I was lucky enough to conduct an interview with Richard Garfield—whom I will assume needs no introduction—for my non-digital game design class last spring at Lawrence Tech. We had a great conversation that (I'm guessing) touches on several of the points in the much-anticipated new book he's written with Skaff Elias and Robert Gutschera, Characteristics of Games. So, without further ado, Mr. Richard Garfield:
Hi Richard, thanks for your time. To start us off, could you summarize your argument about luck and skill?
Sure…though I don’t know if I’d characterize it as an argument, because there’s a bunch of things I believe about luck and skill, but not any one particular thing.
First of all, I don’t think people understand what luck is in a game. They look at a game like chess, and say there is no luck. And, in fact, I think that luck comes from all sorts of places, and really the most useful definition of luck is uncertainty in outcome: as long as you can’t predict the outcome, we’ll call it luck.
With that definition, there’s luck in chess, since your mind can’t take in all the complexities on view there. This doesn’t mean there’s not a lot of skill—there’s an enormous amount of skill—but you take your best guess as to what the best move is, and then you choose it. And, if you're lucky, you choose the right one.
You also see luck in sports. There’s been a lot of studies in baseball and basketball to test the validity of hot streaks and cold streaks, and most of the time, things follow a straight probabilistic model, where people have a certain chance of hitting, and it is a percent that isn’t any more or less streaky than a set of dice rolls.
Secondly, I think game designers have a tendency to design out the luck. Game designers tend to be game players, and game players like things that highlight their skill, and the more luck there is...there can still be a lot of skill, but there’s less opportunity for it shine through.
So what luck can do for you as a game designer...it can make it so that a broader audience can play a reasonable game against each other. In a game like Poker, for instance, anybody can beat an expert player for a single hand, though over an evening it’s much harder. There’s a lot of luck in the game, and obviously a lot of skill, and one of the reasons it's such a broad game is because anybody can compete against anybody, and although maybe they don't win in the long run, they have their little victories. Even a game like Scrabble—a very skill-intensive game—has quite a bit of luck. If you are a reasonable Scrabble player, you can beat an expert if you just happen to be dealt a good mixture of blanks and vowels.
Lastly, luck can be leveraged into variety—that is, if you have uncertainty within the game, you can throw the game into states which players aren't used to, which can create variety.
I actually had a paper recently from a student who is a fairly skilled chess player, and was lamenting that he couldn't find people to play with except for online, because everyone he personally knew was either too good or too bad to be any fun for him.
Yes…and of course, chess has as big a fanbase as you could hope for.
One question that I had after your talk—and I realize this might just be semantics—was about whether luck and randomness are really the same thing. In the sense that there are games with randomness which a more skilled player will almost always beat a less-skilled player at, and so don’t really compress the skill range. For instance, Chess960 [aka Fischer Random Chess, in which the back rows are randomized at the beginning of each game] tests a different set of skills than traditional chess does, but as long as the back rows are symmetrical...it’s a very different thing than if they had been randomized independently, which would give compress the skill range quite a bit.
That's right, though I’d argue that Chess960 does introduce luck in a more subtle way, in that players will come prepared in different ways and have had a different range of experiences. So when an arrangement’s dealt out that they've seen before—maybe not the exact setup, but patterns will emerge—then they've gotten lucky.
Another example of this is in Bridge: before the game, you prepare a bidding system, and that bidding system will respond better or worse to different hands. And if you're lucky enough to be dealt a hand that works very well within that bidding system, you're going to do much better. Duplicate Bridge can be seen as analogous to Chess960—both teams have been dealt identical hands, but they've prepared in different ways.
That ties into what I see as two different types of skill-compressing luck, although they blur…there's intrinsic luck within a game, and then there's the fact that the more less-than-perfectly-correlated skills you're testing someone on, the tighter a skill range you'll have. My favorite—almost cartoonish—example of this is Chessboxing.
[pause] Sorry, say that again?
Uh, Chessboxing: you alternate playing four-minute rounds of speed chess and three-minute rounds of boxing, until someone's either checkmated or knocked out. Apparently, it's somewhat popular in Europe.
Huh…I'm not familiar with that. It seems I would have heard of it. Although I can't imagine I have, because that’s quite memorable.
The thing I'm fascinated by is that they have a minimum ELO score of 1800 for the chess side, so that, like, Mike Tyson doesn't just come in and wreak havoc…but the reigning Chessboxing champ tends to be somewhere in the 1900-2000 range. So it's a very narrow window—which makes sense, since those are two skills so negatively correlated that you'd expect the luck of the genetic draw to compress the skill range. That seems like a much broader version of what you're talking about in terms of different styles, and how chess players actually have a lot of little skills that they tie together in different ways…
I think "style" is one way to put it, but I'd say different preparations. So we could have in some sense the same chess style, but if I've played a friend a lot who likes a particular opening a lot, then I'm going to be very familiar with that opening. And if in a tournament, I'm matched up with someone who plays that opening too, then I'm favored through dumb luck—it could literally have been a coin toss deciding whether I play them or someone else.
