Tuesday, December 5, 2017


Hey friend. I want to tell you about a game I'm really excited about. It's the most expensive game on the market, and the rulebook is so much bigger than all other games they don't even print it. But it's totally worth it because the theme and the art are fantastic and the gameplay is so deep and fun. I mean, when you get to play; Maybe a fifth of all games are decided by the luck of the draw, though you don't always know that until you've played it out.

Let's talk about game design, variance, and non-games for a bit. Non-Games are experiences that look like games but whose outcome is not influenced by choices the players make. Candyland is a non-game. Some games of Magic are close enough to being non-games that it's a problem worth discussing.

Non-games aren't Magic's biggest problem. A steep learning curve, overcoming the perception of a poor value ratio and appealing to the full potential market are #1, #2 and #3 in some order. I would put the frequency of games that are frustratingly decided by the draw at #4. I'm calling these non-games because they aren't decided by decisions players make after the game has begun (after deck-building and mulligans, which certainly factor heavily). While not the most important problem facing the game, it is important, and it's something we can address.

Dicey Spoony

Toy game is a term for a game you'd never actually play, but whose analysis is useful for studying Game Theory. The Prisoner's Dilemma is the best known toy game.
Consider this toy game. Every round, two players roll one six-sided die each. If either rolls a 6 and the other doesn't, that player wins. If either rolls a 1 and the other doesn't, that player loses. Otherwise, the game continues.

Thinking for 3 seconds or less, how often would you guess that roll ends the game? If you said one-third of the time, you'd be in the vast majority. A d6 will yield a 1 or a 6 two out of six times: 1:3. Let's map out the possibilities and see the exact answer.

1 2 3 4 5 6
1 L L L L L
2 W L
3 W L
4 W L
5 W L
6 W W W W W
If I roll a 1 and you don't, I lose. If you roll a 1 and I don't, I win. If I roll a 6 and you don't, I win. If you roll a 6 and I don't, I lose. There are 18 results where the game ends. Out of 36 possible results, that's half the time. It doesn't look like half glancing at the chart, but it adds up.

We knew this game was bad, but it's worse than we thought. No matter what fun/engaging things you give players to do the rest of the time, having the outcome determined randomly half the time is awful. Players need to feel like they have agency and huge random effects hurt or destroy that feeling.

Let's talk mulligans 

Mulligans are difficult. There's a ton of thought and skill you can apply to the decision, but they're also emotionally taxing. As clearly correct as it is to mulligan your one-land starting hand, it still sucks to draw a new six-card hand with just one land… or no lands at all. Pro players try to vulcan up and avoid attaching any emotion to moments like that because it's out of their control and tilting will only hurt their chances even further. Here's the thing about Vulcans: Suppressing your emotions is bad for your health. And it's not fun. (Remember how Magic is a game?) It also takes real mental effort, and that comes out of the pool you're using to play your best game. Importantly, it's also not something you can ask of players who aren't pros—you know, the vast majority of the audience.

Let's forget about that concern and look more generally at mulligans now.

Consider another toy game: Dice Mulligan. Each player privately rolls a d12. If you don't like your result, you can roll a d10. If you don't like that, roll a d8, etc down to d4. Once all rolls are final, reveal your die. The player with the highest result wins.

This game is a simple analogy for mulliganing in Magic.

Draw Pairs

Here's a more complex toy game that models the first few rounds of a Magic game. Players each have a deck with 30 action cards and 30 resource cards. They draw 7 cards each and can Vancouver-mulligan. Then they'll draw 3 more cards each and reveal their hands. Whoever has the most pairs of action and resource cards wins (so if I have 4 actions and 6 resources, I've got 4 pairs).

The entire game is decided by the luck of the draw and how many times you choose to mulligan. Your only influence on the game's result is your choice whether or not to shuffle your hand away and draw another, smaller hand. You definitely keep a 3:4 hand and definitely pitch a 0:7 or 1:6 hand. You probably pitch a 2:5 hand. And those are all the possible 7-card scenarios. If you go to six cards, you still pitch a 0:6 or 1:5 hand and keep a 3:3 hand. Again, you're only thinking about a 2:4 hand. Being able to scry improves your chances of drawing 2 of the cards you need in your next 3 draws, and your best possible five-card hand is a measly 2:3, so you probably keep 2:4. If you don't go to six cards, you're hoping to draw as many of the cards you need as possible in your next 3.

