've wanted to lead a Magic set design since I won the (original) Great Designer Search more than six years ago. I arrived in Renton with a bag full of dreams and a suitcase full of ideas, ready to take on the world. It didn't take long for reality to sink in—designing Magic is hard. The six months of my internship flew by. I learned a lot, but most importantly, I learned that I had a lot more to learn before I would be qualified to lead a set.
Knowing that you don't know is one of the biggest steps along the path from novice to expert. After my six-month internship, I was still a novice—but at least I knew it.
Art by Karl Kopinski
I managed to secure a full-time position at Wizards of the Coast, but as a programmer; technically, I was supposed to be spending my days writing code, not writing rules text. Luckily, I was good enough at the whole "designing cards" thing that I snuck my way onto several more design and development teams during my stint as a coder. During this time, I was also proving myself worthy of greater and greater responsibilities within the Magic Online team. Each set I worked on brought me a step closer to being ready to lead a set design; each year that passed brought me closer to a world where I couldn't spare the time to work on one at all.
Luckily for me, I know how to work overtime.
When Mark Rosewater asked me if I would be willing to lead Dragon's Maze, I couldn't say "yes" fast enough. I knew that this chance might never come again—and I wasn't about to give up on a dream from six years earlier. I was being given a special opportunity to lead a set design without actually being in R&D—something that hasn't happened since the Stone Age of Magic, when sets were contracted out to external teams. Several people were taking a big gamble on my ability to successfully lead a Magic set and I had no intention of letting anyone down.
So I took my job very seriously. I started designing Dragon's Maze before we even sat down to design Return to Ravnica.
Start Before the Beginning
In some sense, this was the most intimidating part of my first set lead. When I envisioned leading a set, I expected that I would start with a nice, well-understood small set. I would be given lots of direction from the previous set(s) in the block. I might even have some interesting cards or even mechanics that had been pushed off. In short, I expected my first set lead to be about getting comfortable with the normal processes of leading a set and fulfilling well-understood expectations.
Instead, I got to start my first set lead by doing something that basically had never been done before. Return to Ravnica was the first block, to my knowledge, where R&D actually sat down in a room and drafted a mocked-up version of the entire block before anyone designed a single card for the first set. A lot of people weren't convinced that the large-large-small block structure with a five-five-ten guild distribution would play well, and we had to either prove them wrong or come up with an alternative proposal.
These early playtests for Dragon's Maze consisted of, basically, a bunch of random cards that I designed all by myself in one or two crunch sessions. Have you ever sat down and designed a full set in a couple days? I can't say that the cards were Great-Designer-quality work. My main goal was to capture what I felt was a reasonable structure for the set—the individual cards weren't really that important. It is interesting, though, to look back through those original playtests and find cards here and there that actually made it through design and development to see print. I suppose that when you make up 150 cards, you're bound to come up with a few keepers.
The team had also thrown together mockups of Return to Ravnica and Gatecrash, using mostly Ravnica-block cards but reorganized into groups of five guilds. We drafted this concoction, adjusted, and repeated until R&D was convinced that the overall block structure was sound. What did we get right from that early draft structure? What did we get wrong? How was I going to fit everything I needed into a small set? How was I going to do all ten guilds justice? How was I going to make my set feel like it had something new, and wasn't just "more Return to Ravnica" and "more Gatecrash?" Dragon's Maze was shaping up to be anything but "normal."
The biggest difficulty with early Dragon's Maze design was figuring out how to fit everything we wanted into, at the time, a 145-card set. We needed cards to represent all ten guilds and their mechanics, enough monocolored cards and mana fixing to support Limited, lots of cool multicolor cards, something interesting at rare, and something new and unique that wasn't present in the first two sets.
Looking back, I'm surprised we managed to squeeze everything into the set, but it took a lot of tricks to pull this off.
Deputy of Acquittals | Art by James Ryman
Every Guild Deserves Justice
As part of the large-large-small block structure, the team expected Dragon's Maze to deliver more of each guild. One of the biggest flaws of the original Ravnica block design was that all of a guild's cards were in a single set. This meant that if you were a fan of Selesnya, then all the cards you cared about were probably in Ravnica. You had nothing to really look forward to in Guildpact or Dissension. Dragon's Maze was going to solve that problem.
This seems like it would make set design simple. I would have ten mechanics to build on, and a large portion of my set would be defined by a well-understood guild structure. This was where I discovered that "well-defined expectations" was not the same as "simple."
The problem wasn't one of expectations, but one of space. There isn't a lot of room in a small set. If you take a normal 145-card set and divide it into ten guilds, you get, at most, fourteen cards per guild. Once you account for unguilded monocolored cards, you'd be lucky to get ten cards per guild. Those cards need to count. I needed to fit the entire payoff for each guild into about ten cards. Some of the hardest parts of leading this set involved constantly having to cut guild cards that were awesome so we could fit in other guild cards that were awesome. Something I learned quickly as a lead designer: you simply can't fit everything in. Don't get attached to a card or a cycle—you'll inevitably end up forcing yourself to cut it after too many agonizing attempts to spare it.
Unique, Just Like Every Other Set
Part of leading a set is discovering ways to make your set unique. For my first lead, I was being handed a structure that included ten guilds, ten keywords, and a number of other nonnegotiable needs. My set had an identity crisis—how was it going to feel unique? What was going to make my set distinct from its two parent sets?
Looking back, this part probably consumed more mental effort than any other part of designing Dragon's Maze. I was constantly looking for ways to squeeze in a bit more "identity." I was also constantly frustrated as each attempt was either soundly rejected by playtesting or quietly integrated into the set design, with no trace of uniqueness visible from the outside. A lot of these changes resulted in a more interesting Draft format, but I wanted something loud—something that screamed out to players that they should buy this set.
