(As with my look at Magic: The Gathering scheduling the previous year, I just about jumped out of my chair when I realized I’d seen something important in the finer details of MTG set releases. What especially tickled me was that I’d used exclusively mechanical circumstances to make a story-relevant observation, even though admittedly I never cared much for the story. Of course, my theory illustrated in the post below was proven 100% correct when the Guilds of Ravnica design columns came around a few months later. There were seven criteria given for the guild arrangements, in order of priority; the top two were functionally requisite to Ravnica sets being color-balanced, and the next two were story considerations. The post does contain various MTG slang, as it was originally posted in an MTG community context. I did make one significant edit to the original piece, which I’ve seamlessly incorporated below. You can see the original pre-edit text, as well as various observed minutiae related to Ravnica guild alignments, in the original Reddit post.)
Two weeks ago, a rumor surfaced that we could be returning to Ravnica, based on a web domain that was likely acquired by WotC. I started crunching some numbers, looking to predict the most likely new arrangement of guilds, assuming two large sets of 5-and-5. That Friday, not only did we get confirmation of Ravnica 3, but we were even told what the new guild arrangement would be. And…. I was way off!
But I looked again, and my number-crunching wasn’t wrong. Six years ago, when I was a newborn Redditor, I discussed why the guild arrangement for Return to Ravnica was the most logical choice, given R&D’s established rules and the new rules they likely applied (most of which I correctly deduced). But this time around, I’m writing about the opposite – why the arrangement for Guilds of Ravnica doesn’t make any sense for all the same reasons Return to Ravnica’s did. I’m going to break down the inherent constraints of arranging two sets of five guilds, the additional restraints R&D has chosen to follow, why the arrangement of “Guilds block” (as I’ll call it) makes no sense, and why that might actually be interesting unto itself.
Warning: Most of this post is hardcore Mel (formerly “Melvin”). But for you Vorthoses, I do have something you might find interesting, if you skip to the conclusion.
TWO SETS OF FIVE
We’ll begin with a breakdown of the unavoidable model that is required for making five-guild sets (a “guild”, of course, being defined strictly as a two-color pair). For this model, we’re operating under the following assumptions, taken straight from MaRo’s design article for Return to Ravnica:
- We want all five colors to have equal representation among the five guilds of a set;
- We want each set to have a mix of ally and enemy guilds (to differentiate Ravnica from other multicolor sets that focus on either ally color pairs or enemy color pairs exclusively).
With five guilds, there are ten slots for colors, so to speak, with each color occupying two of those ten slots. This requires us to establish a chain of guilds that loops back on itself. The chain arrives at a color through one guild, and then leaves through another, giving that color both its slots. This means the chain will never revisit a color, instead winding through the remaining colors until the fifth guild, which returns to the color where the chain started. For an example of this, you could start with white, move to green through Selesnya, move to blue through Simic, move to black through Dimir, move to red through Rakdos, and return to white through Boros, completing a chain. Visually, this chain could take one of two different paths:
We end up with two symmetrical figures, pointing at one of the colors along the color wheel. We’ll call the one on the left the “ally pointer”, because it includes three ally color pairs, and the one on the right the “enemy pointer”, because it includes three enemy color pairs. Note that these aren’t simply two different possibilities; they’re actually partners of each other. The five guilds missing from the left arrangement form the right arrangement. So if you’re designing two sets to share the ten guilds between them, one set will use the “ally pointer” and the other will use the corresponding “enemy pointer”, both pointing to the same color.
