Magic: The Gathering Drive to Work Podcast - #1198: Variance

Episode Date: December 13, 2024

Variance is an important aspect of Magic design (and game design in general). In this podcast, I explain what it is and how we use it.

I'm pulling away from the curb because I dropped my son off at school. We all know what that means. It's time for the drive to work. Okay, so today's topic is variance. This was requested by someone on my blog. And interestingly, after someone requested I do a podcast on variance, someone else said, hey, but he's already done a podcast on randomness. And then I had to say, Hey, wait a minute, variance and randomness are not the same thing. So today I'm going to explain what variances, why it's not randomness

or it is connected to variance. It's a randomness, but they're different things. And I will talk about how we in design interact with variance, how we raise or lower variants. So I guess let me start with the most important thing on a podcast about variants. What is variants? So variants is talking about how often games play out the same. The higher the variants, the less they play out the same. The lower the variants, the more they play out the same, the lower the variance, the more they play out the same.

So why is that important? And real quickly, I guess, let's say you're playing a game of Magic, okay? I have the same deck and I play 10 games. High variance would mean, man, those games are really, really different from one another. They played out completely different. Low variance is, I was basically doing the same thing every game. Now, why do we care about variance? Well, the answer is, and I'm gonna make some generalities here, but they're generally true. High variance tends to be more fun. Why? Different things happen.

You don't know what to expect. You know, or at least when I say more fun, I mean from a casual standpoint. That high variance means just I'm surprised more. Things happen I don't expect more. There are more moments that are kind of unique to each game. There are more moments that are kind of unique to each game. Low variance usually means it's a more skill testing environment. Why?

Because I have a better understanding of what's going to happen. That the decisions are made are more based on having knowledge of things I understand. For example, chess is a very low variance game. You start with the same pieces in the same position every game. That doesn't mean games can't change game to game, but your understanding of what to expect is a lot higher. You can study opening moves because the pieces

always start in the same place, the same pieces in the same place. Now I do want to stress that it is not as if high variance doesn't have any skill testing to it. I would argue that there is some skill testing even to high variance, which is adaptability. The thinking on the fly and doing things that you don't expect, there is a skill to that. So but the higher the variance, the more instability there is. The harder like in a game where you want high strategy, you normally want low variance because

you want people to have a pretty good expectation of what's going to happen. So anyway, these are tensions. There are reasons you want high variance. There are reasons you want low variance. And different players will want different things. That if you are playing in a really casual game of commander, let's say, hey, you're looking for lots of high variance.

You want crazy things to happen. You want games to play out differently than previous games. Cause casual games are more social. Like you want more of the amazing moment to happen than it matters that it's as strategic as possible. But on the flip side, let's say you're playing at the pro tour, you know, and there's lots of money and prestige on the line.

Hey, you want the better player to win. You want the person that's more skilled to win. And so there's this balance that we have to reach in the sense that every magic set wants to have some amount of high variance, some amount of low variance in that there's this tension. The other big thing it has a lot to do with where you're aiming at, but we'll get to that. Okay, so let's start with the first question. How exactly do we raise or lower variance?

I would say there's two major categories. Category number one is deck construction and category number two has to do with individual card designs. So let's start with deck construction. There are three main things that determine your variance in deck construction. Number one is how big is your deck?

The larger your deck, the more variance, and the way to think about this is how often do I see the same card well if I have a 60 card deck I have a 1 in 60th chance of drawing the card every turn if I have a hundred card deck I have a 1 in 100 chance of any one actual particular card I mean we'll get into how many you can run in a second. That's also important. But the simple idea is the larger your deck, the larger the variance.

If I'm playing a 250 card deck, there have been formats, in fact, one called 250, where the deck was 250 cards, super high variance. There's a lot of different things that can happen here. If I have a small deck, let's say for example, I'm playing limited, where I have a 40 card deck. Wow, that's probably the lowest high variance. There's a lot of different things that can happen here. If I have a small deck, let's say for example, I'm playing limited where I have a 40 card deck. Wow, that's probably the lowest in variance. I have the lowest uniqueness of what I'm going to draw. Okay, number two is how many different copies of a card

are you allowed to have in a deck? In standard, you can have four of. In commander, you can only have one of. It's called singleton. There are four matches, though, not super popular ones. For example, early magic didn't have any deck restrictions. Very early magic, didn't have deck restrictions. If you wanted to play mostly plague rats, you could do that.