My earliest drafts had toyed with split cards as a way to sneak more multicolor cards in that could also serve as monocolored cards. Return to Ravnica and Gatecrash used hybrid cards for a similar purpose, but early in design, it was my job to try things that were more "off the beaten path." We knew we could always return to hybrid cards if other solutions didn't work out.
Adding split cards had conveniently led to a potential solution for "what's new in Dragon's Maze." We wanted something that didn't really "feel" like a keyword, because we already had ten keywords for the guild mechanics. The "new thing" also had to pull its own weight and fit naturally into the set—it couldn't just be a random eleventh mechanic.
Fuse seemed to answer all of the right questions. It was added during structural design and didn't change throughout design and development. Fuse was doing a lot of good work at uncommon, giving us cards that could serve as both multicolor and monocolored cards, but it didn't quite pack enough punch to justify a new mechanic. There needed to be something bigger and better we could do with fuse than utilitarian uncommons. So we took things up a notch.
Fuse cards at rare served a completely different purpose than those at uncommon. At rare, they could be splashy and exciting. They could cost a pile of mana and create exciting stories when someone managed to cast both halves. Casting Beck or Call will often be a decent use of a card, but casting Beck and Call? Eight mana may be a lot to ask for, but drawing four cards and getting four tokens is going to have a huge impact on the game. Although we weren't sure at the time, we had found our eleventh mechanic.
Broken Mana, Needs Fixing
Normal sets don't usually have to worry too much about mana fixing. Sure, there will be a handful of cards, possibly a cycle of artifacts or lands, but these cards rarely define the structure of the set. But when you're creating sets with a focus on multicolor, you can't afford to ignore mana fixing.
We knew from Day One that we'd need, at a minimum, ten mana fixers at common to support the multicolor drafting that would emerge as part of full-block drafting. Over the course of the set's design, we tried four or five variations on mana-fixing commons. This process got increasingly frustrating, as each design was found to be interesting, exciting, unique, and... not common. Or not balanced. Or not good enough to put in your deck. Mana fixing doesn't really help if players aren't willing to play the cards.
This is a part of design that I simply was unable to solve to my satisfaction—I handed the set over to development with placeholder mana fixing, deferring to their expertise on what was, arguably, a development-type problem. I was somewhat surprised when I learned development had settled on the same cycle of commons that we had in one of our earliest drafts. Sometimes the obvious, simple idea really does turn out to be the best.
Even though the actual cards were always in flux, the team at least knew we had to have that cycle of commons. But we also knew that it wasn't enough. We had tried several other ways to add additional mana fixing into the set, but late in design, the team was still frustrated at the underabundance of options available when building a draft or sealed deck. No one wanted to spend another ten slots on mana fixing, and even if we did, we were having enough trouble settling on one cycle of mana fixers. Trying to design two cycles sounded impossible.
I believe it was Erik Lauer who put both of these limitations together and came up with a shockingly elegant solution. We already had simple, nearly perfect mana fixers in the form of the Gates. Why not reprint those? Since they were reprints, we didn't need to count them as part of our quota of commons. To emphasize that these were "bonus" cards, we could even replace the basic land slot—players wouldn't even have to give up one of their commons. It was just crazy enough to work.
For most of design, I thought the lesson I was learning here was that, sometimes, a problem just can't be solved, and you just need to work around it. I was pleasantly surprised to also be taught a second lesson: sometimes, even your most basic assumptions need to be challenged.
Art by Michael C. Hayes
Back to Basics
On one hand, I'm pretty proud that I managed to get several key pieces of the third set's structure right before we had done any serious work on the first two sets. But if the structure was sound, what exactly would we do to fill months of design time? There were still cards to design, but I had proven that I could hammer out a full set's worth of cards in under a week. Piece of cake! What could we possibly need so much time for?
Early during Dragon's Maze design, Mark Rosewater sat me down and shared with me one of the greatest secrets to set design:
All new lead designers, according to Mark, make the same mistake. They don't playtest enough. I was already playtesting, but I know how to take a hint. So we playtested some more! During early design, as soon as we had a reasonable set of commons, the team playtested at least once a week. Once the commons and uncommons were nailed down, we usually playtested twice or more a week, with adjustments to the set simply being handled through email.
The process of "playtest, change some cards, and repeat" isn't particularly earth-shattering. There's no trick to it, the team just needs to put in lots of hours playing with the cards—changing a line of text here, replacing a card there—until each card has become the best card that it can possibly be. Then the set is handed over to development, and the process just keeps going.
I'm skimming over the "change some cards" part a bit. Sometimes, I threw out 50% of the cards and replaced them wholesale with new designs. Sometimes, it was simply lots of small tweaks to try and nudge the set's focus in a given direction. The entire feel of a playtest can change from moving some key commons to uncommon or vice-versa, a process that happened several times during design of Dragon's Maze.
As a rookie lead, the temptation was great to sit in a room and debate design all day long, but it's important to remember that Magic is a game. If you're not playing the game, you're not actually designing Magic. You're just making noise.
Looking back at my first design lead, I'm happy with how everything turned out. I'm also happy with how I turned out—the entire experience taught me a lot of useful things. Some of them are fairly specific to game design, and some of them have broader applications. I hope you enjoy the set as much as I enjoyed making it.
I think I'm going to go playtest Dragon's Maze some more now.
Alexis Janson is the winner of the original Great Designer Search. After a six-month internship, she moved on to the technology team at Wizards of the Coast but continued to moonlight as a contributing member of Magic R&D. She currently works as a developer on the Magic Online client team.