Thursday, June 2, 2011

CCDD 060211—What If Magic Were Always Fun

Cool Card Design of the Day
6/2/2011 - Last week, Sam Stod wrote an excellent article on the importance of randomness and — more specifically — land screw in Magic. It's not the first article to take this position; Mark Rosewater and more than one developer have discussed this as well. I'm going to take a stand here and call them all wrong.

To summarize the major points supporting the other side: Randomness is necessary, land-screw allows players to beat better players occasionally, and the land system rewards better deck builders. All of those are true statements and important considerations for the game, but I claim that none of them depend on the system being as flawed as it is.

Randomness is necessary — A game with no randomness is not a game at all, it's a puzzle. Randomness takes many different forms in games: drawing cards from a shuffled deck, rolling dice, spinning spinners, and throwing balls or flicking discs. Note that I'm being lax in my terminology, by 'random' here I really mean 'unpredictable.' Diplomacy is a game with none of the elements we traditionally think of as random, but it remains massively interesting because of the challenge of predicting what other players will do. Chess is a game for the same reason, but to a lesser extent since you're only predicting what one other player will do.

Magic is not lacking for randomness. You could remove the randomness of card draw by handing each player the same deck in the same order and you'd still have a game just from the way those two play their decks (and who goes first). It wouldn't be a terribly good game, but I'm proposing something much less radical: Removing just the randomness of land-screw from the game. There are a few ways we could do this and I'll come back to those in a bit, but the key point is that the game would still contain all the unpredictability of player choice as well as the random order you would draw spells and possibly the random order you would draw land. That's plenty randomness to keep things interesting.

Less-skilled players should be able to beat more-skilled players — But to what extent? Not nearly as much as The Man has been telling you. How likely are you to beat Kasparov in a game of chess or Michael Jordan in a game of basketball? Barring an act of God, it's not going to happen. Likewise, we shouldn't be concerned with non-professional Magic players being able to beat the Finkels of the world. Still a player who's a 6 out of 10 in Magic skill should be able to beat a player who's a 7 out of 10 at least a quarter of the time. While playing and losing to players better than yourself is the best way to advance your own skills, few players can lose ten games in a row without being demoralized and these rare victories can push us to keep trying.

Unfortunately, Magic is so random that players — regardless of skill level — lose games through absolutely no fault of their own on a fairly regular basis. There is no joy or pride in winning a game in which your opponent mulligans to five and never draws another land so there's no emotional upside for the player who walks into the win. In addition, there is a great deal of frustration for the player who has to sit helplessly as his deck refuses to play nice. The four biggest reasons players quit Magic are the cost, the frustration of automatically losing 1 in 4 games, not having enough locals to play with, and the time commitment. I'm guessing, in that order.

Remove land-screw from the equation and players will only lose for one of three reasons, all of which are much more palatable: Playing a deck that is prey to your opponent's deck, drawing your spells in a less optimal order than your opponent, or being outplayed. This would reduce a player's reach in terms of how much better an opponent can be and still have a reasonable chance of beating them. I believe that gap is far too large currently and comparing Magic to any other game of skill will support that belief.

The land system rewards better players — There is definitely skill in knowing how many land to run in your deck, how to build an effective curve, when to mulligan, and when to run cards that are less powerful but will help you in the event of land-screw or flood. That's a good thing. Furthermore, the existence of land and the spells that help you accelerate or fix your mana are integral to Magic and removing them would result in a less sophisticated game. As you'll see in the next section, I'm not remotely interested in any solutions that involve removing or replacing land from Magic.

There are a number of solutions that don't mess with land but help mitigate or eliminate land-screw that require just as much understanding and reward skilled players appropriately. I can't really prove this point without covering a few so here they are:

Two decks — I've had this idea for a long time but I know that many others have as well. It's pretty straight-forward: you shuffle all your spells into one deck and your land into another. Whenever you would interact with a deck, you choose which. You can draw from either one each turn, Rampant Growthing from one and Mystic Tutoring from another, but your opponent can mill either one. This path has some eccentricities that would have to be addressed. Traumatize would need to affect both decks. Goblin Charbelcher is right out unless there are minimum sizes for each deck. We probably need minimums anyhow just so you can't run a seven-spell / 53-island combo deck.

