For last week's feature, I picked my favorite article that wasn't based on inside information, something that could have appeared on another site as opposed to a "behind the scenes" piece. For this second week of "Best of '05" I knew I wanted to feature the other side of the coin, the inside information the likes of which you can only get on magicthegathering.com. In the end, Mark Rosewater's "Getting Your Philosophy" was the article I wanted to go with. This isn't just a simple glimpse inside the workings of R&D and set design. This is a document the likes of which we've never published on the site, and our company reveals far more of its inner workings than any other company I've ever had any experience with, so that's saying quite a bit. A fan favorite, written by one of our most popular authors, and chock-full of inside information, you can't go wrong with this one. (Even if the blacked-out parts drive you mad!)
-- Scott Johns, content editor, magicthegathering.com
This article first appeared on October 3, 2005.
So, here's the deal. Whenever a set is released, Scott (Johns, magicthegathering.com's editor) contacts the lead designer of the set and has them write a feature article about that set's design. Why should Ravnica be any different? The only problem is it's kind of hard to write an entire feature article about design when I already do it weekly in my column. You know, the one on Magic design. (“Making Magic” for the three of you that don't know what I'm talking about. Check it out.)
But I liked the tradition of the lead designer article, so I decided that I would find something that was worthy of a feature article yet wouldn't conflict with all the design stories I had planned for my column. While trying to do this I started looking through the different documents I had on Ravnica. One of them was a document known as the “Control Design Philosophy” (Control was Ravnica's codename, along with Alt and Delete for the small sets.)
You see, design works about a year on a set. During that year we spend a great deal of time thinking about the set's theme and structure and how exactly it's going to work. At the end of the year, we hand off a final design file to the development team. But the card list doesn't cover all the different thoughts we've worked up during the design. For that reason, I (starting with Odyssey) created what I called a Design Philosophy document. This document walks through the design team's thoughts on how the set was put together. (As Head Designer I now require it for all sets, in greater detail for the large expansions.)
It's very important as it lets the developers understand why the team has chosen to do certain things. This way if they want to make changes they can do so in a way that is consistent with how the design team built the set. For Ravnica, for instance, it proved invaluable in helping the development team understand how we saw the guilds and how we built the set around them.Anyway, as I was reading the Control Design Philosophy document trying to figure out new and interesting content about Ravnica, it struck me that this was it. Here was a piece of history, an actual document, which gives insight into how the design team saw the set. How much more insightful could I hope a feature article to be? The last step was to get permission to reveal this document to all of you. I was given permission to show you most of it. The very few remaining pieces will be blacked out. I felt this was a small price to pay to give you guys a peek at one of the key pieces of documentation about Ravnica's design. Sound like fun? One last time, I want to stress that this was the document I handed off when I handed over the design to the development team (not counting the images added for this web article). As you will see, some things changed in development (as they always do), but the essence of what became Ravnica can be found in this document.
Control Design Philosophy
XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX XXX XXXX XX XX XXXX XX X XX XXXXX XX XXXXX XX XXX XXXXX
I'd like to tell you why I had to black out the first paragraph but it's so secret, I'm not even allowed to tell you the reason I can't tell you. I can say that it has little bearing on understanding the rest of this document.
The XXXXXXXXXXX design foundation for Control is the guild model, or in broader terms two-color interactions. Each of the ten interactions has been given a distinct identity based on color pie philosophy. Control's number one job is to show off and delineate these identities. As such, the Control design team set out to create a design that accomplished this goal. This has played out in several ways:
Guild Breakdowns Per Set – Early playtests demonstrated that having too many guilds represented made it almost impossible to communicate the mechanical flavor of each guild. To solve this problem, the team came up with the idea to spread the ten guilds across the block's three sets. Each guild would appear in only one set. After crunching numbers, the team felt that the best execution was 4-3-3.
