I spent three years working with a modular cube, so I have many empirical findings to share.
First, about keeping the modules separated and sort them after shuffling. Yes, it’s a significant issue, but it can be overcome and is not a dealbreaker at all. I used tiny triangles of washi tape on the corner of the sleeves to differentiate the modules. Wife’s idea and execution, worked like a charm. See
https://riptidelab.com/forum/threads/too-many-lands.1521/page-4#post-64703 for photos.
On sorting, it’s some work afterwards, but a tiny minor thing compared to the time sunk looking for cards, sleeving, shuffling, read riptidelab.com, etc.
Now that the logistics are out of the way, let’s go on a journey.
v1: Monolithic good stuff cube.
I had a good-stuff 540 card cube. There were no supported archetypes, it was just aggro/control/midrange/tempo. That was quite fun.
v2: Monolithic with archetypes
After starting to frequent riptidelab.com - yes, this is all your fault - I saw that archetypes were an interesting place to take a cube and not all deck needed to be good stuff. I had some minor synergies, but the cube was just not good for Johnnies. So I took my good stuff cube and swapped out a lot of the weakest cards for cards to support archetypes.
Abzan Falconer,
Goblin Bombardment,
Shrapnel Blast went in. Graveyard, reanimator, artifacts, enchantments, counters, sacrifice, spells, pretty much whatever went around here in 2015 was supported. This new 540, archetype-based cube was v2.
With v2, I ran a 5-player draft. It was a fucking trainwreck. My margins were designed to run a 8-player draft, that is, I was counting on ⅔ of each deck to be present in the draft. With 5 people drafting from a pool of 12, all decks were there, and no decks were really there. One guy to draft enchantments, couldn’t find enough, then moved into artifacts pivoting with
Helm of the Gods, and ended up with a pile that was halfway between them. My sacrifice/tokens deck went great, because it was using cards from both archetypes, which were open, and they fed into each other creating enough consistency.
v3: 3 modules
Deeply disappointed, I obsessed for days about what went wrong and how to fix it. Literally obsessed, I couldn’t think of anything else. Turned out that maybe that design was fine at 8 players, but I never got more than 6 players at the time, and most often it was 4 or 5. With this new assumption, I designed the first iteration of a modular cube.
v3 had 3 modules of 180 cards each. At 5-7 players, I would have used 2 modules. At 8+ players, I’d use the full 3 modules. Each archetype was contained in a single one module, meaning it either showed up, or it didn’t.
Even before the first draft, I found the biggest disadvantage. v3 was a complete nightmare to manage. Look, I love working on my cube, but it was insane. I would have to move cards around modules and even the notes about changes were complicated. Balancing the proportion of mana curves, creature/non-creature proportions, artifact/enchantment removal, board wipes, etc. got extremely difficult. I spent a TON of time over months trying to get this version right. I ran a couple of v3 draft and it was alright, but not great. One issue was that the same deck was always there, and the same cards got played with each other. Another issue was that cross-polination among archetypes could be achieved within a module, but where to put the cards that made sense in archetypes not in the same module? Moreover, with 3 modules I had effectively 3 cubes (AB, AC, BC), which was not enough variety for me. It played more like three cubes, rather than a smooth modular cube.
v4: 4 modules
I rearranged the 3 modules into 4 modules of 135 cards to fix the “same deck always there” issue. This way, I had more like 10 combinations (AB, AC, AD, BC, BD, CD, ABC, ABD, ACD, BCD), which was a big improvement in variety. v4 was even harder to manage and didn’t address the cross-polination problem. In fact, it got worse.
The separation was:
Module A - WBG enchantments and WUR spells.
Module B - WU fliers, WRG tokens and BRG sacrifice.
Module C - UR artifacts, WRG counters, WB lifegain
Module D - UBRG madness, WUB reanimator, UBG graveyard, RG lands in graveyard
The way this version played is: if there were two modules, you either drafted a deck from one module, the other module, or a good stuff deck. It was like shuffling two tiny cubes together and playing two entangled subdrafts.
v5: Sattelite modules
Again, I obsessed for some time and found a new design. One main large 360 card module with the generic cards (that always felt loose on v3 and v4, they were moved around a lot to make space), all the cross-polinators, and some archetype cards that were good enough by themselves. 90 cards were moved from each module to the main module, leaving four 45 cards satellite modules. These four smaller modules contained cards that were not useful if a given archetype was not there: the specific payoffs and support cards that did not have the power level to be playable in a good stuff deck.
The cube became very flexible. I could draft [main], or [main + A], or [main + B + C]. The effect was that about half of the decks were just plain good stuff - mostly coming from the main module but also some playables from satellites. The other half of the decks were archetypes from the modules that were present.
This design was good. It played well. The modules mixed reasonably with the main module because of the cross-polinators. One stingy issue - not a huge one, but it was awkward - was that often two modules were included and there was a perfect cross-polinator for them in the main module, but it did not appear since only part of the main module was there
v5 fixed all the problems I had identified in v3 and v4, and I ran it for about two years, up until a few months ago. Some modules were more successful than others. B was great because all themes were connected, C and D were alright, A was too poisonous and weird and eventually became and energy module. I even started getting more creative and made module E - morph.
v5 did have one fatal flaw though, which I took long to understand and kept everyone but me from liking the cube at all. This is a flaw that I believe to be inherent of modular cubes and the biggest reason why they are not appropriate for all but the most hardcore groups.
People like familiarity. They like to learn and feel rewarded when they use that knowledge to get an advantage.
What v5 felt like, as well as v3 and v4 did before it, was not one cube, but a randomly generated cube. There were dozens of combinations, and each felt different. Even main + A and main + B were a world apart. Draft after draft, I was pulling the rug under my playgroup and they would have to start over and figure out the new environment. On the fly.
This made the cube a completely inaccessible format. In my perfect world, I’d be drafting with that group 2 times a week, and it would be fun after they learned what’s in each module, and how they interact. The modules would make replayability almost unlimited. In practice, we drafted every 2 months or so, and it took almost a year to see all the modules.
v6: Monolithic with archetypes again
This year, I went back to a monolithic cube. I took many steps to fix the issues in v2. More mainstream measures, such as reducing the size, increasing the number of picks for fewer players, and just giving up some some archetypes that were hard to weave with others.
I gave up on modular cubes.
Conclusion
Modular cubes are hardcore. They are much, much more time consuming to design, a bit more fiddly to setup and tear down, but this is all on you, cube owner.
The largest burden is really on your players, who will play in a difficult, ever-changing environment, where you will have a significant advantage upper-hand until you get an inhumane amount of drafts in.
If you have an experienced group that will play very frequently (say, every week), I recommend a satellite design like v5. One core module, as many as you want smaller modules that are either 100% in or 100% out.
PS
I went deeper than that. I had the concept of a rare module, rares inside the satellite modules, a simple module, a lands module. These features are left out because they are not relevant to the main message.