That is the lowest variance. You know what? I'm probably gonna draw a bunch of plague rats. If most my deck is plague rats yeah you know what I'm gonna draw plague rats. So the more copies of the same cards you're allowed the lower the variance the less copies of each card you're allowed the higher the variance. Standard with four of has a lower variance than commander with singleton one of. The third thing that impacts it has to do with how large is the pool of cards.

The smallest pool we've ever, well the very very smallest pool is limited. I'm drafting a single set. Well all the cards are from that single set. And the widest is something like vintage or eternal format in general commander legacy and stuff, where I have access to most of the cards in magic. We've made like 27,000 cards.

That's a lot of cards. Whereas in, you know, in limited, you know, you have like 300 cards. So there is a lot. So now let me explain why that's important. Obviously, when you talk about your deck size, and you talk about how many copies of a card you can have in the deck,

that is straight up sort of just how often I draw the card. Right? That is like, for example, we'll compare standard to commander. Standard four of 60 card commander 1 of 100 card. So if I have a particular card in the deck that I want to draw in a standard game 1 out of 15 chances I have to draw that card. Assuming I play 4, assuming it matters to me I need to draw that card. Assuming I play four, assuming it matters to me,

I need to draw the card. I put four of, one in 15 chance. In Commander, at best I have one in 100 chance, right? So you can see the chance that I'm drawing the card I need in standard is significantly higher than me drawing the card I need in Commander. Now there is one exception to that, which I'll get in a second, which are format rules.

And I'll get to that in a second. The issue about having a smaller card pool is the less unique cards that exist in the environment, the more consistency there is in the environment. Meaning, if I want to do a certain effect, let's say I want to do a certain effect, let's say I want to destroy all the creatures. In any one set, at most, usually at most there's one mass creature removal. Maybe, maybe there's two in the sense that there's a white one and

there's like a red one that does a bunch of damage. But even then, usually there's one mass creature removal. Which means if I'm playing and I can see what colors my opponent is playing, I might know whether they have mass removal. Hey, in this environment, there's a white mass removal and maybe there's a red that does a bunch of damage. Hey, I'm playing against a black green deck. I know in this environment, black doesn't have mass removal. So like you start having more predictability of what's happening because the smaller the

card pool, the more predictability you have on what's happening because the smaller the card pull, the more predictability you have on what could happen because you know the available cards. So variance does go down based on that. Okay, the one other factor, and this is not deck construction, this is format, but deck construction and format are tied together. Normally, your deck construction rules are made by your format. Your format defines your deck construction rules. It's one of the things formats do. The one other thing formats can do is they can have special rules.

My example here is Commander. In Commander, you get your Commander. Your Commander starts outside the game. As soon as you're able to play your Commander, you can play your Commander. That lowers variance, right? So one of the things about commander, the format is general, super high variance. It's singleton, it's a hundred cards that I don't know what I'm going to see. But to provide, to sort of lower the variance a little bit, they give you the commander and the commander is super low variance, but as low variance as it comes. Why? I know when I have

enough mana I can play that card every game all the time so that's very low variance. So the idea is commander has a very high variance set of rules but one really special rule that lowers the variance. And the reason there is the other reason you want low variance not just it's more skill testing but also you want some predictability what's going to happen. If you truly have high variants, so imagine playing commander, but you don't have your commander is just very like, there's not the, the, the games will be a little too all over the board that you want to kind of focus a little bit. And having the commander allows you to focus your deck and say, oh, my deck's about this thing.