In my columns on Ravnica's Design (“City Planning, Parts I, II, and III”) I talked about how hard the earliest version of Ravnica was with all ten color combinations both in traditional multicolor and hybrid. A number of players commented that R&D has to trust that the players can handle complex game interactions and we shouldn't “dumb down” the game. To which I reply, baloney. Simplifying is a key component of design and development. The sign of good design and development of any game is that it accomplishes its task with the least amount of complexity necessary. Complexity unto itself is not a good thing. Magic is plenty complicated. I'll go so far as to say it probably is the most complicated game being currently published. It doesn't need to be any more complex than it already is. If R&D didn't do its job, Magic would be a game that five math PhDs would play for fun over lunch. Actually it wouldn't even be in print anymore because I doubt those five guys between them would buy enough cards to keep the game afloat. Please when we tell you it was too hard, trust us. I don't mean any disrespect, but, it's what we do for a living.
Taking many factors into account (having a mix of ally and enemy guilds each set, popularity of certain guilds, etc.), the design team chose the following breakdown:
Control – U/B, G/W, B/G, R/W
Alt – XXX, XXX, XXX
Delete – XXX, XXX, XXX
The Ravnica design team had to figure out how the entire block was going to lay out. While I can't show you what's coming up (and do you really want me to?), I do want to point out that it was all planned up front.
Mechanical Breakdown Within Each Set – As each guild was only appearing once, the team started to focus on how to mechanically represent each guild. It was decided that strong parallel structuring with loose execution (we are strict that every guild must have X but are liberal in how each guild chooses to make X work) would do the best job of reinforcing the identities of the guilds as well as the contrasts between them.
Let me take a moment to stress the importance of the current parallel structuring. The Control world is going to live and die on us making the guilds matter and making them interesting. How do we do that? We make sure that every aspect of the game plays up how each guild is unique, that it has its own distinctive flavor. But flavor cannot be defined in a vacuum. As an example, let's look at the single colors of the color pie. Let's say you wanted to make the audience understand the flavor of red. We start by making red cards that embody the essence of red. But ultimately to define the color, we have to show the boundaries. And any one color is incapable of doing that. How does the audience know that a red card is on the cusp of not being red?
The answer is using the flavor of the other colors as contrast. You demonstrate what red is not by showing elements in other colors that do not show up in red. A certain quality is not red because it shows up continually in another color. The strongest way to do this is to create two cards that have a similar element but are represented differently in the two colors. White Knight and Black Knight, respectively, seem much more white and black, respectively, because their mirror version shows what they are not.
The same principle holds true for the guilds. There is only so far we can get by showing just the guild. We also need to show how other guilds approach a similar idea differently. This is where the parallel structure is huge. If every guild, for example, has a legend and his followers, it is very telling how the relationship works in each group. In one guild they might be disciples. In another, soldiers. In a third, slaves. This is why the design team has chosen to be so tight in its parallel structure. We want a number of items to show up in each guild. The important part is that each guild has to have enough mechanical freedom that it can show vast differences in how they handle the mechanic.
There is a temptation to do this through number (one guild has three of something while another has one). The problem here is that number is a very hard thing to catch when you're looking one card at a time. Simple variance can hide differences for a long time. Thus, numbers only work if you grossly exaggerate (six to one is more effective than three to one) which eats up a lot of extra slots. Parallel structure does the same work but with much less cards and much easier recognizability (assuming of course creative uses its tools properly).
Also, its important to note that while numbers do a poor job of showing differences, they can do a much better job of showing similarities. Players will more quickly pick up that everyone has the same number of something than they will that things are different. People look for patterns and thus catch on quicker when things have a similar structure.
You might wonder why I spend so many paragraphs on this point. The reason is that it was something that design set up that would be easy for development to misread as simple structuring. When designing a set, it is very easy to set things up such that they run through each category (colors in a normal set; guilds in Ravnica block). It creates a solid structure that is easy to build around. This is why, for example, a lot of cycles work their way into Magic design.
I felt it was important to stress that the parallel structure was not a matter of design ease but rather that it was being used as a means to show contrast between the guilds. As you will see as this document continues, design turned the set in even tighter than it was finally printed. I'll comment on these things as we get to them.