For some reference, I've done something similar in Mini-Master and that works great. Rather than shuffle 3 of each basic land into the booster pack you open, you just shuffle up your pack and set three of each basic land aside. Whenever you draw a card, you can draw from your deck or you can draw a basic land of your choice. This is a very different scenario than 40 or 60-card Magic, so there's no direct correlation for success, but it does support the possibility.

This solution has a lot of potential, but many pitfalls. We totally eliminate land-screw while keeping most of the basic skill-testing concepts and replacing others, but how fiddly do the final rules have to be to keep the format from breaking because of a given card or strategy? I like the flavor of being able — as a planeswalker — to decide each turn whether you want to remember one of your spells or re-establish one of your mana bonds each turn, but I also see that this is a huge change to the game. It may be fun and is definitely worth trying, but good luck convincing tens of thousands of Magic players to convert to a new rule so fundamentally different.

Abundance — Another solution that I'd like to try tonight if my draft group is up for it is giving every player a one-off Abundance draw each game:

So if you're shy on land, you can pop this to make sure you get at least one land in your next draw. If you've made it unscathed into the middle game, you are rewarded for your good land skills with one guaranteed gas draw. If you're in the late game and you're down to your last chance to top-deck, this thing will at least make sure you don't draw a useless land (y'know, or that you do if you're playing Baloth Woodcrasher).

Now a player who knows she can Abundance her next draw may be more likely to keep the sketchy one land hand that would have been a toss-back otherwise. I'm not sure that's a good thing, but I'm not sure that's necessarily a bad thing either. It sounds like it's just a different kind of evaluation.

Maybe each player should get an extra one of these things each time she mulligans. Taking a mulligan never feels good and it's very hard to convince less experienced players to take mulligans even when they only have one land or one spell, but knowing that you'll at least be able to draw spells or lands as needed for an extra turn helps level off the cost of a mulligan.

Testing may prove one or both of these versions bah-roken, insufficient or simply too alien but they're definitely worth trying out.

Something else — I'd be interested to hear any other solutions you might have that don't involve ditching or replacing land all-together. I'm confident that there is room to improve Magic by alleviating land-screw, but it may take a while to find a solution that fits. How many other games do you know that you just straight up lose 1 in 5 or so games regardless of how you play? Magic is such an amazing game, that it's great despite this flaw, but that doesn't mean this isn't a flaw. Don't let anyone tell you your soup wouldn't taste better without the fly in it.


  1. Are you making an honest estimate that 20-25% of games played are lost directly to land screw, or is that hyperbole? And is that estimate based on a tally of some sort, or is it a guesstimate? (Just to be clear: I'm asking honestly, not sarcastically.)

  2. Why don't you and the Goblin Artisans crew play a bunch of Constructed/Limited games with various rules to minimize mana screw, and let us know how it goes? It'd be an interesting project, and I'd love to hear the results.

    My suspicion is that a solution like "two decks" probably breaks the game because it dramatically alters the power of so many cards. As you suggest, combo would be more consistent. Multicolor decks are easier to play. Countryside Crusher is the nuts, as is Oracle of Mul Daya. Landfall is ridiculous.

    "Two decks" also breaks the mana system. Seven drops are much more powerful than six drops, since you have to wait 2-3 turns for the extra land, but now you're guaranteed seven lands on turn seven every game, if you want it.

    "What If Magic Were Always Fun" is more palatable, but would also be abused. In decks like 43 Land (17 spells) or Dredge (10-12 lands), it's a potent pseudo-tutor.

    I think Magic has too many carefully developed cards to just give you free resources or deck manipulation at this point. Any solution should have a non-trivial cost, but, as you say, one that a beginner doesn't mind paying.

  3. Dan, "1 in 5" is a guesstimate as I haven't done any actual tracking and it's more for the beginning player than the advanced player who has probably mitigated it down to "1 in 10" (again pure guesstimate). Even so, I've never played another game, auto-lost, AND returned to that game. Regardless of the rate, it's a critical flaw, not a trivial flaw and not a feature.

    Anonymous, I definitely intend to test at least the second option. Didn't get too last night b/c the draft was crazy with a dozen new players, but I will test it out.