It turns out, these are the only two such arrangements possible, without violating one of the above assumptions. (If you accept this claim, then feel free to skip to the next paragraph while I demonstrate it for any skeptics.) Let’s assume, hypothetically, you start with two ally guilds in a row. (By “in a row” I mean they share a color, and thus are next to each other on your chain of five.) If you add a third ally guild in a row, you’ve now locked yourself into all five ally pairs… well, without violating one of the other rules, like skipping the fifth color altogether. Remember, these models are using abstract colors, so this applies no matter how you spin the color wheel around. Since you can’t have three consecutive ally guilds, having two in a row dictates a sequence of enemy-ally-ally-enemy. Usually any given color has two options for an enemy color pairing, but in this case you’ve already limited those options, forcing your hand on both of those enemy guilds, resulting inevitably in the “ally pointer” above. On the other hand, let’s assume you don’t start with two ally guilds in a row, which dictates a sequence of enemy-ally-enemy. (You do have to have at least one ally guild to satisfy the requirements.) In fact, since you’re avoiding two ally neighbors, which resulted in the “ally pointer”, the last two can’t both be ally guilds. This dictates a sequence of either enemy-enemy-ally-enemy, or enemy-ally-enemy-enemy. Notice how, in either case, you have two enemy guilds neighboring, with at least one of them leading to an ally guild. As before, you’ve already closed off your options for which ally color to proceed to, regardless of whichever end of your two enemy guilds you choose to proceed from. You can either close off an isosceles triangle, which violates the above assumptions, or you can move toward the open ally color on that side. Once you’ve done that, you have only one open color remaining, resulting in an enemy guild, and looping back around with one more ally guild. This sequence results in the “enemy pointer” above, regardless of which direction you attempt to resolve it. Note that those pointers can still be rotated to point toward any color on the wheel, so there are still different possibilities in execution, but your diagram will always look like one of those two.
Using this model, we have a grand total of five possible arrangements for the guilds in a 5-and-5 breakdown, based on how the color wheel is rotated around the pointers – or rather how the pointers are rotated inside the color wheel. We’ll refer to these as “Points to [color]” based on what color the two figures are pointing toward:
Others have made mention of this, and it bears repeating: These five are the only possible arrangements of 5-and-5, if we hold true to the two criteria above. Given that we’re likely to continue returning to Ravnica for some time, hey, that’s worth keeping in mind.
RAVNICA 1
Now that we’ve established the 5-and-5 model, let’s throw it out and talk about original Ravnica. Ravnica: City of Guilds was a bold experiment for its day… but that’s a story better suited for history class. Right now, we’re talking science! Newly promoted Head Designer Mark Rosewater wanted to explore all ten two-color combinations over the course of three sets, with those set sizes being large-small-small. This presented a problem, in that there wasn’t enough space to showcase all ten guilds and to evolve them over the course of the block. The bold choice was then made to abandon the tradition of block evolution, and focus only on showcasing all ten guilds. Each guild would appear in only one set, thus getting its time to shine (along with the two or three other guilds that share the set). Among the three expansions, the guilds would be divided 4-3-3 – Four guilds in the large set, three each in the two small sets. But how exactly do you decide which guilds go where? This presented a tremendous logic puzzle that needed to be solved. To go about finding an answer, they laid down the following criteria:
- There would be an even mix of ally and enemy guilds in each set – this meant the large set would be 2-2, while the small sets would be 2-1/1-2;
- Each color must have some guild representation in each set;
- They wanted a mix of play styles in each set – In particular, they didn’t want all the fast guilds (Boros, Rakdos, Gruul) together in one set and all the slow guilds (Dimir, Izzet, Orzhov) together in another;
- R&D took an internal poll to determine which two-color combos were the most popular, with the hope of spreading the most popular ones evenly between the sets (no such ordered list has been publicly provided);
- Given that these guilds would get creative concepting for the first time, the creative team was more confident in executing on some guilds than others, so an effort was made to spread those guilds around;
- A number of other criteria, including unspecified measures taken to improve draft.
Rules Manager Paul Barclay did everyone a favor, and designed a computer algorithm to weigh all possible combinations against these factors and provide options for full block layouts. As the story goes, this computer program yielded one possible result. And so that was the guild arrangement they went with:
What’s interesting to note about the Ravnica: City of Guilds expansion is that when you chain four guilds in this fashion, you effectively get one of the “pointer” models above, but missing one guild. In fact, you could say RCoG was “Enemy pointer, Points to red, Missing Izzet”. That is a mouthful, but that’s what happens when you have thirty possible four-guild combinations that abide by the given restrictions, and 120 total possible combinations for the whole 4-3-3 block. (The explanation of why there are 120 and what they are is interesting, but long, so I’ll put it in the comments.)
RAVNICA 2
Seven years passed, and it was now time to return to Ravnica. This gave R&D the opportunity to attempt a new take on the guild structure, and to fix anything they thought wasn’t ideal the first time around. For instance, one of the things MaRo didn’t like about original Ravnica block was, if your guild wasn’t one of the first four, you never got a chance to draft exclusively your guild.
