfy most of the atomic numbers we need. Once we had the variables within reasonable range. I went back to the Jeopardy coefficient and noticed that our wrong answer was off by a factor of 2. That meant we had to write 4 words instead of 2 words. From there we finally realized that we have to add "What is", and finally understood what the feedback was telling us. Definitely not intended, but I enjoyed the journey!
Wow! Definitely not intended!

The Puzzle Potluck Crew

We didn't quite expect to see teams sacrificing their pets, so the first team that sacrificed their dog (the Bowser Bunch) prompted Lindsey to first send an email response with just this video, followed by the email with the actual answer one minute later. Within that minute, they attempted to solve the Soul puzzle with THISISUNACCEPTABLE.