Once we decided to have a strict parallel structure, we started looking at how to execute it:
What we're about to get to is one of the most important elements of the design philosophy. The reason it's so important is that it sets up how we structured the guilds both for mechanical definition and to help the creative team figure out which cards were supposed to represent the guilds. Later it would also help with figuring out which cards got the guild watermark.
Defining Levels – One of the first exercises the design team took was figuring out how cards applied to the guild system. We realized quickly that cards interaction with the guilds could vary significantly. As such, we came up with a system of levels to show how close a card is to the guild model:
Level 1 – These are cards that cannot be played unless both colors of a guild are used. This level is basically multi-color cards.
I say “basically” because technically the rare dual lands also count as Level 1 as they are only used in a deck that is supporting both colors.
Level 2 – These are cards that require both colors of a guild if they are to be optimized. An example of this is the guildmages that are playable with one color, but far more powerful if you have both colors.
As you will see, this level proved to have the richest space for design innovation.
Level 3 – These are cards that only require one color of a guild to play, but are associated flavor-wise with the guild. These cards benefit from playing both guild colors only in the fact that there might be synergy in numbers. The best example of this level are the keyword mechanics. For example, if you want to build a crittercast deck, it will be optimized in green/white.
This is the level that causes the most confusion. Level 3 is the first that has a connection to the guild but yet doesn't require playing both of the guild's colors to maximize the card's effect. In my mind, it is this level which was fundamental to making the guild model work. If each guild card was only playable in its own guild, the block would be too fragmented to work. Each guild just wouldn't have enough space to breathe. The guilds would be strangling one another for cards. The key to making it work was to have a section of cards that felt connected to a particular guild but was playable with other guilds.
The simplest way to think of this was to imagine that the keywords only existed in Level 1 and 2 cards. This means that each guild would have access to only one keyword. But in the current set-up, each guild has access to seven keywords (three for each color and one for the colors combined). This is why the system works. Level 3 allows the cards the surface appearance and definition of a guild without cutting off the ability for other guilds to make use of it.
Level 4 – These are cards that are not guild related but have some loose synergy with the larger goal of the guild. An example would be Raise Dead. The card has no connection with the black/green guild, but it does have synergy with the guild's recursive flavor.
While Raise Dead is not an incorrect example, it is far from the best I could have given. Better examples would include green's token production or blue's milling.
Level 5 – These are cards that have no synergy with any guild.
It was actually important that the set have some of these. The rare Hunted cycle and the common auras with comes-into-play abilities (in fact, the entire aura sub-theme) are examples of this category.
After some discussion, we decided to define the guilds mechanically by the top three levels. (Where creative draws the line, for example who's in the guild versus affiliated with the guild, has yet to be determined.)
Creative also drew the line at Level 3. This means that the card bears the guild's watermark and that the card creatively represents someone or something from within that guild.
Level 1 (Multi-color Numbers) - Once we knew how many guilds existed in each set, we crunched the numbers to figure out how many multi-color cards we could fit in each guild. For starters, we decided that we wanted all the guilds to be of an equal size. A general balance of colors (normally we deal with single colors, but color pairs work very similarly) is a standard for design and the guild system only pushes us to maintain it.
A number of players have expressed to me that they are worried that the other six guilds will not get equal billing as they appear in small sets. You need not worry. The entire 4-3-3 breakdown was chosen because it would allow us to showcase the guilds at equal size (and by equal I mean approximate as the guilds do vary slightly).
The smaller ratio between guilds and overall set size in the smaller expansions dictated our upper size. The team wanted to push multi-color so that it showed up in a greater rarity in all three rarities than Invasion. Also, experience with multi-color showed that it had to skew towards rare to keep deck building from being too difficult/restrictive in limited. In the end we ended up with the following mix:
Common – 4 per guild, 16 (large set); 12 (small set) Uncommon – 7 per guild, 28 (large set); 21 (small set) Rare – 8 per guild, 32 (large set); 24 (small set)
While the actual numbers got shuffled a bit in design, all the guilds did end up with an equal number of multi-color cards (both traditional and hybrid).