The entire game comes down to your decision at the start. What if you had to play for 5-15 minutes before finding out who won at the outset? How many games of Magic are won or lost before the first turn in almost this exact way?

Most Magic games aren't decided so early, and those tend to be fun. Even some of the games that are can be fun because you get some good plays in and the game ends closer than it started. We rarely even know which games were pre-determined by our shuffle and which we could've changed with different choices. If we did know, we'd be frustrated more often. It's good for the game that we don't, but it's bad for the game's design because the problem is obscured, and so the drive to solve it is dampened.

A Designer's Responsibility

I was watching Shawn Main's talk at roguelike_con and the last question was from a game designer asking how to communicate that his game is balanced to players who get clobbered in their first game.

Imagine a game that's really fun, but where a player will lose due to randomness 1 in 100 times. If you print 5000 copies of that game, 50 players are going to lose this way in their first game. Even if you're okay selling that experience, those 50 players are all going to rate your game terribly online, where only the usual percentage will rate it 'fairly.' The players who watched their friend lose this way will also judge the game harshly. And the groups who didn't experience this catastrophe will play again, and 49 more will hate your game. And so on.

In fact, the better your game is, the more a flaw like this will hurt, because more people will play it more often and suffer from the flaw more frequently.

A game designer is responsible for the average experience their game creates, but they are also responsible for every experience their game creates. If your game is "balanced" so that it gives the intended experience more often than not, or even almost all the time, your game isn't ready for publication. It has to deliver every time. That means very different things for different games. I'm not saying you can't have luck or elimination. But if that's not the core experience, and it can happen and wreck someone's fun, you need to address that. And if you're blaming the players for making poor choices—nah. Nah, you put those choices in the game, dear designer. You failed to communicate the costs, the risks, and/or the consequences of those choices. Players don't play games wrong, they just play wrong games.

The point is, Magic creates far too many bad experiences. Even if it's only 1 in 10 games that suck, that's still many tens of thousands of bad games every day. Wizards of the Coast needs to do everything it reasonably can to mitigate this flaw.

What can't we do?

If we could order every deck before every game—remove all the variance of the shuffle—we could guarantee every game competitive. But we can't assign a game designer to every kitchen table to do that. We can't let players order their own decks, because that creates perfect information that kills the fun of the game (and they'd choose optimal draws that ruin the challenge). We can't let players order other players' decks because they're not all game designers with a complete understanding of how to create dramatic game experiences. Even if we wanted to program an algorithm into digital versions of Magic, how do you write logic that can analyze any matching of any decks and order them to create natural and meaningful experiences for humans? (Machine Learning might actually be able to, but that doesn't help with physical games, and so isn't sufficient here.)

We can't eliminate land flood or mana screw entirely. First, it's not possible without changing the game so fundamentally it's no longer Magic. Second, we don't want to do that because it's important to the game—see Mark Rosewater's excellent article Mana Action. We can't eliminate the factors that make misfires possible without removing the drama, the tactical adaptability, and the deck-building challenge that makes Magic MagicThere will always be non-games of Magic. But we can recognize a problem with our game and take a few small steps to mitigate it.

What has R&D already done?

Magic has employed some techniques that help considerably for many years.

As imperfect as the Vancouver mulligan is, it's better than the Paris mulligan before it, and that's better by miles than no mulligan.

Every new set is internally required to include a smoothing mechanic like explore, scry, or cycling. This is very much one of the reasons Magic is doing better now than it ever has. If you're somehow skeptical, go play Sealed with a set from the '90s.

R&D committed to making strong dual lands in every set about a decade ago, with the intent and result of reducing color screw.

You could even argue that the rapid increase in the power level of creature cards at common over the past several years reduces the relative impact of rares, and mitigates the difference between the games where you draw your rares and the ones where you don't. Or that it makes each individual play more impactful, and thus creates more dynamic play.

What are our options to address non-games of Magic?

The obvious solution then, is to double-down on what we've got now:

1. Make mulligans easier and more effective.

2. Make smoothing cheaper and more pervasive.

3. Make land flooding less frequent or less painful.

4. Make color screw less devastating.

5. Make automatic losses shorter or more fun.

6. Obscure automatic losses more or give players more hope to overcome bad shuffles.

The best solutions in Magic tend to be card-based. It's also possible to modify the existing rules in a way that achieves any of the above universally. Such changes need to be considered very carefully, however, because they can add more complexity when that's the last thing we need. Even when such changes are definitely positive (like fixing the stack, eliminating mana burn, or the Vancouver mulligan), disseminating that information to a player base numbering in the millions is a serious challenge and there will be resistance. So we can consider rules changes, but any time we can affect progress by designing cards and sets differently, that option is much preferable.

