Dominaria’s MTGO Collation Issues

This article went longer than I planned, but it needed to explain booster collation so I’m not posting a TL:DR. It’s 2000 words, give or take.


Over the weekend, both paper Magic and MTGO had their prereleases for the new set, Dominaria.

Players were reminded just how good Icy Manipulator is in Limited, and got to play around with a whole lot of new legends.

If you are passing this card in draft, you are probably making a mistake.

The set has a real Kamigawa feel to it – there are Legends everywhere (albeit with names that are easier for an English speaker like myself to pronounce), and a number of intriguing build-around-me cards at the higher rarities. Although the Kamigawa block was polarizing, it has stood the test of time well, introducing at least two cards that are Legacy staples (Glimpse of Nature and Umezawa’s Jitte), many Modern staples and archetype-enabling cards, and it is one of the most significant blocks for casual formats like Commander.

I was hyped for Dominaria for this reason – and on top of that there’s the nostalgia of returning to the plane that the first set I experienced (Ice Age) took place on, and Saga cards that explicitly attempt to turn the stories of these sets into cards.

The MTGO prerelease, however, was marred by a significant issue.


Card Collation

Booster packs are not perfectly random, and this is by design.

Not considering factory errors, in paper Magic, you cannot open two of the same card in one booster. However, in packs with 10 commons, duplicates would actually be a very common occurrence if boosters were truly random.

To prevent this, Wizards de-randomize the packs. The first uncommon added to a pack is effectively random, but it then determines the other two uncommons in the pack. A more complicated system is used for commons, but the same principle applies – the first common put in the pack influences the other commons that can be added, and this is done in a way that prevents duplicates.

Foils break these rules (so you can open Dauntless Bodyguard and a foil Dauntless Bodyguard in one pack, but not two non-foil copies, assuming the system works).

So packs are not perfectly random. But this also applies to booster boxes.

If packs were randomized within boxes, players that opened entire boxes would regularly encounter duplicate and triplicate rares. While opening three copies of Steel Leaf Champion in a box would be a happy outcome, opening three copies of Two-Headed Giant would be disappointing or even infuriating. And most rares printed are not chase cards.

The probability of getting 3 or more non-foil copies of Steel Leaf Champion in a perfectly randomized box would be 2.148%. This isn’t high, but when you factor in that there are 53 different rares, you have around a 70% chance of having a triplicate rare in your box, around a 30% chance of having two or more triplicate rares, and a 10% chance of getting 3+ triplicate rares.

(Disclaimer: I’ve modelled these estimates with binomial distributions, which those of you with a statistics background will recognise are not perfectly accurate here. I’m confident that the estimate will still be close. If you want to run a Monte Carlo simulation of opening ten million boxes, feel free to report your findings, and the maths geek in me would be particularly interested in a rigorous analytical solution to this problem).

To avoid the extreme feel-bad of opening a box with three triplicate bad rares, Wizards of the Coast use a variety of sorting algorithms to decide which rares go into a box that minimizes duplication, just as they do with the commons and uncommons within a pack. They then add in further tweaks to reduce the predictability introduced, which led to the ugly phenomenon of ‘box mapping’ around six years ago.

The end result is a very complex collation system with a lot of moving parts, especially when paired with the 121 card print sheets Magic has used for a very long time.

Dominaria’s ‘every pack has a Legend’ adds to this collation complexity.


The MTGO Impact

That’s all been about paper Magic. Why does this matter for MTGO, where no informed customer ever cracks bulk boosters?

MTGO is designed to simulate paper Magic as accurately as possible, albeit with a few improvements (cheating is basically impossible, and the chess clock most judges would like to see in paper if it were feasible). It also has a few silly bugs, and an omniscient judge at every table.

MTGO is programmed to reflect most of the card collation decisions made in paper. This is true within packs (if you ever see two copies of Adamant Will in a Dominaria booster, one of them will be foil).

It is also true to an extent across multiple packs, making it impossible for a draft to open 3 copies of a given rare in one round of packs. I believe that the 8 packs opened simultaneously in each round of the draft are intended to mirror 8 packs that come from the same box. (This information is based upon my personal recollection of a now long deleted post on the old Wizards of the Coast forums from around ten years ago, so it may not be exactly correct).

This combines to give MTGO a very complex set of algorithims for card collation. And like we all know, when you have a very complex system, something can go wrong.

And in Dominaria, something did go wrong.