    CC and OoM are two more cards that break in the two-deck system and indicative of many more (Future Sight comes to mind). I'm fine banning or errata-ing one or two cards, but a dozen or a score just sounds like a mess.

    I would be very curious to see what the most extreme use / abuse of "What if Magic Were Always Fun" could be. Particularly w/ and w/o the mulligan bonus.

  4. I don't think the 2-deck method would feel like Magic - you lay lands every turn until you get 6, then you draw gas every turn.

    I wish there's a way to just eliminate games where you draw 10 lands in a row.

  5. The "Abundance" idea is broken because the designer assumes players will play decks with a balanced ratio of cards, which is the same reason why Oath of Druids is broken. For example, Manaless Ichorid plays one single land, Bazaar of Baghdad, so this rule would allow it to always start the game with it in play.

  6. Having played so many CCGs without mana screw, I gotta say, it's much better with it.

    You focus only on the extreme aspects of mana screw; you either are screwed or flooded and it costs you the game.

    What you're not paying attention to is that mana screw doesn't always determine the outcome of a game. Sometimes it just determines which cards you play that turn.

    Much more likely than a win or loss is the moment where you have a 6-drop in hand, but only 5 lands, and thus, that turn you play a 3-drop or 2-drop, or both, or play out a combat trick. It's not a binary situation of "it messes up the game" or "it's invisible". There's subtlety to the unpredictability, that is lost terribly when you are guaranteed your resources every turn.

    In a CCG, when you are guaranteed your resources each turn, the early game is always the same. LORD OF THE RINGS, STAR TREK (both editions), BABYLON 5, THE X-FILES, DUEL MASTERS, VS SYSTEM, OVERPOWER... all these games had elements that worked and didn't work, but all of them had guaranteed resources, and everytime I played them the initial turns were always the same. And how dull it was. Top decking didn't exist, really.

    The most common result of the land system in Magic is that it makes you play outside the curve. This is far too valuable to lose.

  7. I absolutely agree with you, Bass, which is why I'm not interested in any solutions that entirely circumvent the land system or random draws. I haven't tried the two-deck method, but I don't think it's even worth wasting a single playtest on because it removes a lot of the elements that you correctly describe as being important to the game.

    The once-per-game Abundance solution (which I am now inclined to call AbdundOnce) doesn't mess with this dynamic (much). I've now experienced an eight-man swiss draft using the rule and can say that it helps alleviate the all-land/no-land problem a bit, but it doesn't remotely eliminate the possibility.

    Similarly, it has allowed players to keep otherwise unkeepable hands in a few marginal cases, but mulliganing is still necessary and skill-rewarding.

    It has also been used, unsurprisingly, by players with good hands to guarantee a land-drop on an important mid-game turn, or to secure on more piece of gas. I'm not sure that's a bad thing, so long as it isn't quite as great for players with good hands as it is for those with weak ones. Needs more testing.

    I'm not prepared to call it a success after just one night, but the first test definitely suggested a positive impact. I encourage you all to try it out and let us know how it went.

  8. I think the two decks and the abundance all have the same problem: if you want to draw a land, you draw a land. That fundamentally ignores the whole reason the land system exists; you don't get to choose.

    The best method for dealing with screw is cycling. You lose a card you have to have a chance at drawing the card you want. Landcycling and even spellcycling are tutors and work fine to lessen the screw too, but only in small numbers. These are tutors and the problem with too many tutors in an environment is it reduces variance (removing the land system inherently reduces variance).

    If you want a variant, say all cards have "Cycling 3" unless they actually have Cycling. Cycling isn't a tutor, it's a second chance at drawing. Landcycling on everything would be as problematic as the two-decks idea or the VS system/Duelmasters idea of any card becoming a land. But Cycling on everything lessens screw by giving you more draws, but doesn't actually alter the randomness of the draw.

    Again; I think the fun of the land system is in finding new ways to lessen it, within each deck. I feel that is a huge aspect of deckbuilding, and if there is a blanket-answer to it in a format, you will inherently lose more variance than you will gain, and it will be less fun.

    It's much more exciting that Shards lessens mana-screw with cycling, Zendikar with landfall, Rise of the Eldrazi with level up, and Scars with phyrexian mana, than if every block used the same mechanic to try to solve it.