Share suggestions of cards/mechanics/rules to help prevent non-games, or observations of such that Magic or other games have used and how successful they were.


  1. Great post!

    This is something I think about a lot in the context of Eternal. (PS: If you're reading this and you aren't playing Direwolf's Eternal, you should be!)

    The Eternal reddits are FLOODED with people complaining about mana screw/flood (which is actually a smaller problem in Eternal than it is in Magic). As Eternal is completely digital, it is rightly pointed out that it would be very easy to smooth the probabilities of drawing lands and spells, ensuring that no one was every stuck on two lands for 7 turns.

    That would also ensure that the game the other day that I WON while stuck on two lands for 7 turns would never happen. That game was awesome! Is that a worthwhile tradeoff? I can't say.

    What I can say for sure is that for many players, losing games due to mana screw/flood is far more frustrating than it is for me. I often see players concede turn three out of no where, presumably because they were about to miss their third land drop. Those players, I think, very often quit. Worse, when they don't quit, they complain and flame and sour the experience for others. I'm sure everyone reading this has played against that person on MTGO chat.

    A lot of this is, I think, community responsibility. Teaching newer gamers how to think about and handle randomness. And, to be clear, I don't mean telling them to suffer in silence, I mean helping them develop a healthier attitude that it is just a game and letting them move onto the next one.

    A big part of this is making individual games of Magic/Eternal low-stakes. If I'm just playing a random ladder game, who cares if I get mana screwed, it is probably over in 2 minutes and I can just requeue and play again. When people get really upset by variance is when they have a financial stake in it. Unfortunately, MTGO is designed entirely around paying to play games of Magic which then have prizes skewed heavily towards the winners. When you draft a deck you think will be great, and then you 0-3 because you get mana screwed game after game, that sucks, especially if you have very limited funds for gaming. Then you're in a position where you're like "that was my gaming money for this week, I guess I can draft again next week," and that is a place where anyone would feel upset.

    My sincerest hope is that they get the psychology of money right in MTG Arena. Your post gave me a bunch of other thoughts as well, I'll likely list those in different posts.

    1. I strongly agree with the financial point. I only care about “not playing magic” in tournaments. When i cube I can always shuffle up for a game 3 or 4, and usually do.

      This is kind of wild brainstorming but maybe we define a “non-game” somehow (a loss where you drew 2 lands/2 nonlands or fewer?) and they count as half a loss instead of a full loss? That would make too many matches go to time...but maybe something similar could be workable.

      Alternately you could use Eternal’s deck building rule of at least N% resource cards in your deck, and then allow all 6-1 or 7-0 hands to be mulliganed for free, on top of Vancouver mulligan rules for 5-2 or 4-3 hands.

    2. Well said, Tommy.

      The entire reason we play matches best-2-out-of-3 is to reduce the cost of one bad draw. Pro finals are best 3-out-of-5 because the stakes are even higher.

      I can't advocate for longer matches across the board because Magic events drag a bit long as it is, but we can consider options in the same vein.

      And yes, the community would benefit from teaching players that tough games can be won (and that those are the most satisfying to win), and that losing about half your games is simply expected at all levels of (well-matched) play. I think while digital introductions to Magic are great, this is one of the places they fail worst.

  2. When the Vancouver mulligan was introduced, it was explained that you didn't get to scry X, where X is the number of cards you're down, because that might be too strong for combo. Have we actually tested that enough to know? If not, we should. If so, I honestly wonder if we shouldn't address the source of that problem—those combo decks that are too good with the right four or five card hand—rather than the solution shining light on it. (Granted, that involves banning/restricting cards, and that's something we should rightly be hesitant to do. But the long-term health of the game ultimately has to take precedence—that's why bans ever happen.)

    For the vast majority of decks, having five cards and scrying 2 is still strictly worse than having six cards and scrying 1, and so on all the way to one card and scry 6. There's a very fair argument that scry 6 is bad for the game no matter the circumstance because it removes six turns of uncertainty and cripples the game's drama. Coupled with the fact that you're still probably going to lose, it's clear that we don't want players to mulligan that far.