And the reason I know that is the one card that is not high variance, I know what it is and what it does. Okay. That has to do with deck construction a little bit with formats. Now a lot of what we care about in R&D is individual card side variance. How do we in the individual card effects affect variance? It's not by the way we don't care about deck construction rules and stuff,

but that's more determined by format. So when we're designing by format, we understand that formats have deck construction rules and can have other rules like the commander that will impact variants But at least when we're designing for a format that stuff is locked meaning if we're making stuff for commander Hey, we know the rules of commander. We know deck construction commander like we that's a given not that it doesn't impact things It does and how much variance?

You have your format has a huge impact on how much variance you have uh, you have, your format has a huge impact on how much variance you have. Um, but as card designers, one of the big questions is, okay, we're designing magic cards. How do we in the actor designing cards impact variants? And so we spend a lot more time on that. Okay. So basically we can do things that raise variance or lower variance. How do we do that?

The key is the more we let you play out the games the same, that is us lowering variance. The more we add in elements that make games play differently, that is raising the variance. We do both. So let me start with talking about how we lower variance. Probably the most famous way to lower variance is tutors. We do both. So let me start with talking about how we lower variants. Probably the most famous way to lower variants is tutors.

So a tutor is I go into my library and I get the card I need. The reason that lowers variants is if you're talking about how much games play out differently, the ability to like, the big variants built into magic is the card draw. Right? You have a 60 card deck that you've handpicked every card, but to make sure like, for example, the highest variance of magic, a format most people do not play or very few people play. So there is a, I'm gonna make up a format

and the format is called exact magic. And the way it is played is It is 60 card for of it's like standard, but You order your deck and then you draw your deck exactly in the order you order it There is zero randomness. Well, sorry zero deck randomness There's randomness in the matchup. But okay, so in that version, oh, I know exactly what's gonna happen. I know exactly in what turn I'm gonna do what. Because it's all I've planned my deck out.

Now maybe my opponent interacts with me in ways that disrupts my plans, that just because I think I'm gonna do something a certain turn, and maybe I won't because there is randomness in what my opponent is doing. There's interaction with my opponent that creates randomness. But Richard Garfield was a smart man and there's a reason that deck games

is the deck's shuffled most of the time. And that idea is it adds some variance. You want some variance. Variance is fun. I'm not saying that playing exact magic wouldn't be fun the first few times you play it, but you'll get tired of it. It's not, it really is a lot more,

the most fun in that is just to see what happens the first time you do it. It is not the most exciting format to replay. The first time you do it, okay, what happens, you know? But it's the kind of format that gets dull pretty quick. So we want some variants and Richard smartly, and like I said, trading, it's not as if card games don't do this.

Randomness and shuffling is a big part of card games and drawing from the deck. But given that, that is the core randomness of magic. That is the biggest element of randomness that we have. So one of the ways that we lower variance is by doing things that lessen the variance of the shuffling your deck. So tutors is a really good example, right? Tutors means okay I gotta go get exactly what I want. Well that is greatly lowering of the variance because you have control, you have exact control of

what you get. We need to be very careful with tutors because tutors more than anything lower variance and that if tutors are too much a part of your gameplay it majorly lowers your variance. Every gameplay is not the same and that one of the things we want, I mean we do want a lower variance sum because we want some skill testing to be in it, I mean we do want a lower variance sum because we want some skill testing be in it but we Magic is more fun. Like the reason you shuffle your library. It's more fun if there's a certain amount of randomness. So Our goal on some level is not I mean if you look at the extremes no randomness is like exact magic I know exactly what I'm getting when high randomness is

You know every time I'm getting when high randomness is, you know, every time I draw it's from what's the format on a digital, what is it called? Where on turn one you draw any card that exists, that's a one drop. And then you draw any card that exists, that's a two drops. So that's like the highest randomness. Like I could draw anything, you know, who, who knows what you're gonna get. So we're playing within that. So tutoring, other things that we do, card drawing in general, the more cards you draw, the more chance you have to see exactly what you need.