While there is a little wiggle room, be aware that a change in any one guild is a change to every guild.
This was the trickiest part about getting development aboard for this block idea. Once they committed to Ravnica it greatly tied the hands of the Guildpact and Dissension design teams.
Next, the Control design team created a number of cycles that fell within multi-color cards. (I'll be talking a bit more about Control cycle philosophy in the next section.) Here they are: (This list is all Level 1 except I've included the keywords – level 3 – to give you a sense of how much extra space there is for non-cycle cards.)
Common (4 multi-color cards)
Small Guy (Followers) – Each guild has a mechanic that shows up on a common and rare creature. The larger creature is a legend and the general flavor is that the small guys are his/her followers. Each guild will flavor this relationship differently.
One of the cycles that fell by the wayside was a big guy/little guy theme that ran in common and rare in all the multi-color cards. While the cycle went away, one copy from each cycle did manage to stay in the file. They are: Dimir Cutpurse, Flame-Kin Zealot, Selesnya Evangel, and Vulturous Zombie.
Keywords (2) – Each guild has two common multi-color cards with the guild's keyword.
Development decided that it wasn't necessary to be so consistent with the number of keywords per guild. They used as many as felt right on a guild-by-guild basis.
Uncommon (7 multi-color cards)
Enchanted Land Seal – Each guild has an enchant land that grants the enchanted land a tap ability that affects creatures of its colors. These cards exist as a means to show the seal or crest of each guild.
It's interesting to note that the cycle was created in the first place as a means to show off the guild's symbol. This goal was accomplished through the signets and this cycle was deemed no longer necessary.
Keywords (1) - Each guild has one uncommon multi-color card with the guild's keyword.
Rare (8 multi-color cards)
Guild Leader – One of the ways to give the guilds flavor is to highlight some of the key players in the guild. (This also has a secondary bonus of reinforcing Kamigawa's legend theme.) This card, a legend, is the creature that serves as the leader of the guild. What this means will vary from guild to guild. Creative has dictated who these creatures are.
Other Guild Member – Each guild has a secondary legend. The role of this legend will depend on the structure of each guild. In some guilds it may be the second in command. In others it runs the army. In others, it's the bodyguard of the leader. Creative has dictated who these creatures are.
The legendary creatures proved to be important to the set for two reasons. First, legendary creatures are very flavorful and did a good job of helping define the guilds. Second, it was my goal to make Ravnica block blend as well as possible with Champions block. To do this I felt that Ravnica had to have a minimum amount of legendary creatures.
Big Guy – This is the legend that is the companion to the little guy. He has a larger version of the same mechanic. Flavorwise, he will be the person who the little guys follow. How they do this will, of course, vary by guild.
Keyword (1) - Each guild has one rare multi-color card with the guild's keyword. Note that this is the only rare card with a keyword.
Equipment – Each guild has a piece of equipment that cost CD or much higher generic cost to equip. The equipment is flavored to do something that mechanically makes sense with the guild.
The equipment got morphed in development to be an artifact connected to each guild.
Dual Lands – These are essentially tap lands that give the player the ability to pay 2 life to have them come into play untapped. In addition, these lands take advantage of the new basic land keywords to grant the lands both basic land types. This cycle has the highest percent chance of being the same throughout the cycle.
As you can see design turned in the dual lands exactly as the got printed. Our inspiration was twofold. First we knew that we wanted to capture some of the fun of the original dual lands by having them count as both basic land types. Next we liked the idea of allowing the land's controller to decide whether the land was a pain land or a tap land. The pay 2 life ended up being a shortcut to make the land feel like a pain land. I should also note that development was originally very skeptical about these lands as they felt at first blush they looked a little “hacky”. Luckily the design team was able to convince them that the pain or tap feel was more elegant than they first realized.
Level 2 (Mono-Color Cards With Multi-color Influence) – Once we the multi-color numbers were in place, the team started looking at the next level down, mono-color cards that needed a second color to be optimized. After much discussion, we came up with a number of cycles.
