Hearthstone's Celestalon has provided some great insight on the creation of the Barrens Mysteries achievement and puzzles, an event that brought the community together to solve four mysteries to unlock a sweet new card back that came from China's card back design competition.

We've recapped the details if you'd prefer to not read through all the tweets, but we encourage you to do so because its a great read!

  • The Hearthstone Devs do a "Free Your Mind" week each year in November or December.
    • Team members can work on anything Hearthstone that week.
    • Celestalon decided on his Free Your Mind week to build a secret puzzle.
  • Celestalon has previously worked on puzzles much like this one within the realm of World of Warcraft.
  • The only people on the Hearthstone team that knew about the puzzle's solution were Celestalon and two members of their QA team (Mary & Eric).
  • There was supposed to be a different reward but it fell through. It may still come out in the future.
  • Having the flavor text riddles come out in a patch earlier than the puzzles let it not look suspicious.
  • If Celestalon could go back, he'd have made the Rogue puzzle cheaper to participate in. He didn't want collections to be a limitation.
  • Celestalon is unsure if there will be more puzzles as they take time to create, but he loved to make them.
Puzzle #1 - Rogue Combination Lock
  • Slightly inspired by the locked chests in World of Warcraft's Torghast.
  • The reason why 4 was selected as the number is that there is no reason.
  • It was simply the solution of the first random combination he liked the look of.
  • Here is the spreadsheet used to create the puzzle.
Puzzle #2 - Druid Maze
  • Inspired by a puzzle in World of Warcraft called "The Endless Halls" which is also a maze.
  • Potentially could have had a 5D space, offset overlapping edges, trap rooms, and more complexity. Celestalon went easy on the first maze.
  • The technical implementation of this in Hearthstone was a huge challenge.
  • Here is the spreadsheet used to create the puzzle.
Puzzle #3 - Warrior Caravan
  • Celestalon likes river crossing puzzles and loved the Barrens flavor that could be applied to one.
  • Providing an easier puzzle that could be solved by an individual and could give the community easy initial progress was something he liked.
Puzzle #4 - Hunter Bazaar
  • Difficult to design and develop because the solution needed to be seriously hard.
  • The initial implementation, before redesigning, would have required 6 hours to input into Hearthstone.
  • He was able to get the solution down to 10-12 minutes after redesigning. Still a bit long but acceptable.
  • Here are the spreadsheets (one and two) used to create the puzzle.

Dev Transcript

Quote From Celestalon

Thread about the making of the #BarrensMysteries! This was a ton of fun, both to create, and to watch the community solve. I've seen requests for hearing more about how this crazy thing happened, so here's a (probably long and rambling and unorganized) thread about that! https://t.co/c2PxiOG8PZ (Source)

Every year (usually Nov/Dec), Team 5 spends a couple weeks doing something we call 'Free Your Mind'. It's a fancy way of saying 'work on whatever you want within the world of Hearthstone'. Last year, I chose to work on building secret puzzles. (Source)

I had worked on these before on WoW, along with the puzzlemaster @Muffinus. Parts of the Lucid Nightmare and most of the Hivemind were my work, and I *loved* how the community would come together to solve these brutally difficult puzzles. (Source)

But how could I bring something like that to Hearthstone? Obviously, WoW is a much more open-ended game than Hearthstone. You can go everywhere, do all sorts of things, and things could be hidden anywhere. Thus, even getting something into Hearthstone was a challenge. (Source)

I also wanted to be efficient about this; the more I could do of it myself without needing to ask for others' time, the more likely it'd actually ship. Additionally, usually when designing new things on Team 5, we are super open and collaborative about things. Not here though. (Source)

As Muffinus and I learned making WoW puzzles, spoilers totally spoil the fun, and the more people involved, the more likely some little tidbit is to accidentally get leaked. So this was one of the very few parts of development where secrecy was paramount. (Source)

Outside of myself, and 2 of our awesome QA (thanks Mary and Eric!), not a soul got to see the whole thing. A few designers helped playtest 1-2 puzzles each, without knowing how they fit into the full thing. Most barely knew it existed. Anyway, let's talk about the puzzle design! (Source)

Oh, one other thing first: The reward! We didn't actually know what the reward was going to be until very late in the process. We had something else lined up, but due to the random realities of game development, that fell through (which I still hope to deliver someday!) (Source)