Looting, rummaging, impulsing. So looting is draw and discard, rummaging is discard and draw, impulsing is look at the top, end cards of your library. You usually pick one, sometimes more than one. The rest go in the bottom randomly. Any kind of thing in which, you know, any kind of card filtering, card drawing,

anything in which I'm just increasing your opportunity to get the things you need. That also includes land fetching. Land fetching is tutoring. It's a little safer because, look, you have a lot of basic lands in your deck and you're not getting specific functionality. But in general, anything that's smoother your lands

are also lowering variance, right? One of the various in magic is the lands. And we want lands to have some amount of variance. We don't want things to play out exactly the same every turn. The fact that you get different mana in different turns and different colors, that really affects how games play out differently.

So we want some amount of randomness to lands, but there is definitely an randomness that gets too high. Games aren't fun if I never draw a land. Games aren't fun if I get color screwed at constantly. So we do give you some tools to create some consistency, but every time we create consistency, we're lowering variance. Now it's not, and land tutoring is a good example,

or just land fixing in general. Things are not the same. Letting you tutor for a basic land is not the same of tutoring for any spell. Those are both lowering variants, but they're lowering variants in slightly different ways. The reason that the one tutoring we still let you do is like basic land tutoring is, hey, that is not real. I mean, it's lowering the variance and that you more often get to play your spells You less often have games where you can't do anything

But that's a variance rolling to give up Whereas I keep drawing the exact card. I need to play out have my strategy play at the same time. Maybe we'll turn Now there are ways we can lower variance we can let you see hidden information Peeking at the opponent's hand for example lowers variance. That's in a way that we're not real fond of. Hidden information does a lot to make the game more dynamic. The more that you know, the more that you can plot and plan and that just understanding what's in the deck will lower the variance. And one of the reasons for example standard gets lower is because of the nature of standard. As soon as I play a couple of cards, my opponent has a lot more information about what I'm doing

that in standard, if I play two or three cards, especially in a competitive standard, you have a pretty good idea what I'm up to. Maybe you don't know exactly my deck. Maybe I've tweaked a little bit. Maybe I have a few rogue cards, but you know what I'm up to. You know my general strategy. In commander, I mean, given I have my commander, my commander tells you a lot, but if I draw a couple of cards, you don't quite know what I'm up to until you know, a few cards don't really necessarily give away my whole strategy the same way that a few cards in standard will give them a strategy. So seeing any kind of hidden information that includes stuff like milling, any way in which I'm getting more information about what's in your deck gives me knowledge that will

lower the variance. Now that is a little more, so when we talk about variance, I should say some of the variance is just like, how often do I get things? And some of it is how often do I know things? We're much more willing to let you know things than we are willing to let you get things. But even then, for example, early in Magic, we did a lot more look at the opponent's hands effects. And what we learned is, you know what? It's just less fun the more you know.

And that we do a little bit of it. Mostly we do it now with discard because if you want, and it's choice discard, I have to see your hand to choose a card to discard. Or if it's discard where it demands a certain thing, I need to make sure that that's true of what's being demanded. But other than that, we really don't let you look at a hand very much anymore because that just lowers variance in a way that we don't think is fun. The other way that you can lower variance is you can have things that repeat.

So things like flashback or buyback or rebound. Anytime I do something or adventure, anytime I do something where I'm cluing you in to what is going to come, hey, you know, I did this spell and you know that spell is going to happen again, or I did this spell, you know, this other spell is going to happen. The more I give you future knowledge about things, the more that lowers variance. Okay. But it's important to talk about how there are ways in which you can lower variance or sorry, there are ways in which you can raise variance, but not lower skill testing. That's another important thing. Okay, so that's a fine, I said in generally, when variance is low, it's more skill testing, and when it's high, it's less.

How do we raise variance but add skill testing? So doing that is through choice. The most common version of choice are modes, where hey, I have a card that can do multiple things. And doing multiple things means that it's not going to play out the same, that I, you know, different games will play out differently, but because it gives me choice, because I the player can choose which effect I want, I can custom tune it to what's going on in

the game So choice is a nice thing that we can add that lowers variance but increases skill And so that's one of the things I guess it's important to understand is when I talk about things of hey in general Loaning variance raises skill and in general raising variance makes things more fun, but those those aren't a hundred percent true makes things more fun, but those aren't 100% true. Choice is a good example. Well, I think choice makes things more fun and choice makes things more skillful. So it's possible to point in direction of fun and skill testing don't inherently