By this time, it was more and more common to replace a small set of a block with a large set. But this time, instead of upgrading the Spring set, the Winter set became a large set for the first time. The idea was, the Fall set would be large (as it always is) and would showcase five guilds, and the Winter set would also be large and would showcase the other five guilds. Both of these sets would be drafted by themselves, allowing all players to draft their favorite guild. The small set in the Spring would then include cards from all ten guilds, giving them a little block evolution they never had before, while also tying full block draft up into a neat little package.
Additionally, with having done a Ravnica block once already, there was an evolution in the criteria being looked at for dividing up the guilds. Here were the rules guiding the structure for Return block:
- There would again be an even mix of ally and enemy guilds in each set (so one set would be split 2-3 and the other 3-2);
- As in the diagrams above, each color needed to be represented twice in each set;
- Again, all the fast guilds couldn’t be together, and neither could all the slow guilds;
- Groups of guilds that all appeared together in the first Ravnica should not all appear together again;
- Since the three guilds of Dissension got shorted last time, two of them should appear first this time.
In my analysis six years ago, I accurately identified the first three criteria, while missing the last one. The fourth criteria, about breaking up groups of guilds, I took to mean breaking up all groups of three or more, and splitting them up as evenly as possible. In other words, I thought they had made a point to split the four guilds of RCoG two-and-two between the new sets. As I’ll explain, even though it did turn out that way, that doesn’t appear to have been their exact goal.
In his preview article, MaRo describes throwing all these criteria “into a blender”, which provided them two options for the whole block. Turns out, we have access to the same blender, and it produces some interesting results. Let’s go back to that 5-and-5 model and see where we stand:
![[Direct link to image]](https://i.imgur.com/tW3i2tV.png){kind=link}
The numbers under “Guild division” represent how many guilds from these two hypothetical sets of five appeared in each of the original Ravnica sets (RCoG, Guildpact, and Dissension). As you can see, “Points to black” lumps all three Guildpact guilds together, “Points to red” lumps all three Dissension guilds, and “Points to green” does so for both. If you also take into account my assumed criteria that the RCoG guilds would be broken up 2-and-2, “Points to white” is the only viable option. However, MaRo did say they had two options to choose from. A re-reading of his criteria in his own words suggests he was in fact okay with a 3-1 split of RCoG guilds:
Regardless, they did end up using “Points to white”, with the “ally pointer” coming first due the “two guilds from Dissension” criteria. Because of the emphasis on breaking up previous groups of guilds, “Points to white” featured the most first-time guild pairings of the available options, making it the logical choice all around.
RAVNICA 3: WHAT WE COULD’VE HAD
Everything has made sense so far, right? Good, because that’s gonna stop.
Having a good idea that we were about to return again to Ravnica, I dredged up my research in an attempt to forecast the next guild arrangement. (And you thought you were safe, heh heh.) I also took a second look at the criteria from Return block. One thing R&D does not like to do is hit the exact same note twice. Even if they do the same thing, in the same setting, they will figure out how to do it a different way. Thus, I assumed, some attempt would be made to break up repeat pairings, especially the double-repeats (those guild pairs that appeared together in both original Ravnica block and Return block). Breaking up repeats would necessarily result in more first-time pairings, of which seventeen were available. Of those, one guild pair was of particular interest: Azorius and Gruul were the only two guilds to not yet appear together despite not sharing a color. (In a five guild set, any given guild will coexist with two of the six guilds it shares a color with, as well as two of the three guilds with whom it does not share a color. Because of this disparity of opportunities, any two guilds that don’t share a color will be paired up much more frequently than two guilds that do. That said, as I’ll show in the comments, there are fifteen pairs of guilds that don’t share a color, so it is inevitable that one such pair will fall through on two consecutive visits.)
Anyway, I punched all this into a new chart, and arrived at these results:
![[Direct link to image]](https://i.imgur.com/BsX3DAl.png){kind=link}
On the far right, you see two numbers and a symbol under “Repeat rating”. The first number is how many double-repeats would return with that arrangement. (That’s bad.) The second is how many pairings would appear for the first time. (That’s good.) The sign indicates whether those first-time pairings include Azorius-Gruul, with a plus being “Yes” and a minus being “No”. I included “Points to white” just to be comprehensive, but of course it scores poorly across the board – it would be a guild-for-guild repeat of Return block. Testing the options against repeat pairings, the best arrangement was clearly “Points to blue.” It only included one previous double-pairing (Dimir-Boros), and had more new pairings than the other “Points to” arrangements.