    But I put it to you that I'm far less likely to mulligan from 5 cards to 4 if I know a scry 2 is coming. And even if I do go to 4 cards, when would I consider going to 3 cards when I can scry 3? Going to proportional scrying will increase the number of times players mulligan from 6 to 5, but should reduce the total number of mulligans across the board.

    1. If we got rid of combo to a point where it was possible, I still think you’d have issues? Linear decks would suffer post-board (always drawing Stony Silence) exacerbating the “silver bullet” problem in Modern.

    2. I think there are infinitely many reasons to ban combo decks in all formats into the dirt. This is a great one!

    3. I have a problem with arguing for easier mulligans based on high-level (i.e. GP+) Constructed play, because the decks there are 100% under player control and already factor mulligans into their design. "Draw-Perfect Temur" from the recent PT is a great example-- the designers accepted the risk of auto-losing based on mana screw because the upside of playing haymakers in 4 colors (many of them with multicolor or double-color costs) was so great. If mulligans become less punishing, pros will just respond by moving further down that curve, and decks will become greedier as a result-- which may actually increase variance as played.

    4. I agree, but I would phrase it like this: As designers are job is to make sure the best strategic choice and the most fun choice align as often as possible.

      If the best strategic choice is a deck that wins 70% of the time, but the 30% of the time it loses is due to mana screw, people will play that even if it is miserable.

      One of the points I did not make in my below post about partial Paris mulligans is that they encourage players to try and skimp on lands leading to more frequent mana screw later in the game.

    5. I'm arguing for easier mulligans not because of high-level play, but despite it. Hell, if Wizards wants to say you get more scries at REL 0 events and fewer at higher events, I'd get behind that. Or more likely, simply distinguishing between 40-card, 60-card, and 100-card play.

      To be clear, there's absolutely a point of diminishing returns and we don't want to remove too much randomness.

      Another factor in the viability of 4- and 5-color good stuff decks is the prevalence of strong dual (and tri-) lands, something R&D tremendously ramped up about a decade ago precisely to address non-games. Not saying we need more or less fixy lands, just observing corrective action that's already taken place.

  3. Here is a precise way to frame the question: Suppose I take a very skilled player (Jon Finkel) and a lower skill player (you can set the exact skill you want).

    I give them each a competitive deck, and pick a random seed (that is, the order of the decks, and the outcomes of all other random things that will occur). There are two outcomes: Jon wins or LSP wins.

    Now I rewind time, and use the same random seed, but switch the player seats. There are two outcomes: Jon wins or LSP wins.

    In total, that makes three types of outcomes: Jon/Jon, {Jon/LSP or LSP/Jon}, and LSP/LSP. We presume that there will be almost no LSP/LSP games (he is Jon Finkel after all) so really there are two:

    LSP/Jon or Jon/LSP: These are games determined by the random seed, that is games where the player skill (or I'd prefer to say player choices) did not really matter. These are non-games. Now, as you note in your article, there are a lot of flavors of non-games, and some might be exciting for a while but then one deck draws 10 lands in a row and loses regardless of who is piloting it.

    Jon/Jon: These are the games that offer both players enough meaningful choices to win the game if they play it correctly. These are not all equal either. Some will be incredibly even, that is Jon will be indifferent to which side he plays. In others, they will be very unequal: that is, Jon will win from one side very easily, but from the other side will win only through a long series of clever players that most players would not have seen.

    Ultimately, though, I think one can ask the "non-games" question as "What percentage of games are LSP/Jon or Jon/LSP games, and what percentage are Jon/Jon games?"

    My instinct is that it is about 50-50. This squares with the fact that the very best players have win percentages at GPs that are something like 70-75% (which is what you'd expect if it was really 50-50).

    Unfortunately, we cannot treat this problem in isolation! If, as I approximate, it is 50-50, then LSP's winrate is somewhere around 25%. If we implement a policy that will radically reduce the number of non-games, that is, say, we increase the number of Jon/Jon games to 80% and non-games to 20%, now our LSP is winning only 10% of the games he plays. Is LSP going to continue playing a game if he is losing 90% of the time?

    I know what you're thinking: Well, games where a player is mana screwed aren't fun for anyone. The LSP won't feel any sense of accomplishment beating poor Jon while he sits on 3 lands and one color of mana. But I think this is completely false. I think most of the time if I put my cards away and asked a LSP "What colors was my deck that game" he could not answer, much less noticing things like missed land drops, flood, poor draws, and more subtle problems.