have to contradict each other. But a lot of things that make games fun, a lot of things that make games skillful at testing can butt heads against each other. So when I talk choice in general, there's a lot of different types of choices. Modes is I have different effects I can choose. There are also some flexibility in targeting. Oh, you know, like if I just let you hit creatures, that's different that you can hit creatures

or artifacts or, you know, that the more choices I give, the more flexibility flexibility I give the more that I can choose and so that's done through modes that's done through targeting it sometimes can be done through sometimes we give you a choice like fabricate was a mechanic from Kaladesh where you could choose whether to get tokens or counters plus plus one, plus one counters. And so like, oh, what's better for me? Also sometimes we can do things, another way to increase variance is we can do things where there's a dedicated choice built into it.

My example might be Miracles is a mechanic from Avacyn Restored where when you draw the card it's cheaper the turn you draw it. So there is effectiveness but we pinpoint when you have to get that effectiveness. Another effect would be impulsive draw. Oh I get XL cards but I can only play it this turn or until the end of next turn. Like I give you the window of when you can do that. I find in general, those type of effects,

raise, they do raise variant, I'm sorry. They do lower variance in the sense that you have less choice where to do them, but that is sort of the opposite of giving you less a choice where I'm restricting what you're doing. It is something that lowers variance, but I think lowers variance in a way that is, once again, when I say skill testing, it's skill testing in that I have to, there is skill testing to figure out how to deal with things in the moment. So that is skill testing.

But by lessening your choices, this is sort of the opposite, right? Where I'm doing something that decreases variance, but I'm not necessarily making it more fun. And there is skill testing there. But it is lessening. On some levels, those kind of effects can lessen skill to a certain extent because I have less options of what to do. The other thing that we can do in the game is one of the ways to lower variance is there's certain types of effects that lower them.

Things like discard, counter spells. Games can play out more the same if I have ability to attack your resources. Land destruction as well, right? That if I counter everything you do, if I make you discard all your cards, if I destroy all your land, there is, there will be a familiarity of the gameplay. Not a particularly fun familiarity. Like in early Magic, for example, there were decks where using Dark Ritual and Sinkhole and Strip Mind where you never get to play land. Well, guess what?

Those games play out really similarly. It's not fun. I would even argue not particularly skill testing, but I mean maybe skill testing for the person that does not do anything. But the idea is there are elements in the game that we can do that lower variance that cause problems and This is a another example where I just is less fun that not getting to play the game is less fun and so that is something that even though and

It is something that doesn't really make fun gameplay and it doesn't really just make fun social play. It decreases options. So we think that a little bit of discard, a little bit of counter spell, a very little bit of land destruction can be useful. We do want to have tools to help you. And so we've not gotten rid of discard, not gotten rid of counter spells. We do believe they can add some skill to the game, but in volume, they actually will pull away some of the skill and some of the fun. So much like I said, choice can

add fun and skill. There are things that can erase fun, that can erase fun and skill. So fun and skill are not, even though they sit at different parts on this particular spectrum, it's not that they're inherently opposites. We can do things that are skill testing and fun. We can do things that aren't skill testing and aren't fun together. Okay, real quickly, let's get into randomness. So the final bit of things is, one of the things that we can do in gameplay is we can make effects random. So what random means is you, the player playing, you have less control over what's going to happen.

I talk about choice being really good. There's random choice. Random choice is, I don't control the choice. I have a choice, I don't control the choice. I have a choice, I don't control the choice. Now, random choice can be quite fun. Exciting things can happen. But it really nibbles away at our ability to make it skill testing.

So a classic example, die rolling. So in the D&D set, Adventures in the Forgotten Realm, or the first D&D set, we added die rolling. Well, one of the cornerstones, one of the main tools of Dungeon Dragons is the D20. And in the game, for those that don't know, Dungeon Dragons is what's called a role playing game. Basically there is a dungeon master who's a storyteller,

and everybody else is playing characters, and then they get to figure out what they're going to do But there are rules about how combat works and stuff like that And so you use the dice to dictate randomness to understand what's going to happen But because there's no winning or losing in Dungeon Dragons if something goes horribly wrong Hey, that might be a fun story moment. Oh, no in the middle of battle my sword broke Oh, that's horrible. Or I got a critical hit. Like if you roll 20, I got a critical hit and all. I did this amazing damage and in one blow, you know, beat one of the creatures we're fighting.