Also, per Rosewater’s criteria of mixing fast guilds with slow guilds, I gave each arrangement a calculated “Speed balance”. Here’s MaRo’s list, based on Ravnica 1:
Since MaRo said the middle of his list was arbitrary, I gave the middle four guilds (Selesnya, Simic, Azorius, and Golgari) each a factor of 0. Gruul got a +1, Rakdos got +2, and Boros as the fastest guild got +3. On the other end, Orzhov got -1, Izzet got -2, and Dimir as the slowest guild got -3. I added up the totals to see how much each set is off-center. (Since the two sets offset each other by an equal amount, I only included one number in the chart.) Very interestingly, “Points to green” got a five, almost as high as the theoretical max of 6 if you just handpicked any five guilds regardless of color to be super fast. (So if they ever do decide to use “Points to green”, one of those limited environments will be significantly faster than the other – which isn’t necessarily a bad thing.) “Points to white” is a perfect balance of zero, but “Points to blue” is close, only offset by one, while the last two “Points to” options were offset by two. Note that this is a simplification. R&D may be looking at speed balance as more nuanced than this, so a “one” may not be objectively better than a “two”. Also remember that Gatecrash would be considered a speed-balanced set by this metric, yet those who remember Gatecrash limited will recall the speedy Boros consistently wrecking everyone. So this metric is admittedly not perfect, but it should provide a basic indicator of what they’re looking for, based on their own stated premise.
With all the criteria we had to work with, “Points to blue” is clearly superior to the alternatives. Not only that, but it was already their second viable option for Return to Ravnica, with that decision simply coming down to which guilds Ken Nagle wanted to work with more. This means that, regardless of whatever subjective or undefined criteria R&D may also be considering, “Points to blue” already has their blessing.
Now, you probably noticed I didn’t limit my examination to the five “Points to” arrangements. Even though I thought it was unlikely, I did take a look at sets of all ally guilds and all enemy guilds. That arrangement actually hits every criteria even better than “Points to blue”. It includes no double-repeats, it introduces nine new pairings rather than eight (including Azorius-Gruul), and like “Points to white” it has a perfect “Speed balance” rating of zero. While unlikely given Ravnica’s foundation as the world where ally/enemy distinctions aren’t recognized, I couldn’t rule out this possibility – especially now that the ally/enemy distinction is hardly a thing anymore. But I still considered “Points to blue” the favorite to win.
RAVNICA 3: WHAT WE DID GET
I’d assumed we would yet be kept in the dark about the Fall’s guild arrangement, as happened with previous Ravnica announcements. Instead, we were told the guild arrangement of the new set right up front, and I was left scratching my head.
“Points to red!?“
That’s right! This fall’s arrangement is “Points to red” (with Guilds of Ravnica being the “Enemy pointer” and Ravnica Allegiance in January being the “Ally pointer”). In fact, aside from “Points to green” being a ridiculous outlier on speed balance, “Points to red” scores lowest by nearly every measure. It only includes six out of seventeen possible new pairings. It repeats all four double-repeat pairings. And it’s a speed offset of two. This means Ravnica Allegiance should be noticeably faster than Guilds of Ravnica… and yet, they still somehow managed to put Boros in the “slow” set to wreck all those slow guilds.
If you’ve been paying attention, there’s a very simple reason why “Points to red” scores the lowest on each of our criteria. Recall that RCoG was “Enemy pointer, Points to red, Missing Izzet”. As some people have noticed, the arrangement of Guilds of Ravnica is the same core arrangement that was used for Ravnica: City of Guilds! Of those six new pairings in Guilds block, two come from Izzet filling in the chain where it was missing in RCoG, and the other four come from smooshing Guildpact and Dissension together into one set (minus Izzet). Talk about hitting the same note twice! Nearly all these notes have been played before. (Okay, sure, original Ravnica was 15 years ago, but many players do remember it. You certainly don’t have to repeat it when you have better options available.)