Then the cardback contest art came in, and that was just great! It wasn't thematically matching the puzzles or setting, but that was made up for by how epic looking it was. And it served as a bit of a red herring (which are VERY IMPORTANT in these sort of puzzles), too! (Source)

PUZZLE #0: The Riddles I had come up with drafts of a couple of the puzzles, but needed some way to get people into them in the first place. What I came up with was triggering them via a specific series of cards (or possibly deck code), in Practice mode. (Source)

Thus, the riddles placed in flavor text were born. The puzzles weren't quite ready until 20.2, but I got to get the riddles into the flavor text in 20.0, so they didn't get immediately called out as something 'new in 20.2 so probably related', which would have made it too easy. (Source)

That brings me to an important point about difficulty. These puzzles are designed for an entire COMMUNITY to solve, not just an individual. They have to be *ridiculously* hard to solve. Thousands of people collaborating will notice every subtlety, try every possibility. (Source)

I always have this moment, just after something like this launches, where I think to myself, "Wait, I made this too hard. They have to notice these flavor texts, correctly guess that they need to connect phrases to cards, correctly guess the cards, decide to play those cards… (Source)

…in Practice mode, all in one turn, without any other cards. Omg, that's SO many hoops with no hint that they're on the right track. There's no way they'll find that." But then I remember that I *always* have that feeling, yet they solve it. (Source)

A surprising amount of the time, the global aspect of this community becomes the key to solving some riddles, because things may be more or less subtle in different languages, or be suspiciously similar or dissimilar in different languages. (Source)

In this case, people noticed the Traditional Chinese flavor text for the 4 relevant cards were more blatant about the cards required. As soon as someone noticed that, everyone was off to the races and the hunt transitioned from the 'try things wildly' to 'we have a lead!' phases. (Source)

I also know that there are many different kinds of puzzle solvers. Some people like word puzzles, some like math puzzles, some like logic puzzles, etc. So I like adding some parallelization, so people can work on whichever suits their fancy. It also let some be easier, some hard. (Source)

One revision I'd make here if I could go back in time would be to choose cards for the Rogue riddle that were a little bit cheaper rarity. Overall, I didn't want your collection to be a limiter on being able to do the puzzle (within reason anyway), and the Rogue one pushed it. (Source)

Puzzle #1: The Cubes This was the number puzzle of the bunch. I wanted to do some sort of combination lock. I was a bit inspired by the locked chests in Torghast in WoW, but wanted to take that to an extreme level. More tumblers, different effects of each tumbler, etc. https://t.co/SEtBbHaYbX (Source)

As with most number puzzles, I knew that someone writing a script/program/spreadsheet would probably be the most likely solution, so I was sure to throw in a number of red herrings to at least slow them down a little: the 3 keys, and them changing attack. Plus, an unclear goal. (Source)

As for the solution, people have asked "why 4?". The goal I had in mind was just to get them all the same, 4 just happened to be the solution of the first random combination I liked the look of. No special meaning there, sorry. :) (Source)

I also honestly wasn't sure if there were multiple solutions. I didn't think so, but that was really hard for me to prove. Turns out, there are! The solution the community found was different than the one I reverse engineered the puzzle from! BUT, it also ended in 4. https://t.co/Sh5GSoS6t0 (Source)

Puzzle #2: The Maze I *love* mazes. The Endless Halls is probably the most notorious example of this from WoW. But how could I possibly put a *maze* in *Hearthstone*?! I decided I HAD to find a way. (Source)

(Oh, I called it the "Doors" in secretfinder discord, because I didn't want to give away that it was a maze too quick.) (Source)

So yeah, I put you in an imaginary maze, with controls to turn left, turn right, and move forward, and an indicator of whether there was a wall in front of you or not. That was enough to be a maze, and you could meticulously explore it and map the place out. (Source)

I pondered adding 5D space, offset overlapping edges, trap rooms, and more complex victory requirements, but I figured I'd go easy on the Hearthstone community for their first maze. 😉😉😉 I did add limited lefts you can make, for a bit more challenge though. (Source)

As you might imagine, the Hearthstone engine isn't exactly tailor-made for this sort of thing, so the scripts involved are… quite something. Just the technical implementation of this was a huge challenge. https://t.co/efry7w8h4y (Source)