Like, the high variance in role playing is great. It's exciting. But we were adding 20 side of dice to magic play. And the problem there is that high variance like that can cause problems, especially in tournaments, right? We don't want to have a situation where if I roll one, I lose the game, I roll 20, I win the game. Oh, well, whether or not I win the game is not about what I did up to this point. It all has to do with this die roll. Not very exciting.

And so one of the challenges is how do we do die, like when I do unsets, I put die rolling and unsets, I variance is great. I want high variance. One of my favorite cards for example is elvish impersonators. You roll two six sided die, one for the power, one for the toughness. I can get anywhere from a one one to a six six. That's a huge amount of variance. That is not something we would probably do in a normal game of Magic. It's a little too much

high variance or at least in a standard set. That's fine for Commander, fine for other formats, fine for casual formats. But when we were making Adventures of the God and Realm, we ended up making charts. So we actually limited how much like die rolling actually had a few outputs and we could control how often it happened. And even when we did a few like oh roll a 20 something happens, it was small effects that gave you a little bonus but didn't swing the games wildly. And that is the issue with randomness that randomness is a lot of fun and I did an entire podcast on randomness so if you want to listen like there's some general rules about front load the randomness let

people respond to the randomness don't make the game hinge on randomness don't make the game end because of randomness there are a lot of ways that you can and like I said randomness is generally a lot of fun there are a lot of ways that you can, and like I said, randomness is generally a lot of fun. There's a lot of exciting moments where I don't know what's going to happen. And there's a lot of tools of randomness. You can flip coins and roll dice, which are kind of obvious random tools, but also just drawing cards off the top of your library or flipping until you get to a card or randomly pulling a card from your opponent's hand or from your own hand. That, that can add some dynamism that In too high volume can cause problems and tournaments, but there's something really cool. So

Randomness does tie in I said at the beginning. It's randomness in general increases variance Randomness isn't variance but randomness generally increases variance and so when we look at randomness it has to do with how much variance do we want. And once again, where we make something, if I'm making a card for a standard set, I know that this is being played in tournaments. Like I know there's a lot of focus on this in a competitive area.

Hey, we need to control the variance on that card. I'm making something for an unset in which, look, no one's playing this in a tournament or even when we made eternal cards, look, the format that you could play them in was legacy, the power level legacy is way, way higher than standard. So we had a lot more room. No one's playing Elvish impersonator in legacy. I mean, competitively. So the idea is where you're designing the car will dictate how much variance you can put in. The more we're pushing to a high variant format, the more we can do high variance.

And the more it's for competitors, the more we have to watch it. So like I said, variance is interesting in the sense that it's not something that you always want to go in one direction or the other. It's not like magic just wants high variance or magic just wants low variance. Magic wants both kinds of variance, but it's important to understand, like I said today, how are you doing the variance? What's the... Where are you, like... And variance isn't in a vacuum. It's not like I... There's certain things that make it more predictable, like getting land

is generally a good thing to do. And there's other kind of variants where I know exactly the spell that's gonna win me the game That's a bad kind of variance. And so there's there's a balancing between those two things. It's not that we want Total chaos, you know, like everything's high variance and it's not that we want everything prescripted where you know exactly what's gonna happen all the time There's there's sweet spots between that and like I said different formats have different sweet spots so when you're designing cards you have to know where you're tailoring them because it really lets you change what you're doing. But anyway guys that is my talk on variance today so for those that didn't know what variance was or thought that variance was randomness now you've learned something. Hopefully you listened to my podcast because you learned things I hope and because you're entertained. But anyway guys I'm now at work so

we all know that means it means the end of my drive to work. So instead of talking magic it's time for me to be making magic. I'll see you all next time. Bye bye.