On MTGO, most players prefer non-foil cards, at least for cards in the various post-8E frames. The animation on foils performs poorly on many computers that can run MTGO just fine, and as a consequence people who open foils tend to dump them on dealers quickly. Dealers then compile them into complete sets, and these digital foil sets mostly get sold to paper MTG stores. These stores then use MTGO’s redemption program to swap these digital foil sets for foil sets of traditional paper cards. (Side note: Individual players can use redemption too, but it does not appear that these players are a large factor in the MTGO economy. Most redemption appears to be done by people intending to resell the paper cards).

A few hours into Dominaria prerelease events, the large MTGO dealer chain Goatbots realised that they were not being offered any copies of most of the foil Legends in the set.

Possibly because of the complexity of Dominaria’s collation, these cards simply did not get inserted into Dominaria packs online. At the time this article is being written, they still do not exist.


Why This Matters – for MTGO players

If you play MTGO and draft a lot, you probably sell most cards you open to dealers, in exchange for Event Tickets. You probably sell every foil you get, with the possible exception of nearly-valueless foil commons.

Dealers only pay for these cards because they are confident that they can compile sets and then onsell these sets in bulk to the owners of paper MTG stores.

Assuming one foil rare or mythic per 36 packs (this ratio has never been confirmed, and could be as low as 1:27, but I’ll write this assuming 1:36), every 4356 boosters of Dominaria opened online generates enough mythics for 36 normal sets and 1 foil set to be redeemed, as well as a bunch of excess foil commons through rares that don’t assist redemption as mythics are the bottleneck.

This isn’t quite true for smaller numbers of packs (e.g. if only 43560 packs were opened, variance would dominate and you would not have close to 10 of each mythic), but over the number of packs that are drafted during a set’s redemption timeframe on MTGO, these variances become less significant.

This sets a minimum price for cards on the MTGO secondary market. Assuming prices are reasonably similar to the last set of the exact same size (Amonkhet), 36 regular sets will typically sell early on for around USD 2520, and the foil set another 210. This $2730, spread over 4356 boosters, or approximately 60 cents per pack, is what I term the ‘redemption equity’ of boosters on MTGO – it’s the value boosters would be expected to have if you assumed that the only value of digital cards was to compile them into sets to sell to someone who will redeem them.

At the moment, it is impossible to compile a foil set of Dominaria. This means that foil cards have no redemption equity at all, and the redemption equity of each booster is consequentially around 5 cents lower than it should be, and this has been the case for every Limited event run so far.

We don’t know what caused the collation issue, and it’s entirely possible that the missing foils have been replaced by foil cards of the same rarity. However, this does not restore the missing redemption equity.

While a 5 cent rip off here and there doesn’t add up to much for an individual player – even if you entered six release events you’ve only opened 36 boosters of the new set for a grand sum loss of $1.80, it is still something Wizards should make good on.


Why This Matters – for Paper Players

If you don’t play on MTGO, this may seem like a storm in a teacup.

However, MTGO redemption has a large impact on the secondary market for foil cards in paper Magic.

Redemption is a huge source of cards – I know of one MTGO dealer (who is far from the largest one) that sells 600 digital sets a year to their third largest customer alone. Whilst I’ve never redeemed more than 20 of any given set, large paper chains redeem so many sets that they are sent on pallets.

And because foils are not particularly popular online but are hugely valuable in paper Magic, very large percentages of the available foils on MTGO are redeemed.

I won’t go through my methodology (as I’m not confident it is correct), but my best estimate is that about 8% of the copies of a given foil mythic in the paper game come from MTGO redemption.

If you are a paper player and desire foils from Dominaria, this will result in you paying more for them, and having a harder time finding them.


What’s The Solution?

I believe Wizards are right to have MTGO collation mirror paper collation, to ensure that booster draft online is a similar experience to paper draft.

And with systems this complex, mistakes will happen. Programmers are humans.

One of the best things about MTGO is that, unlike in the paper world where it is very expensive for WotC to replace a defective product once it has been sold to a wholesaler, then again to a dealer, then again to a player, on MTGO it is relatively easy for the system to retroactively give players additional digital objects for events they entered.

I believe the best way for WotC to fix this mistake is firstly, to fix the collation issue for future packs.

Then, they should give everyone that has participated in a release event a small promo pack for each prerelease they entered.

Some of these promo packs should contain one of the ‘missing’ foils, at a relative rarity that is carefully calculated so that these cards are no more and no less rare than other cards of the same rarity. My estimate is that about 1 in 5 packs should contain these foils.