    What do you think the percentage of Jon/Jon games is? How much does it depend on the format (Standard vs Modern vs Limited)?

    1. Why is LSP playing all their games against Jon Finkel? This is a match-making problem, not a game variance problem. Two players should only be paired if you can't say that either player's victory requires a non-game.

    2. So I, as someone who has played Magic for 20+ years as my main hobby, should not be able to teach my friends who have never played and enjoy playing with them? That seems like a really serious problem, if so.

      For what it's worth, that is exactly why I no longer play chess, match making is impossible!

    3. I didn't know you were as good as Jon Finkel!

      Seriously, though. I've also played for a very long time and am leagues beyond a first-time player. (And 'leagues' is particularly appropriate here because sports use leagues to match players of appropriate skill level. To an extent, Magic does too.) I can play a game with a brand new player and expect to win 90% of the time. That's fine for their first few games, but then they're going to want to play with someone closer to their skill level until they level up. That's how games of skill work.

      The variance in Magic is what keeps that from being a 100% win-rate; some of that comes from non-games (which you're right, brand new players won't realize) and some will come from games on that fringe. Those games on the fringe are the best ones. So there are two more reasons we wouldn't want to eliminate non-games entirely (even if we could). But in the same way that we don't want to flip a coin to see who wins, that doesn't mean the percentage of non-games that currently exist is ideal either.

      Sports also use handicaps to help players of different skill levels play together. Surely when you're playing a new player, you choose the weaker deck?

    4. I would say this statement sums up a big difference between me and a lot of Mag players: "they're going to want to play with someone closer to their skill level until they level up"

      I play games with people I like, rather than playing games I like with people. I don't have a great sense of the exact breakdown, but I know that I'm unusual in Magic. That is, most players choose to play Magic, and then play with whoever shows up at FNM etc. I have more of a board gamer mentality, I have a game night at my house, people show up there, and we play a game, sometimes its Magic, sometimes it is one of the other 400 games we have (In other news, I have a problem...).

      I would say it is a HUGE pro of Magic that I can teach a new player and with a little training have them cube drafting well enough that they beat me some times without too much work.

      But then, Magic has this near infinite path of ways you can get better, additional edges you can seek, you're always learning something, so that you can feel like you're getting better at Magic for decades.

      If a player wants to improve quickly, Duels is a huge help, particularly for improving a player's skill at limited for more vanilla sets. It gives you a lot of practice attacking and blocking with flying, trample, first strike, etc.

    5. Yeah, I want to play the game I love with the people I love, and to make 'fun' win ratios, that requires (1) them to have the endurance to withstand losing a lot (2) me going easy on them without patronizing them or (3) luck to turn in their favor.

      Generally, I just play faster, or with weirder and more unreliable strategies, or with themed decks, or intro decks, or like I did on a memorable train ride, making a deck from random leftovers they have. But that isn't always enough to reduce the win rate to 'fun' levels.

  4. Finally, I'll note what I do for games at my house. I exclusively play limited/cube so I'm not worried about turn one combo decks.

    At my house, we always use partial paris mulligans with one free mulligan, only reshuffling after all mulligans are complete. That is, if I don't like 6 cards in my opener, I'll pitch 6 and draw 6. Then, if I still don't like 4 cards, I can pitch those to draw 3, and so on.

    With this mulligan format, it is VERY rare to run into mana issues early. I will note that Paris Mulligans do you have a lot of problems, but I won't go into them at length here as I've already written as much as Jay in the comments.

    1. Can you clarify how you mulligan?
      Is the example that I can set aside 6 of 7 cards from my starting hand, and draw 6 new cards as my free mulligan. Then, if I want to mulligan again, I choose 4 of my 7 cards to set aside and draw 3 to replace them. And then I shuffle the 10 cards I set aside?

    2. Correct, then you shuffle those into your library. Not shuffling is largely a matter of making this process faster, as at my kitchen table I'm definitely not worried someone is cheating or stacking their deck.

      And to be clear, the 6 and the 4 were just example numbers, you can mulligan anywhere from 0 cards to your whole hand each time.

