What transpired between those two endpoints was surprisingly complex—even with this specific vision, Scars of Mirrodin had more iteration and dead ends than any other design team I've been on. What follows is my attempt to describe just how much work goes into a set that never makes it onto a single printed card.
Returning to an existing block presented us a puzzle. We knew what Mirrodin was, and we knew what represented it mechanically. We wanted to maintain that familiar feel as much as possible.
Our first order of business was looking at the existing elements from Mirrodin block. Artifacts and some sort of "artifacts matter" cards were the obvious inclusions. Equipment and indestructible would likely show up, but these had become evergreen and carried little weight here. Scry had just been reprinted in Magic 2011, and entwine and sunburst were too unrelated to the core Mirrodin flavor we were aiming for. That didn't leave a lot of existing space to explore, but we dutifully picked through it all.
Artifacts that involve color: This was the first design space to be mined. Fortunately for this article, Shards of Alara had already used one of the most simple and interesting evolutions available to us—artifacts with colored mana cost. Had Esper not existed, our journey might've ended here. Without an elegant way to reinvent this space, we decided it was better to avoid it for now.
Without a "Get Out Of Jail Free" card, we had to get to work. We toyed with artifacts that cared about colored mana; different ways of playing with colored mana in your mana pool; Crystal Shard variations; Spellbomb variations; and many other "color matters" artifact mechanics.
Even though we tried a dozen new mechanics in this space, we ultimately settled on simply using colored activations, with a small tweak on a new cycle of Spellbombs. Everything else felt too mechanical or complicated, and nothing was fun enough to make up for it.
Affinity: Seeing as how this worked out so perfectly in the last Mirrodin block, we figured there was nothing left to explore here and moved on. I had a great paragraph explaining why we skipped affinity, but like everything else related to affinity it got banned.
(Note: MaRo's article challenges this timeline and says affinity came before imprint, but this is wrong—we didn't seriously put affinity into the file until late.)
Myr: Our earliest Scars of Mirrodin design had Myr. Actually, it had small artifact creatures that happened to say "Myr" on the typeline. Since playtest stickers don't have art, I couldn't really be sure whether they were real Myr, or if it was just some sort of myrsquerade. It wasn't until later that our Myr gained some identity.
Fairly early in design, we needed more artifacts that would stick around and help enable the various "artifacts matter" cards and mechanics floating around. We also needed a bit more mana-fixing. Naturally, this is where we hit upon reprinting the five mana-Myr—except, of course, we didn't. Design needs to justify their existence by creating new cards, not by blindly reprinting entire cycles.
I proposed a cycle of Mocks (pronounced like "Mox")—simple two-mana artifacts that tap for a specific color of mana. These quickly became important glue holding our set together. Everybody was excited about the awesome game play this cycle enabled, until someone just had to ask: Why weren't we just using the Myr?
Obviously, they satisfied the same needs as the Mocks—which, of course, happens to be the same useful purpose the Myr served in the previous Mirrodin block, making them flavorful, relevant, and exceedingly cute. They're also still good at picking up an axe and hitting your opponent in the face.
Still Haven't Found What We're Looking For
By now, the Mirran mechanical identity had been mostly hammered out—at the time, it was a Rosewaterian dreamscape of counter manipulation, zone changes, sacrifice-matters, and grinding resource conversion, but it was something development could work with. What was missing was a new keyword to lead the Mirran mechanical identity into battle.
Fused: This mechanic linked an artifact to a revealed card. You might wonder how this differs from imprint. It didn't. We decided we would repeat imprint, but that still left us needing something new.
Rebel: I don't remember what this keyword did; it came from an attempt to focus on war-like mechanics, rather than artifacts. We didn't know it at the time, but this was flavorfully anachronistic. In Scars of Mirrodin, the Mirrans aren't actually at war with Phyrexia yet! We deleted this mechanic with prophetic prejudice.
Chain: This mechanic was fun, interesting, and balanced, and will probably see print at some point. Unfortunately, it went on colored cards, and didn't interact with artifacts. Oops.
Flurry, Fuelback, Multiply: By now, we had decided that the important concept to aim for was "artifacts matter." We went through several keywords that triggered on every artifact, or had an effect that multiplied for every artifact you controlled or had played this turn. Most of them had the same flaw—they would have to be costed assuming your deck was mostly artifacts. While we wanted to reward you for playing artifacts, we didn't want players to feel they had to play a deck consisting solely of artifacts to reap reasonable rewards.
Affinity: Although we had initially spurned affinity, at this point we were running out of ideas. We decided that even this menace deserved a second chance. The team (and R&D) was pretty harshly divided over affinity. Some people felt affinity had a lot of interesting space to explore, and if done properly would be fun and exciting. However, many other members of R&D felt that affinity was far too risky—cost reduction is an inherently dangerous mechanic, and its PR firm hadn't done it any favors (one of the most lauded decks of all time was named after affinity) But what harm could come from repeating past mistakes?