For example, if there have been an average of 120 copies of each foil mythic in DAR ‘printed’ so far, the promo packs should be assigned probabilities that produce an expected 120 copies of foil Lyra Dawnbringer. The same should be done with the missing rare and uncommon foils.

Variance may then decide that there are 116 or 127 copies of a specific mythic foil legend printed, but that same variance would have applied to ‘normally’ printed ones anyway.

The remainder of these promo packs could then contain a tiny bogey prize – 5 or 10 play points, a random Modern-legal rare/mythic, or something similar.

Humans make mistakes. This mistake is pretty easily fixed, and should be fixed ASAP.

  • sirgog

20,192 total views, 2 views today

Initial thoughts on Brawl – The ban list should change.

Edit 23-Apr-2018: It turns out that I was partly wrong in a prediction here. Baral is hegemonic in 1v1 Brawl according to the first major tournament (a $25 buyin event on MTGO).

Baral made up 6 of the top 8 and 11 of the top 16; results better than Skullclamp Ravager Affinity Aggro ever put up in events during the month after Darksteel’s release.

While I’d expected The Scarab God to be the biggest danger in the format, it’s now clear that Baral needs an emergency ban. I’ll leave the article unchanged, just take my conclusions with a grain of NaCl.



Brawl was announced recently.

As someone that leans toward the Spike end of the spectrum, I can’t get into Commander. So you might ask why I’d give Brawl any thought at all.


An aside on Commander.

Feel free to skip this section down to the ‘end rant’ text below. It explains why I’ve never liked Commander despite thinking the idea has a lot of potential.

Commander promises fun times until you try to optimize decks at all. At that point, you feel obliged to play a whole lot of decidedly anti-casual effects – absurd combos like Prossh (as a commander) and Food Chain, completely broken mana sources like Sol Ring, and utterly broken consistency engines like Demonic Tutor.

But unlike Vintage, where even stronger combos are common and all of the individually broken cards are also allowed, Commander sharply restricts the available counterplay. Efficient and powerful answers like Mana Drain, Mental Misstep or Force of Will are restricted to 1 per deck, while you can play a lot of tutors.

More powerful in combo than it is in any other archetype, this card epitomizes the consistency a 100 card singleton format should rail against.

When you play to the letter of the Commander rules and observe the official ban list, and then you attempt to win, the optimal strategy is to minimize interaction while you aggressively tutor and assemble some sort of killer play. This could be Tooth and Nail, or a two-card combo where your Commander is one of the cards, or a two-card combo where your Commander can play a role in assembling the combo.

This then ruins the spirit of a singleton format. Singleton formats are (supposedly) intended to increase variance – the amount that games differ from each other. The existence of Magic’s most broken tutors, however, adds a degree of consistency at least equal to that found in Modern.

It also squeezes out all the zany plays possible in a singleton format.

A part of all Magic players – even Spikes like me – has wanted to put Followed Footsteps on Rhox Faithmender while you control Doubling Season.

Zany board states, enormous plays, crushing reversals – these are three of the things that make Magic fun.

Commander could have that feeling, and in groups that agree not to abuse the weaknesses of the ban list, it does. But it is hard to push away the Spike tendancies, and to any Spike, the official ban list screams ‘play combo, or lose to people that do’.

This factor results in me simply giving up on Commander, which is a shame because some previously popular singleton formats (like Australian 7-point Highlander, or Canadian Highlander) have been a lot of fun.

Australian 7-point Highlander in particular was something I have fond memories of. If you are unfamiliar with the format, it’s a singleton format that has a pseudo-Restricted list. Certain cards are allocated ‘points’ and you were limited to running 7 points in total. Cards like Demonic Tutor had huge point scores (4 when I followed the format), and so playing them was only possible if you did not spend a lot of your 7 deckbuilding points on powerful combo payoffs. You could play DT, you could play Vampiric Tutor, Imperial Seal or Grim Tutor, but you could not jam all of them into a deck alongside explosive mana and multiple two-card combos the way you can in Commander. Nor did you begin the game with a Legend already tutored up for you and protected from interaction as well.

End rant.


So that’s been a bit of a rant about why Commander breaks when you attempt to optimize it. The raw power of tutors available allows you to build a consistent deck with an extremely proactive game plan, alongside generic good cards to try to win when you do not pull off your combo(s).

Brawl does not appear to have this problem.

