Monday, May 20, 2019

At What Price Variance?

Unpopular opinion: Mana screw is bad.

The number of non-games that Magic’s core system regularly produces would seem to indicate that most players would get frustrated with the game and leave. Yet, Magic has one of the highest player retention rates of any game invented in the past 50 years. Why is that?

Mark Rosewater frequently states that mana screw (and flood, and etc.) are unfortunate byproducts of the game’s variance, but this doesn’t illustrate the fundamental difference of how Magic works compared to other games. You see, while most games labor to provide an optimal play experience each time, Magic throws a billion gamestates at the wall and sees what sticks. And surprisingly, this approach works really well, but it only works because Magic has the commitment to create a level of randomness that sometimes produces bad situations.

Strategies and Tactics

“As long as players can continue to encounter meaningfully different game states, they will enjoy making decisions on the fly. If they decide that they have seen everything your game has to offer, this is the moment when it transforms from chess to tic-tac-toe.” –James Ernest, “Strategy is Luck” from The Kobold Guide to Board Game Design

Most games have strategies.

Chess has a number of optimal openings that have to be memorized, as well as several deterministic ways to checkmate the opponent’s king in certain endgame situations.

Scrabble has a list of high-scoring short words that has to be memorized, and anything other than the optimal gameplay of attempting to bingo as often as possible while squeezing QAT onto a triple word score won’t be effective in the long term.

Texas Hold’em has a big chart of probabilities and rules-of-thumb as to whether you should call, fold, or raise with various hole cards and from various positions.

Magic doesn’t have strategies. Instead, it has tactics.

There are certain heuristics that most decks can follow, like “play lands” and “try to reduce your opponent to 0 life before you are reduced to 0 life”, but in the end there are no 100% prescribed strategies because what you have to work with is different every time. You instead have to read the situation and make the best decision, which might be different every game.

As an example, let’s take Ad Nauseam, one of the most rigid decks in Modern. The only way the deck can win is to cast Ad Nauseam into Lightning Storm or Laboratory Maniac– it has, at least preboard, literally no other path to victory. Everything in the deck that isn’t a win condition is a way to get to your win condition faster or more consistently. And even then, things can go awry. You might have to use Angel’s Grace as an emergency Fog against the Burn player who got a good start or alternate the target of your Lightning Storm against a pesky Spellskite. The primer for the deck on MTGSalvation is filled with corner cases and obscure uses.

The reason that even Ad Nauseam players can’t rely on doing the exact same series of moves every game is because their strategy is chopped up and given to them at random along with their resources. Large deck sizes, combined with this variant but swayable combination of lands and spells, mean that matches that are fairly rote can always have the occasional surprise.

The Variance Is Out of This World

“A Game of Infinite Possibilities” –Tagline for Cosmic Encounter

One of Richard Garfield’s influences when creating Magic was the game Cosmic EncounterCosmic is another long-lived game, celebrating its 42nd birthday last year, and seeing printings from five different publishers. Its current edition with Fantasy Flight has resulted in six expansions and a longevity of 10 years.

Cosmic, like Magic, subscribes to the theory of “give the players a huge number of play experiences.” Most famously, each player is assigned an alien whose special power alters the rules in some way. The Zombie never dies. The Leviathan sends entire planets at its enemies. The Sniveler can whine about being behind so the other players are forced to let it catch up. In each game, players discover how aliens interact, what’s a threat, and how to win.

But alien powers are only scratching the surface of Cosmic’s commitment to variance. Opponents are chosen at random through a “destiny deck” – this also stops players from always picking on the leader until it’s the leader’s turn. The strength of an individual hand can vary wildly, from something with the most powerful attacks and strong “artifact” cards to a grip full of attack 03s and negotiates. But most importantly, this variance can be turned into a win from the worst position if you’re crafty enough.

There certainly are games of Cosmic that end badly. Sometimes, you never get a chance to use your alien power before the game ends. Sometimes a player’s position is so strong that they can single-handedly win 5 colonies based on their alien power and opening hand. Sometimes the game drags along forever because the Lightning constantly picked off people’s foreign colonies. But in exchange for F-grade games, Cosmic grants the player group A+ games, something that would not be possible if it wasn’t willing to have so much variance. It’s not hard to see the progression to Magic.

Commanding Chaos

“The play is much improved over traditional chess because you don’t need to analyze or memorize any book openings. Therefore, your play becomes truly creative and real.” – Svetozar Gligorić on Bobby Fischer’s randomized Chess960

If you want proof that Magic’s biggest asset is its variance, look no further at the astounding popularity of the Commander format. More importantly, Commander demonstrates that high variance only has true benefit if the games are markedly different from one another.