I know there's a typo in that paragraph but I wanted to keep the document as authentic as possible. Editing doesn't do a pass on internal documents.
Before I get to these cycles though I want to discuss a little philosophy about how we handled them. Unlike most sets, a cycle in this block is referring to ten cards. If you want to make a new dual land, for example, the public will expect you to make all ten. This means that the first set does a lot to dictate the rest of the block. While some expectation is good, the team was very worried about it removing further excitement for the expansions. We don't want the players feeling as if they know everything about sets 2 and 3 just by looking at set 1.
This led the team to make two important decisions. Number one – Control cycles have to be much looser. Traditionally when we make cycles, we try hard for as much structure as possible. Let's take Invasion's guildmages, for example. Each one was a 1/1 for C that had two activations “D,T” and “E,T” (D and E being the ally colors of C). When looking at Control's Guildmages, we decided that we didn't want to lock down mana cost or power/toughness. The activations alone we felt (along with creative help) should dictate their guildmage-ness. This way, while players would anticipate guildmages, they have a lot less ability to figure out what they'll look like.
Number two – cycles can be looser in the block than the first set dictates. Or in other words, each set can approach a loose cycle differently. For instance, let's take a look at the common lands. In Control, the common lands are “filter lands” (the lands tap for colorless and let a player filter the two guild colors). Alt has the freedom to do a different style of common lands. The key is that those three must have a common link and there needs to be one common land for each guild in that set. The only restriction is that Alt dictates whether or not Delete follows suit (if Alt, for example did filter lands, it forces Delete's hand). Which cycles are looser than they first appear (meaning Alt and Delete will do their own versions) is planning to be decided by the Alt and Delete design teams.
Several points here. First, you can see a number of things that got changed in development. The guildmages got turned into hybrid cards when development chose to bring back hybrid and they ended up ironically being more structured. Also, the common cycle completely changed. (Definitely for the better.) Next, another issue that's been raised by some players is the guild system will make the small sets too predictable. As this section above shows, we were conscious about this issue from day one. And yes, there are a bunch of surprises coming your way in the next two sets.
That said, here are the Level 2 cycles in Control:
Guildmages (2) – The guildmages in Control are creatures who have two activated abilities (currently both for “C,T”). The first is weaker in the in-color. The second is stronger in the off-color. The idea behind these guys is that you might play it if you're playing the base color, but you're almost definitely playing it if you have both colors. Also note that each guild has two guildmages as they have one based in each of the two guild colors.
When development chose to bring back hybrid, this was an obvious choice. I will say that although I agree that making them hybrid was the right choice, I did like the above version.
Creatures with Off-Color Activation (2) – Each guild has two (one in each of the guild's colors) creatures that have an off-color activation. Like the guildmages, we've tried to design them such that they will see some play when only the base color is being used, but are optimized when you play both colors. To distinguish them from the guildmages, the off-color abilities do not include a tapping cost. As with the guildmages, each guild has two of these, one centered in each color.
Filter Lands – These are lands that tap for colorless and can filter the two guild colors for 2. The reason I stuck these in as Level 2 and the rare lands as Level 1 is that the filter lands are often used when you are only playing one of the two colors (especially when you are splashing the color) while the dual lands are really only valuable if you're playing both colors.
These lands were designed to make it easier to splash an extra color or two. Development rightfully figured out that this wasn't exactly what we wanted from the common land cycle. These lands wanted to help cement you playing two colors (or possibly three) not enable an Invasion-like five color build.
Guild-Friendly Creatures (2) – These are creatures who get a bonus for the existence of another color in play. These are designed so that you are less likely than the commons to play these if you don't have both colors.
The spirit of this cycle moved to rare in multi-color where there is a cycle of cards that encourage you to play cards of the guild's two colors: (taking over the role of the non-guild leader legendary creature) Agrus Kos, Wojek Veteran; Circu, Dimir Lobotomist; Savra, Queen of the Golgari; and Tolsimir Wolfblood.