I took affinity and walked through half a dozen iterations we could try; we picked to simple versions—"affinity for artifacts" (déjà vu!) and "affinity for creatures"—and sprinkled them throughout the set. It played well, but it didn't feel very "new." Having two variations was confusing; conversely, just having "affinity for creatures" wouldn't suit our needs. After iteration and deliberation, we determined that we would be unable to find a good balance between "new," "fun," "feels Mirran," and "doesn't break Standard in half," so we moved on.
Presence: I've only mentioned half a dozen keywords so far, but dozens more populated our emails and design meetings. The cycle repeated itself over and over—review our design goals, brainstorm mechanics, select one or two to try, playtest, find out the mechanic was relatively fun, and then discover some reason to euthanize our latest experiment and try again. We were only a month or so away from handing the set off to development, and we still needed a new Mirran mechanic. Having ditched affinity like last season's fashion, we set out to brainstorm some more.
With limited time to work, we immediately slotted in a standby mechanic that Mark Globus just had sitting around called "presence." Originally designed to count permanents, it was tweaked to care about artifacts, and design started creating theme decks.
ASIDE: When you have a new mechanic you want to test, don't sticker up cards and play sealed or draft. Instead, build theme decks around the mechanic. Make up cards showcasing the mechanic as you go, and fill the deck out appropriately with cards from the latest set(s) or cards from the set you are designing. As you play, feel free to change cards on the fly as needed—not just for power level, but to make them more fun, more interesting, or try out a new variation of the mechanic. This method really distills a mechanic down to its core with only a couple hours of work; as a bonus, you often have a bunch of cards already designed when you decide to keep the mechanic.
Luckily for us, this story has a happy ending: metalcraft turned out to be easy to understand, tons of fun, and made artifacts matter the way we wanted. The eleventh-hour mechanic that required the least amount of deliberation turned out to be exactly what we needed.
Stars of Mirrodin
Mechanical identity was important, but we also had to make tons of awesome artifacts to showcase Mirrodin. In some ways, making individual cards should be easier than designing the mechanics and structure that supports your entire set; somehow, it never feels that way.
Although Mirrodin, being an artifact set, was allowed to have some number of quirky and awesome artifacts at uncommon, the majority of the marquee cards would end up at rare—so I'm going to spill some dirty secrets about rare design.
At first, a typical set doesn't actually have any rares in it. Since we start by working on the core mechanical identity of a set, and we want that identity to show up in Limited, we naturally start by designing commons and uncommons. Rares languish in the back, patiently waiting their turn, while we hammer away at themes and keywords. Over time, uncommons that are interesting but unfit for uncommon move up to rare; designers also slot in rares that come to mind as they brainstorm commons and uncommons.
Eventually, the set has a full complement of rares that are rarely fit to print. Two problems permeate the rare slots at this stage. These rares all tend to link into the mechanical identity of the set that we are so busy molding. This is fine for some rares—but we also want rares that are just awesome cards, distinct from the themes weaving their way through common and uncommon. More importantly, however, is how most of the rares at this point were generated from people thinking about commons and uncommons. These rares, frankly, tend to suck at being rare.
What typically happens at this point is the design team meets and reviews all the rares. By the end of this meeting, the file is usually decimated; we're lucky if half of the rares survive the bloodbath. This is followed by a round of hole filling—where team members specifically brainstorm rares and then discuss which cards to add to the file. A design file can go through three or four of these iterations before the rares begin to feel less like awkward designer gimmicks and more like something you'd be excited to find in your booster pack.
Given that this is before development even touches the file, getting a rare to go from design to print unscathed is a rare event. I'm proud to say Scars of Mirrodin may contain one of my favorite "brain-to-print" cards ever.
Mirrodin is artifacts—that part was easy. (Clearly my definition of "easy" includes "infinite iterations" and "squeaking in under a deadline.")
But what is Phyrexia? To quote an email from Mark Rosewater—Phyrexia is toxic, viral, unrelenting, and adapting. We just needed to capture that flavor mechanically. No problem!
We knew going in that we would use poison, so our initial designs focused on that. I won't bore you with another twenty paragraphs iterating all the mechanical variations involving poison we tried—partly because many of the unused variations show up later in the block, and partly because Mark waited fourteen years for his thunder and I'm not about to steal it from him.
I don't remember which came first—infect or proliferate—but I do remember Mark dancing a little jig when he found two core mechanics for Phyrexia that interacted so well. Proliferate also interacting with charge counters? I think there were tears of joy.
Don't let my relative brevity trick you into thinking this process was any easier than finding metalcraft. We went through salvage, link, convert, delicious, [CENSORED], and [CENSORED], among others, before deciding that maybe infect and proliferate filled enough space for one set.
I know it sounds like there's a lot of effort and awesome ideas that go to waste—such is the life of a Magic designer. For every ten cards you design, you're lucky if one sees print—and even luckier if it's still recognizable after development finesses it. For every keyword that sees print, dozens more are brainstormed, debated, and tested. As designers, it is our job to sift through all of this rubble, and find the gems worth printing. Ultimately, it is this endless iteration that lets us ensure every set, mechanic, keyword, and card is something you can be excited to find in your next Magic booster pack.