Commander’s format – giant, singleton decks – produces decks that are about as inconsistent as can be while still falling vaguely within the boundaries of reason. Persistent commanders and color identities don’t demonstrate a specific strategy as much as an outline for how the player could win the game. For example, if my opponent has a Mizzix of the Izmagnus deck, I cannot think “ah, this player is going to win using the card Expropriate”; I can only think “this player is going to win using any number of expensive instants and sorceries.” 

Since these victory outlines are different from deck to deck, each game feels different in a distinct way that completely randomized formats like Battle Box* aren’t as capable of accomplishing. This creates a variance to expectation ratio significantly higher than other Constructed formats but that doesn’t lean completely into a random hand of singleton cards.

* Battle Box is a really fun format that I like, but I have trouble remembering anything I did in the format while I can remember Commander games from several years ago. This is something I wrote about more in my article on impact.


Magic is as popular as it is, not because it’s the perfect strategy game, but because it makes a commitment to the game it wants to be. It can be difficult to reconcile the fact that a game can be a good design even with real flaws; the issue is that most of the time, games have to have flaws. There is no clever outside-the-box solution that will give you huge variance and perfect strategy, or complexity and ease of mastery or freedom and not digging yourself into a hole. The best games pick the ratio of these things that are strongest for their needs, and Magic’s variance ratio has proven over 25 years to be the formula that makes it the most interesting.


  1. I'm not sure what your definition of 'strategy' is, but it's quite unlike mine: A plan of action. In any game, I have a plan to win the game from the first moment information is revealed until I win or lose, or concede because I cannot fathom a winning plan. In Magic, I have a plan before I even sit down.

    The details of how I execute that plan shift game to game—and within a single game, if it's dynamic enough—but that doesn't mean I don't have a strategy. Strategies can be vague (gain control of the game / be as aggressive as possible / claim the most resources) and they can be specific (use Ad Nauseam to fuel a lethal Lightning Storm or empty my deck for Laboratory Maniac). They can also shift during a game, as we adapt to our opponent's play.

    That games play out differently doesn't mean there's no strategy, only that there isn't a single sequence to execute my strategy.

    The difference between Commander and Battle Box is strategy: I can build my Commander deck around an ability and expect to hammer certain kinds of plays; I can plan to be aggressive or controlling, disruptive or unpredictable.

    I thoroughly agree that Magic is flawed—largely because there is too much variance—and that it overcomes those flaws by providing a variety of rich experiences.

    1. In some ways, my use of "strategy" to refer to rote game-winning moves was hyperbole; I was commenting on the difference between a strategy as a long-term plan to win the game and a tactic a short-term way to accomplish a goal, and that Magic focuses more heavily on the short-term because there'll be situations that you have never encountered before.

      Magic absolutely has an element of strategy to it, I've listened to too much LR to forget that "have a plan" is an important way to be good at Magic, but the components of that plan are so variant from game to game that it frequently relies on execution on the spur of the moment rather than sticking to a plan.

    2. I don't think I agree that Magic focuses more heavily on short-term tactics than long-term strategies. Almost every short-term decision in Magic is made (or should be made) in the context of the long-term goal. That's exactly the point of the MTG strategy staple "Who's the Beatdown?". Perhaps the most important Magic 'tactic' is card advantage, which in itself is only an advantage in the long-term.

      Deckbuilding as a whole certainly leans to the more strategic side of things, while sideboard hate and silver bullets are more tactical. Transformational sideboards like those of Mardu Vehicles or Abzan Midrange are also much more strategic in nature. The whole goal of those sideboards is to make sure the strategy your opponent plans for matches up poorly against your own.

      I'm not saying that short-term decisions don't happen and that you don't have to deal with the unexpected at every turn, but the context of all of your short-term decisions should be relative to your long-term goal.

      Do I develop my board or pop off a removal on my opponent's flyer? Well that depends on if I'm looking to pressure my opponent's life total or simply preserve my own. Do I think I can win the long game or do I need to race to have a chance to win? Who's the beatdown? There is no tactic without the strategy.

    3. I classify Magic as being highly strategic and highly tactical. The two are in now way opposed, like luck and skill (which Magic is also full of).

    4. My point is more that you can’t go into a game saying “I will pop off a removal spell at my opponent’s flyer” like you can go into a game of chess saying “I will use this opening”. Improvising your play to better fulfill your strategy is much more important in Magic than in lower-variance games.

  2. "This is something I wrote about more in my article on impact." Can you link that article? The blog doesn't have a way to search by author, at least not that I've found.