Despite Standard having experienced two bannings related to combo strategies in very recent history (Felidar Guardian from the Crazy Cat Lady Combo deck, and Aetherworks Marvel from  yet another deck that could play either a combo game or tapout control), Brawl’s nature as a singleton format with only one half-decent tutor ensures that there is no way to assemble combos reliably.


Brawl’s Ban List

At present Brawl inherits the ban list of Standard. The following seven cards are banned:

Aetherworks Marvel

Smuggler’s Copter

Felidar Guardian

Attune with Aether

Rogue Refiner

Rampaging Ferocidon

Ramunap Ruins


Marvel and Guardian are banned in Standard because of combos that were too strong for the format, but that are unlikely to be good at all in the context of Brawl. Felidar Guardian is not legal if Saheeli is your Commander, and if Saheeli is not your Commander, you have to naturally draw both pieces of the combo or use a janky tutor like Djeru or Mastermind’s Acquisition to find them.

This card is one of two tutors in the format that could get a piece of the Crazy Cat Lady Combo, were it legal in Brawl. You can only play one copy.

Aetherworks Marvel is even less exploitable in the format. All the energy cards are either banned (Attune, Refiner) or limited to one copy, and it’s just impossible to get the required critical mass of energy generators. Even if you do manage to set off the Marvel, big deal – the highest impact play you can spin up is probably Bolas, and you are only allowed one copy of him.

The other cards on the ban list are all there because of highly tuned decks that assembled a critical mass of synergistic cards.

Ramanup Red put two cards on the list, as did Temur Energy. Both decks simply would not function in a singleton environment.

For this reason I think that every card on the Brawl banned list, with the exception of Snugglecopter, should definitely be removed. Snugglecopter I’m less sure about but my initial thought is that it would merely be a good card and an incentive to try to play aggressive strategies, rather than the incredible powerhouse it was in Standard.

At the same time, Brawl’s nature as a format where you can guarantee access to specific cards empowers two of Standard’s more powerful threats to levels that might render them oppressive.

I’m talking Baral, Chief of Compliance (a card that was an absolute terror in MTGO’s 1v1 Commander format until he was taken out the back and shot), and The Scarab God.

The Scarab God is one of Standard’s best threats, and his existence shapes many removal choices. Players run Vraska’s Contempt largely because Vraska can actually answer TSG, unlike removal spells like Murder that would be better in most Standard formats.

However, in a format with the Commander ruleset, these cards fail to answer TSG. He can be recast from the Command Zone when the opponent draws their one Vraska’s Contempt or their one Cast Out. Instead of answering TSG, these cards simply throw it off balance for a short time, much like Unsummon would.

These same concerns about the TSG posing a nearly unsolvable threat also apply to The Scorpion God and The Locust God, but those cards are not close to as powerful as their blue-black cousin.

Baral may not even be good in the format, but his effect is dangerous when you can rely upon always having access to it.



I’m not usually an advocate of minimalist ban lists.

Standard has improved tremendously with each banning that has happened over the last while, and we’ve reached a point where the major design fuckups of the last 3-4 years have all rotated (Collected Company), or been banned (Crazy Cat Lady Combo). Standard was terrible for a period, but it’s now quite a decent format.

That said, in the case of Brawl, I’m advocating a clean slate.

Brawl’s banlist should be decoupled from Standard’s banlist, and Brawl should start out with nothing banned.

Three cards – Snugglecopter, Baral and The Scarab God – are moderately likely to prove themselves troublesome. There may be others that I haven’t considered (Bontu and Oketra come to mind in particular as Commanders that are potentially troublesome).

Even if the cards named wind up fine, if the ban lists remain coupled and Brawl is supported in the longer term, it’s almost certain that some future card like The Scarab God will cause issues in Brawl but is a positive force in Standard.

I don’t want Wizards to be faced with having the choice of either letting that card ruin Brawl, or having to hit Standard with a banning that is purely collateral damage from an action to save Brawl.

In the event that one or more of these cards actually proves itself troublesome, it should be promptly banned. But unless this transpires, Brawl provides a great opportunity to play with some of the cards that are banned in Standard but are too weak for the larger formats.

In conclusion: Decouple the ban lists, just as Legacy’s ban list was decoupled from Vintage’s B&R lists about 15 years ago. Then unban everything in Brawl, but be ready to throw Snugglecopter, Baral, The Scarab God or potentially other cards under the bus if/when needed.

2,470 total views, 1 views today