    3. This is almost identical to a mulligan rule I was going to propose. Here are the big differences between what you do and the status quo:

      1. You don't pitch your whole hand, just the cards you feel you'd be better off replacing. This increases the number of choices per mulligan from 1 to 5041, but shouldn't be so bad in practice. It definitely gives you more agency. Being able to ditch your four CMC 5+ spells in a land-light hand would feel great. Being able to ditch all but 3 land (of your choice) would feel great.

      2. Not shuffling those cards in before re-drawing does save significant time and effort, but it also reduces the chances of drawing back the very cards you chose to pitch, making the original choice more effective.

      3. The one free mulligan isn't something I was thinking to pair with the above two, but it's how single-game matches work, so there's clearly a lot of merit to it.

    4. I will note that usually when we cube draft there are four of us and we play one game against each opponent (with extras if there is time permitting) so it is nice having a mulligan system that makes that game feel like it counts.

      A "downside" to this mulligan system is that it makes the average game of Magic much longer. A match with this mulligan system is probably twice as long as a typical match because there are so few non games.

  5. Unstable has been on my mind lately, mostly because I love seeing how cards can push the envelope. That said, the set also pushes the game in some unusual directions that I think are worth pursuing, particularly in the addition of side decks.

    A suggestion I've seen floated for addressing Mana problems is the creation of a land deck, so players choose to draw a land from the land deck or a spell every turn. That's too far in the scale of removing variance, but I think it's an interesting impulse. I'd suggest the following, inspired by Split Screen:

    Players present two decks of the same size and shuffle them separately. If a player would draw one or more cards (including their opening hand), they choose one deck and draw all those cards from that deck. If anything refers to your library, choose one of your decks for it. Between games players may move cards between decks and sideboard so long as the decks meet the criteria listed above when presented.

    This keeps draw step tension, emphasizes deck construction, while giving players more agency with their draw steps. Maybe it's too un-Magic-y, but it's something I'm interested in trying with my friends.

    1. It's definitely a variant format if not an entirely different game. But it's also something I've noodled about. The hard part is finding the rule that determines how you can split cards between your decks. All land in one deck and all spells in the other is a non-starter. Can a GU deck put all its forests and green spells in one deck and all its islands and blue spells in the other? Is that a problem?

    2. Actually, game with five decks, each with the respective land and color of spells, might be super fun. I may work on a variant along these lines!

    3. I think drawing the opening hand entirely out of one deck solves that without needing a ton of extra rules. Keeping an all land or all spell hand isn't ideal, even if you're certain to draw cards of the other type every turn. Same with splitting colors entirely. Rather, I'd imagine you'd wind up with an Early game/late game deck where cheap shield and more land go into your first deck and less lands and expensive spells go in your second. Transitioning and building between those decks seems skill testing and like it would smooth out non games more than a mana solution alone.

    4. I failed to extrapolate from "when you draw, only draw from one deck" to the opening draw. That is a clever solution, but I think a lot of players would be fine drawing 7 spells when they can guarantee their next 4-8 draws are lands (even if they can't guarantee which lands they are). I sure would.

      Keep thinking on this. I sense there's a solution, and I'm confident you're the guy to find it.

    5. Maybe! And I'm fine with giving risk adverse players options to in deck configuration. Magic has always allowwd players to play a single color to reduce Mana problems, but the color pie hopefully checks the power of such strategies. The risk of color screw similarly checks the land/spells splits, my theory creating only goes so fast ultimately.

    6. Players present two decks of equal size: An early deck and a late deck. Players draw their entire opening hand from the early deck and continue to draw cards from the early deck until they decide to replace it with the late deck at the beginning of their turn. Players only ever have one library at a time and once you've moved from early to late you can't move back.

      It might still be too good for combo (30 card deck? Yes please!) but it captures what I'm going for.

  6. This is a fascinating question.

    I mostly play casually, in fact. And we often say "oh go on, mulligan to 7 again" if someone is unlucky (and isn't being competitive in planning to take advantage of it).

    One thought is, would the variance of "just lose" be better if the average land count were higher?

    Another thought is, I used to dislike land-specific mulliganing rules because they seemed to enshrine the *usual* deck (i.e. mix of lands and creatures) as the deck preferred by the rules. e.g. if you have some wacky land-only or nonland-only deck, the usual rules don't work, even though magic should let you build decks out of anything.

    But now I wonder, that's so normal in basically all decks, would a rule that allows a mulligan specifically for having a particular number of lands (or a slightly easier mulligan) be preferable after all?