And that’s not even the end of it! Remember how MaRo made it a priority to feature two Dissension guilds in the first set? In fact, here is his exact wording:
This idea of putting previously shorted guilds first wasn’t an offhand thought. It wasn’t a “Maybe if we can do that, it’d be nice.” It was a numbered rule: Guilds that got shorted last time won’t get shorted again. Seven guilds have appeared last in Ravnica blocks: three in Dissension, and five in Gatecrash, with Simic the only guild to be last twice over. Ravnica Allegiance, the latter set this time around, consists exclusively of guilds who went last previously, including the ever-delayed Simic. (Since Ravnica Allegiance can only have five guilds, the other two got to go first by default.) Three guilds – Simic, Orzhov, and Gruul – have never gotten to go first in a Ravnica block; all three of those guilds go last yet again. Three other guilds – Selesnya, Golgari, and Izzet – have never had to go last in a Ravnica block; all three of those guilds go first yet again. Despite it being a priority previously to change which guilds got less playing time, this time out it’s a complete blank on this front. But what makes this point even more peculiar is that it’s really arbitrary. At least with some other considerations, moving this guild to that set means another guild has to move, perhaps resulting in unforeseen consequences. But for this, they wouldn’t have had to shuffle around set arrangements at all. They could just swap Set A and Set B, and fully achieve what they previously made a point to do. All the guilds that never got to go first would go first. But no, for whatever reason, these five have to go in the Fall, and these five (which includes three of five from Gatecrash and all three guilds from Dissension) have to go in the Winter.
None of this is to say Guilds of Ravnica won’t be fun. This is just very puzzling. If one or two of these things were off, if the arrangement was marginally suboptimal, I could accept it being the result of factors we aren’t privy to (like guild popularity or something). But for them to completely whiff on every objective criteria they themselves had previously identified, while passing over a superior alternative that already has their blessing, has me very curious.
CONCLUSION
I don’t make any qualms about not being a story and flavor guy, so I don’t usually have anything to offer the Vorthoses of the crowd. However, this time I may.
I’ve exhausted my analytical toolbox trying to make sense of this without just guessing. I can only come to two possible conclusions:
1) R&D was so satisfied with previous iterations of Ravnica (especially original Ravnica block) and its play patterns that they wanted to mimic that as much as possible, or
2) This arrangement was driven by the story.
I think we’ve established that R&D is not prone to just trotting out the same thing they’ve done before. They’re always looking for new permutations on old ideas. Guilds of Ravnica should have some surprises of course, but the easiest way to help ensure this Ravnica is fresh is to use a new guild arrangement in addition to whatever else you’re cooking up. That way, at the very least, even if their other surprises are duds, you’re not left smashing Selesnya into Golgari for the third Ravnica in a row. Heck, Gavin Verhey himself, in the “Announcement Day” video (linked in the comments), hyped Guilds of Ravnica as including new and different guild combinations, meaning such a shake-up is clearly seen as a positive. So I don’t take conclusion #1 seriously.
And that leaves me with conclusion #2: This guild arrangement was driven exclusively by the needs of the story.
What exactly does that mean? Well, we know this is a three-act story told across Guilds of Ravnica in the Fall, Ravnica Allegiance in the Winter, and “Milk” in the Spring. To use an example, let’s say your story required Dimir to be major players in Act 1, but completely absent in Act 2. Well, then it would be pretty weird if Dimir and its characters didn’t even show up on the cards until Act 2. Synergy between Magic story and the cards in your hand is at an all-time high. While it’s true R&D will not sacrifice basic elements of Magic playability such as color balance to serve a story, there is still a lot of room for story considerations. Taking the core 5-and-5 model we established as a given, it wouldn’t take many story directives to lock the whole block’s structure in place. For instance, Izzet are likely to be a key guild in the story, with Ral Zarek having ties to both the Izzet and Nicol Bolas. (Okay, I am passingly familiar with the story.) If your story required that Izzet and, say Boros both appeared in the first set, congratulations, you’ve just limited yourself to two arrangement options for both sets, and one of those two options is all-enemy followed by all-ally. Different directed guild placements offer different numbers of options, depending on whether those guilds share a color and whether they share ally/enemy status. Regardless, given the firm rules R&D has not and is unlikely to compromise, directed guild placements quickly close off your options. Not every guild has to be where it ended up for the story they have in mind. It may only be two or three guild placements in either set that decided it for everyone else.
Anyway, aside from some bonus trivia in the comments, that’s all I got. Vorthos, it’s on you now.