Boostable Spells (2) – These are spells that have a second ability if you used the affiliated guild color when playing the spell. These cards were designed such that you quite often will play them without the second color. Having the guild colors turns these from good cards to great cards. This is the kind of a cycle that heavily pulls you to play the second color.
The idea behind this mechanic was to have a spell that could be played mono-color but heavily encouraged you to play it with two colors. This is one of my favorite new things in the set.
There are currently no Level 2 rare cards. If you want to add guild cards, rare is the only place to do it.
Level 3 (Keyword Mechanics) – To reinforce the guild model, the team felt that it was crucial that the keywords for this block were each associated with a guild. That dictated a 4-3-3 keyword structure. While ten keywords is a bit above average (Onslaught and Mirrodin each had around eight, depending what you count), the guild structure will force each keyword to exist in smaller quantities than normal. The team felt that this was in an acceptable range.
Next, we looked at how to execute the keywords within the guilds. Limiting the keywords to multi-color seemed very problematic. This would mean that a player choosing any one guild combination would have access to only one of the block's ten keywords. The next logical option was to allow each mono-color in the guild to also have access to the keyword. This would allow a player playing any one guild combination to have access to seven out of the ten keywords, one of which should be at its strongest as the player has access to every card with the mechanic.
Trying to hold to the strict parallel structure, the team chose to put the keywords in the same allotments in each guild. In addition, playtesting helped give a sense of proper balance. In the end, each guild was given twelve cards with keywords, four in each subsection – mono-color C, mono-color D and multicolor. The team felt that twelve felt weighty enough to be a mechanic, but small enough to be managed in only two colors. Because of the weighting of the gold cards towards rare, the keywords on the multi-color skew more towards rare. Here is the breakdown of keywords per guild:
Common – 6 (2 for each of the mono-colored cards, 2 for multi-color) Uncommon – 5 (2 for each of the mono-colored cards, 1 for multi-color) Rare – 1 (for multi-color)
Development loosened up this restriction (rightfully so). Certain mechanics needed more space while certain others needed less. The disparity of cards with the keyword is the major factor for why the guilds have different numbers of cards with the guild symbol.
Level 4-5 (And the Rest) –
While most of the cycles in the set are guild affiliated, there are a few cycles (traditional five color ones) in Control that have nothing directly to do with the guild model.
Comes-Into-Play Creature Enchantments – These are creature enchantments that have a comes-into-play effect. The cards are designed such that the comes-into-play effect is a larger part of the card than the creature enchantment part. The creature enchantments feel like a nice bonus.
Bring Your Own Enemies – These are creatures that when they come into play give some number of tokens to an opponent. These cards are meant to be flavorful so creative needs to be careful in selecting the creature type and token types. (The design team tried to help in our first pass.) Also, be aware that the design team made it so that one card makes one token up to a fifth card that makes five tokens.
Both of these non-multicolor themed cycles made it through relatively unscathed.
Guild Definitions – The most important aspect of Control design is using the mechanics to capture the feel of each guild. This was accomplished in several ways.
First, as explained above, a parallel structure was set up allowing each guild to show how it executes a similar function. This structuring is covered in the last section.
Second, each guild was given a mechanical identity to match its flavor identity. That is, we figured out how each guild would win based upon its overall guild philosophy and the mechanics available to the two guild colors. Here is a brief synopsis of each of the Control guilds along with a description of its mechanical identity:
Most of this will seem a bit repetitive if you read “City Planning, Part III”. The cool part was how much all of the individual guild philosophy was traced back to design. The one fun part will be to see how the keyword mechanic is described, as there were a number of changes with the keywords in development.
Blue/Black (U/B) – U/B is the sneaky, shadow guild. They are power-hungry and secretive. They want to control the world but behind the scenes. Of all the guilds, it's the one you least want to piss off because they will definitely retaliate but not in a way you'll see coming. U/B is the most mental guild. They're greatest weapon is their mind. They're smart and resourceful. As such, they're favorite target for attack is the opponent's mind.