Have you as a game designer thought about how you introduce that kind of luck into a game, without necessarily introducing explicit randomness per se?
I think that while I haven't consciously said "I want that kind of luck in the game," I have spent a lot of time thinking about how to change the framework of the game from the outset, or within the game in some consistent way. An excellent example of that is Dominion…there's many things to like about Dominion, but I've always put at the head of my list the choosing of the 10 Kingdom cards which define the game. I think that gives you such a wide variety of games, and such character to each one. Another game I like very much is Ascension, which has variety in a more traditional sense of not getting the same hand in the same order, but that’s much muddier than in Dominion. There's many reasons I think Ascension is very good outside of that, but the one thing I think Dominion really dominates it in is the game state from the outset. And that's very similar to Chess960, and games where in the beginning, there's some setup that may favor a particular style or preparation.
In some ways, you see that in Magic: when new sets come out, different deck types will become stronger and weaker…and in some circumstances, an entirely new deck principle will be created. At the beginning of the season—say the attack decks are very strong, so players who have an affinity for those are favored. And everyone else needs to either catch up or figure some way out to make their skill set work in the new environment.
I know that you're retired to some extent, but is that something designers of Magic sets are consciously thinking about: there's a certain personality type or player type that we've been neglecting, and here's an opportunity to let, say, deck depleters have a shot at competing this time around…
Absolutely…I'm not involved in that element of Magic very often these days, but that's certainly been part of the thinking: we look at the environment and say, "Well, people who like this type of deck haven't been doing well in tournaments, so let's try to make some cards which really support that sort of play." And they make sure that the different things players are looking for are available to some extent in sets. And very frequently you'll see that the environment is becoming dominated by a particular deck or a particular style of play…say, Control, which makes the top-level games become very slow. And there'll be the idea that next year, there will be fewer Counterspells, and less effective control cards in order to nudge things in the right direction.
One thing I talk to my students about is how Magic’s rise was more or less contemporary with that of the Internet…it was certainly the thing that prodded me to start checking out USENET at college, looking up whether you could Terror a Wall of Swords etc. And there was an interesting effect that I’ve always been ambivalent about, in that the group I was playing with had come up with some very strange and probably wrong strategic ideas. For instance, we all thought that Glasses of Urza was very valuable, since it's a card game and obviously seeing your opponent's hand in a card game is the most valuable thing…and other Galapagos Islands-ish mutations like that.
Yeah, one of the delights of the pre-Net days were these different local variations…which you still see among lower-level players now, but back then it was all levels of players.
And as someone who I know is…this is an odd way to put it, but you're a fan, as I am, of low-level play. That's one of your points about luck, and in your Game Glimpses you interview your kids and take their opinions seriously. What's your take…what sort of skill level do you tend to design your games for now, and how do you think about that?
I spend a lot of time thinking about the low-level play: people who are reasonably intelligent, as those tend to be the people who play these games, but I don't assume that they have a strong background in game designing, and what I want is…paramount for me is that their play experience is fun. And there's a lot of ways in which that can fail. So I want to make it so that there's an avenue for excellence within the game, which rewards the player for investing a lot of time, but not so much that they can't play with casual players. And that's hard to get right…usually you err to one side or the other, but that's what I shoot for.
There a lot of things to look for to make sure that the casual player has a good early experience. You want to make it so that there's easy strategies for them to pursue, which are reasonably effective. And another is to make sure that the things that they want to do are not strategically terrible. For example, in a game in which one of the things you can do is purchase cards, like in King of Tokyo, a lot of players really like to purchase cards, because they're flavorful and fun. And it would be terrible if that's just a bad strategy. And there's a lot of games where that's the case, where you have something that's really tempting, but you really shouldn't be doing it.
So that's a subtler point about theme matching mechanics in a way…that the theme, the implicit theme that is what's attractive should match up to what you want to do…
I think in my conscious mind was that I have this game here that's basically a King of the Hill game, and what can players fight over that's exciting? And after brainstorming a bunch of ideas, one of which…the nearest serious contender was King of New York, where you were gang members who were fighting over turf in New York. I wanted something where fighting was central, because I wanted to make a game where it felt like interaction was okay. And there's three basic strategies in King of Tokyo: go for victory points, go for attacks, and go for cards. And I wanted to make sure that whichever of these three basic strategies people were drawn towards, that they had a place…that sometimes games were won by that strategy.