    I still wonder if there's any way to make the "specify when you draw lands and nonlands" work, without allowing everyone to just curve out perfectly. Or have a rule that you could pay 2 life to get a cavern instead of a card under some circumstances?

    1. I was designing a tcg of a sort this summer with the mulligan rule that you can discard any number of cards from your starting hand to draw that many new ones once. This cost was significant put because your deck was your life total.

      Anyhow, there's a fundamental A-B mechanic in the game and I was looking at ways to enable that, and one that occurred to me was the idea a mulligan needn't be restricted to your opening hand (as long as there's a limit or a cost): Players could mulligan every turn if they wanted.

      GY strategies make that too good to directly port to Magic, as does a lower risk of decking, but the principle could be applied. Maybe each card you mulligan costs 1 life and is exiled.

    2. In Eternal, you can mulligan only once, but when you do, you get a full hand of 7 cards that is guaranteed to have between 2 and 5 lands.

    3. I like this a lot! What about:
      If your deck contains >= 33% land (20 lands in 60 card, 13 land in 40 card), you may reveal hands with 0, 1, 6, or 7 land from your opening hand and take a free mulligan back to 7.

      This might incentive players to run the absolute minimum number of lands and just play the game with their guaranteed 2+ lands from opener though. That would be good for aggressive decks.

    4. The 33% rule is both prescriptive and hard to enforce (even as modest as it is). And yet something like that does seem necessary if we're going to allow free mulligans. Emphasis on the plural there. I think a mulligan rule where players can mulligan once or not at all regardless of their hand wouldn't need stipulations (but has other issues).

  7. Several comments suggested more generous mulligans, but raised concerns about combo decks. Here's a non-intuitive but powerful way to compensate for that:

    Reduce the duplicate card limit in Constructed formats from 4 to 3.

    The amazing thing is that this accomplishes almost all of what Wizards should want. It gives midrangey decks a buff and combo decks a nerf. It makes deck construction and gameplay more interesting. It makes super-greedy mana bases harder to construct.

    The main con, of course, is finance. The demand for any given card just decreased by 25% and it's not clear how much of that would be made up by players switching to currently less played cards.

    Why is this relevant to mulligan decisions? I give you: Commander. A free mull to 7 turns out to be much less abusable when you have a 100-card singleton deck and can't dig for key spells. Obviously going from 4-ton to 3-ton in 60-card decks is much less radical, but this is the intuition for why reducing duplicate count could help support friendlier mulligans.

    1. My concern is that you'd be simultaneously making players stretch for 6 extra quality cards while decreasing the redundancy of their Mana base. Normally I'd imagine that leading to more non-games, but it depends on the power spread between the best card in your deck and the worst. An extra free mulliganing is huge to help, but it still seems really good for combo.

    2. My theory is that combo is most dependent on a few key cards and has tons of the power spread you describe, so it's most likely to feel the pain from this. Though you're right, this approach might work better if Wizards were to print more "Constructed playable" cards per set.

  8. Great ideas! I like a lot of the proposed rules changes, and the general thought that, yes, the rules change would affect more things than just mana screw/flood, but that's acceptable. Probably not a risk WotC wants to take though.

  9. From the card based solutions, I think lands we might consider to be too strong solve some of these problems.

    Easy Peesy Cycling
    T: Add U to your mana pool.
    Cycling 3.

    Just gotta make better than basics!
    More radically, add it to all basic lands:
    Basic land - Island
    T: Add U.
    Cycling 3.

    The reverse problem:
    Library Wish
    At the beginning of your end step, if CARDNAME is in your library, you may pay {2}. If you do, search your library for CARDNAME, reveal it and put it into your hand. Put a card from your hand into your library, then shuffle it.
    T: Add {C} to your mana pool.

  10. I'd love to playtest this: All cards have cycling {2}.
    My guess is it's busted with madness and graveyard strategies, but otherwise does more good than harm.

    1. I think I'd go with "All lands have ETB Scry 1" if I wanted to try out a radical change. It is a lot less dangerous in terms of interactions like you mention, and helps both screw and flood.

    2. I think my brain would melt.

    3. Graveyard strategies boost (and New Perspectives combo in Standard :) is a solvable problem:

      Safe cycling {3}: Put this card on the bottom of your library, draw a card.

      I don't think this is abusable by anything (at least on lands, maybe Grenzo for creatures but that seems acceptable).