Mechanically this plays out in two ways. First, U/B is the king of the library. This means it's the guild with the best ability to manipulate their own library and attack their opponent's. Let's start with the former. The strongest way to manipulate your library is tutoring, the ability to get the card you want straight from the deck. U/B's keyword mechanic is a tutoring variant called transmute. Cards with transmute have the ability, while in your hand, to exchange themselves for a card in your library with the same converted mana cost (Mono-color cards currently pay 2U or 2B respectively to transmute where multi-color cards cost UB).
Besides tutoring, U/B manipulates the library by either manipulating the order of the top of the deck, optimizing draw options (usually by drawing multiple cards to chose from), or by removing cards from your deck that are unnecessary thereby increasing draw quality. In all the abilities that aid you in deck manipulation blue is stronger than black. But because the U/B tries to maximize its cards' ability to go either way (help you or attack them), black has a number of cards designed to attack the opponent that occasionally can be used to help you.
Just about everything that U/B can do to its own deck it can do to the opponent's. In addition, U/B does one other thing that it doesn't tend to do to itself. It can mentally deplete the opponent through milling. While both blue and black deplete the opponents' deck, blue tends to do so through milling while black uses lobotomy type effects to eat it away from the inside.
U/B's second big strategic strength is card advantage, both positive (card drawing) and negative (discard). U/B's sneaky, subtle attack is represented in how it gains card advantage over time. Let U/B do its thing and your chances of winning shrink over time. Little by little U/B will get the advantage. This is how U/B makes use of some of its other mechanics. Creature removal and permission allow black to fend off the opponent as it builds up its card advantage.
In the end, U/B's route to victory is as sneaky as the guild itself. U/B will use its mental resources to maximize its own draws in both quantity and quality. And it will use its destructive powers to go after the opponent's mind (deck and library). U/B is the one guild that can occasionally win through decking. (It was the design team's intent that U/B win some non-inconsequential percentage of the limited games this way.) If nothing else, U/B will use its resources to build up card advantage over time such that the opponent can be handled with whatever threat U/B has managed to put on the table.
The one interesting note to this section is that design's “milling level” proved to be a little too high and was brought down by development (although still at a level where it will matter).
Green/White (G/W) – G/W is the guild that most associates itself with the group. The two least selfish colors, G/W thinks of itself as if it were a single entity. Its actions are all aimed towards improving the group. A key factor of this is taking whatever steps it can to enlarge the group and thus strengthen it.
Mechanically this plays out in two ways. First, green is the best at populating itself with lots of creatures. G/W has the highest percentage of creature cards. G/W is the best guild at creating token creatures. And G/W is good at getting creatures from the library into hand or play. No other guild matches G/W's ability to pump out creatures. Remember though that G/W's strength lies not in its quality of creatures but in its quantity.
G/W's keyword mechanic is a perfect example of G/W's ability to let its creature beget more creatures. The keyword is crittercast. Costs for any spell with crittercast can be partially (or fully) payed by tapping creatures for mana of their color.
Second, G/W is very good at protecting its creature and ultimately its wizard. Defensive second to only W/U, G/W has a lot of tools to hold off the opponent. G/W is capable of destroying everything but passive creatures. It has numerous defensive creatures with abilities like regeneration and vigilance. In addition, G/W is best at pumping up its creatures en masse both defensively and offensively.
G/W route to victory is two-fold. First, it can simply attack with some or all of its many creatures. Failing this, G/W goes into Plan B, stall and grow. At some point, G/W's army will grow so big that it can simply overwhelm the opponent.
Not much to add here. Everything said above ended up in the final version.
Black/Green (B/G) - B/G is the creepy guild that wants to infect the world with its own twisted sensibility. Like a disease or fungus, B/G promotes growth as a means to establish control. B/G sees life and death as tools in its quest.
Mechanically this plays out in three ways. First, B/G is constant pumping out threats especially creatures. This is the growth part of B/G. This growth is played out in several ways. Green gives B/G access to token creatures while both black and green allow B/G to grow its creatures with +1/+1 counters.