Puzzle #3: The Caravan Here's the logic puzzle. I like river crossing puzzles, and I had the idea of doing a river crossing with the 2 sides of the board. This was what I came up with, when limited to the minion slots on our board. (Source)

I also loved that this one had a ton of room for really flavoring it up to match the Barrens. - Could make the boat a caravan. - Wolves could be Quilboar. - Orc could be Sheep. - Centaur could attack the caravan in cases the crossing was invalid (easy pickin's or too juicy). (Source)

I knew that given the small number of moves involved in the solution, this would be solved relatively quick. I liked including a puzzle that was feasibly solvable by individuals, and also gave the community some progress (1/4, woo!) soon after the riddles. https://t.co/oW6B42HEA8 (Source)

Puzzle #4: The Bazaar Hooo-boy. This was very difficult, both to solve, and develop. I wanted there to be a seriously hard one in there, that would take the community some significant time to solve. I wanted something that wouldn't be solved by a quick script/program. (Source)

I thought I could probably do that with a trading minigame, using a variety of commodities that you exchanged back and forth. I got to get a bit of flavor in here, with Merchants (sellers, top row) and Adventurers (buyers, bottom row). (Source)

I went around and around on the mechanics of this puzzle for quite a while. Ultimately, I ended up redoing quite a bit of it, twice. First time, because I realized that my initial implementation would have required 6hr of game time to input the solution. (Source)

(Yes, the original version was EVEN LONGER. You could buy items one at a time.) The solution to that problem was actually an ah-ha moment, as I realized I could make it less tedious, AND way harder, by just forcing you to always max-buy everything. (Source)

The second time I redid it, I realized I made a dumb mistake early in my solution series (which I did all manually, BTW), which invalidated everything after it, thanks to my new max-buy rule. Huge thanks to Eric for his testing and patience as I iterated on this multiple times! (Source)

My spreadsheet for this one is… uhh… a little more involved than the others. It's also quite the mess, due to the multiple redos. Not sure if Twitter will butcher this due to the size… https://t.co/0hbJn7jrOz (Source)

Yup, butchered. OK, try this: https://t.co/uPAx0hWU1D (Source)

And here is the nicely formatted solution that we used internally. I think it differs slightly from the community found solution. (Full version, and a 'lite' version that just has the action you need to take for quickest reference.) https://t.co/CDIN0metb4 (Source)

Using that lite version, Eric and I were able get it down to about 10-12min to input the solution. Still a little longer than I'd have liked overall, but acceptable given the difficulty I wanted to achieve here. (Source)

And that's about it for the puzzles! One question I've seen asked a bunch is… Will there be more of this in the future?! I honestly don't know. It takes time to do, and I don't know when I'll have time to again, but I'd really love to. (Source)

And I think that about wraps up for this behind-the-scenes episode. I'm gunna make a separate thread to do some Q&A about it (this thread is long enough as is). (Source)


@Celestalon This Spreadsheet is awesome, thanks for the great puzzle! I was Storm on Discord btw. I started working on a script to solve it, and got the puzzle represented programmatically, but a solution came in before I fine-tuned a search algorithm. Great fun!

@StormWindBoy Aye, I listened to you a lot, and enjoyed the memes about you! :D I think I recognize your name from the first google doc of theories and things that had been tried; you were also one of the first to theorize about the flavor texts, back on day 1, I think? (Source)

@StormWindBoy We enjoyed watching that doc, and the discord chat. I remember screenshotting this at the time, saying "OMG THEY'RE DOING IT!" https://t.co/0FTveGAaNF (Source)


@Celestalon This is absurd

@RidiculousHat @Celestalon Absurd doesn't even begin to describe whatever that is. O_O

@dan0play @RidiculousHat Thank you! (Source)


@Celestalon Whoah whoah whoah This took only ONE WEEK to make?!?!

@bbrode @Celestalon The power of @Celestalon with no real approval loops to go through (Source)

@IksarHS @Celestalon I feel like if you told me to make an excel sheet as big as that one he posted and just fill it with literally gibberish, that would take me longer than a week right there

@bbrode @IksarHS Heh, passion does crazy things for me. :D (Source)


@IksarHS @bbrode To say that I appreciate the trust that you put in me on this on me is an understatement. :D (And to be realistic, it was probably about 2.5wk of my time, and 1wk across other people, including the revision and polish.) (Source)

@Celestalon @bbrode Sometimes the best thing to do is just get out of the way 😁 (Source)