Next is death. B/G, particularly the black part, is good at using its creatures as a resource. One side grows them, the other side sacrifices them. B/G sees this as the cycle of life.
The most potent part though is the final piece: recursion. B/G is the king is the graveyard and thus is the best at getting its cards from the graveyard back into play. You can kill everything that B/G makes, but B/G will just keep getting it back. B/G mechanic, called reclaim, is a recursion mechanic. Any card with reclaim can put into the hand from the graveyard in the place of any draw. Cards with reclaim can thus be used again and again.
B/G's battle plan is simple. Keep creating threats. When it runs out of them, it just gets more from the graveyard. B/G's threats just keep coming and coming. Eventually, it just wears down the opponent's defenses.
Remember that at this stage, the milling cost for dredge (then reclaim) had not yet been added.
Red/White (R/W) – R/W is the fighting guild. R/W is always itching for a fight. It likes to solve its conflicts with its fists. But R/W has a conscience and as such tries to find proper ways to use its force. R/W tends to find a good cause and then pursue it with any means necessary. They are vigilantes but with heart.
Mechanically this plays out in two ways. R/W's greatest threat is its creatures. This means in order to win it has to attack. And quickly. Think White Weenie meets Sligh. R/W doesn't twiddle its thumbs. It goes for the throat from the very first turn. This is where the red half of R/W shines. Direct damage helps clear the path for the creatures to charge forward. And it doesn't hurt to help knock down the opponent's life.
Like G/W, R/W understands the value of the organization. It thinks of itself as an army carrying out its mission. As such, R/W has a number of spells that help the larger army. Sometimes that means enhancing them aggressively while other times it protects them defensively. R/W's keyword, called radiant, fits this role to a tee. Radiant spells always target creatures. Then depending on which mode you choose (color or creature type), the spell targets the creature and all other creatures that share the chosen characteristic.
R/W is by the far the fastest and most aggressive guild in Control. The other guilds all have means by which they can gain strength over time. Not R/W. It wins quickly or, more often than not, it loses.
If you played at the prerelease you know that we didn't back down on making red/white the aggressive deck.
A third technique was to make sure that each color that overlapped two guilds (white, black and green in Control) was given an ability that played well but differently in each of its two guilds. Here are the key features in each color:
White - The overlap between G/W and R/W is that in each case, the guild has to think in a larger group sense. Both guilds put out a lot of creatures. White's role in each guild is to enhance and protect them. As such, the design team made sure to give white a larger amount of spells and abilities that let it enhance all your creatures.
Black – The overlap between U/B and B/G is its willingness to use its permanents as resources. U/B and B/G are the two guilds that create the greatest card advantage (U/B by drawing cards and B/G by getting cards back from the graveyard) and thus are more able to trade this card advantage for other effects. The design team understood this and made sure to give black more sacrifice costs (especially creatures) than normal.
Green – The overlap between G/W and B/G is the reliance on growth. W/G overwhelms while B/G wears down, but each side is reliant on creating a lot of creatures. As such, the design team decided to give green more creature token production than normal.
Designing the overlap between the mono-colored cards and their respective guilds was one of the most important pieces of this design. You'll see as the block unfolds that the design and development teams took great care to make sure that the mono-colored cards have homes in the different guilds. As such, there are going to be cards that shoot up in value as new guilds enter the environment.
As you can see, Control design has many pieces. Hopefully, this document will help you have some larger sense of what the design team had in mind. If you have any questions, please feel free to contact me.
I'm eager to see what the development team does with Control.
P.S. Below is a complete list of the guild requirements put together.
Guild Breakdown 
And there we go. I hope this insight into an actual piece of the design process was illuminating. Let me know if you'd like to see more of this kind of stuff in the future.
I hope you all enjoy Ravnica as much as I, and my design team (Tyler Bielman, Mike Elliott, Aaron Forsythe, Richard Garfield) enjoyed making it.