So when people play King of Tokyo, they will at first maybe just go straight for 20 victory points, and they've actually got a chance of winning. And then a master player—and with a game as light-hearted as King of Tokyo, having that mastery level isn't necessary, but it is there—might begin with going for victory points, but will move flexibly between strategies depending on what cards come up and what the competition's doing…they'll pay more attention to what the environment is favoring at that moment.
One design consideration, then, is that players need the flexibility to either decide on a strategy from the get-go or figure things out as they go.
Maybe that's a good way…I hadn't thought about that, but I've thought about it in terms of "I want players to be able to pursue simple strategies, and I want them to be able to do things that are fun." But perhaps another way to think about the issue of beginners, there's a lot of beginners who want to pick a strategy and not have to think about what their opponent is doing, or what the environment is. And sort of play in an auto-pilot kind of way. And I think a lot of games support that—I think you can do perfectly fine in King of Tokyo, or a lot of other games, that way, but it's not going to be the best way.
Right...one of the things that hooks new players into Dominion so well but is often a complaint about the base set is its inclination towards letting you figure out a strategy at the outset and then play more or less on autopilot.
And in fact, getting back to Ascension, that is one of things which I think is absolutely brilliant about it—and where I think Ascension totally blows the other deckbuilders out of the water in terms of interactivity—is that when you first play Ascension, you can absolutely play it in the way I'm describing: you can pick a strategy, pursue it, and ignore what the other players are doing. But the more you play it, the more you want to pay attention to what they're doing, react to it, make counterpurchases, and so forth…there's a lot of interaction. Ascension hits what I'm talking about very well…if you have a game in which a player can do okay by focusing entirely on their own strategy and not paying too much attention to the opponent, I think that's a great place to start. And if then, in the long run, it matters a great deal what their opponents are doing, that's a great place to go.
So you have a game like Yahtzee, which doesn't quite have that back end. You can do very well in Yahtzee by not paying attention to your opponents—you can get a few percentage points by pressing your luck or playing it safe depending on where you are. Golf…you can play a good game of golf without paying attention to your opponent's score at all.
It's interesting…the first real game I made was when I was 19. I had started playing Magic a month before, and what struck me was that none of the traditional card games I liked—since I was the sort of kid who would actually go through books of card game rules to find weird ones—but none of them were nearly as interactive—as just "Beat the hell out of the other guy"—as this. And at that point, that was just the greatest thing in the world to me…(laughs) as it still sometimes is.
My game was…basically, you make a column of 21, and you get those cards, but you also get one of your opponent's columns. And that was basically it—there were different combos that flipped that positive feedback loop, but the basic idea was that you don't just succeed, but hurt the other guy while doing it. And it entirely just came from that realization about there being so many different kinds of interaction, and for whatever reason, traditional card games…trick-taking games are quasi-aggressive, but then you have games like Cribbage or Gin Rummy with much more indirect interaction...
Gin Rummy really falls into the Ascension model, where you can play it focused on your own side and do okay, but if you really want to get good, you get a lot of payback for paying attention. In general, though, the more indirect styles of play are excellent for multi-player games, and the more aggressive styles of play are excellent for two-player games. One of the problems with aggressive multi-player games is that they tend to pick on the leader too much, so then being in the lead doesn't really mean anything, and it becomes a political game. But that isn't a problem with two-player games…with two-player games, the main problem is that you might get a feedback loop that doesn't end fast enough…I might be ahead and you feel like you have no chance of winning…and in fact, you don't, but it’s still going to go on for the next half-hour. Which is a bummer.
My last question is about the intersection of theme and mechanics. I've always wondered...do you think Magic could have been as successful as it was if it didn't have a fantasy, dragons-and-wizards theme? I mean, I've thought hard about this, and I'm fond of games with innovative or esoteric themes, but I simply can't think of anything that would be as instantly accessible and let people know what this aggressiveness is all about, yet still have the variety of thematically intuitive effects that you need.
I think it’s hard to…I think other things could have worked, but I don't think they would have been as good. You need that variety, but you also need this commonly shared, widely understood framework. In a realistic setting you don't have that variety, and in science fiction you don't have that shared framework—there's so many understandings of what a science fiction universe is that if I show you a particular creature, you'll have no intuition as to what it is or what it might do. But if I tell you there's an ogre, you kinda know what they are. Regardless of your upbringing or what you've been exposed to, you know it's gonna be kind of tough, and not too smart.
A lot of things that players brought into fantasy really helped. You can imagine Magic being done more abstractly, like Go or Uno, and could Magic succeed as something like that? I don’t know, but certainly less so. Theme’s important, even when it’s subtle. Playing chess with a Simpsons chess set, or Gin with an artist-driven deck of cards…I find those things really hard to play with, and one of the reasons is that the simple connection of these classic shapes in my head—king, knight, jack, etc.—are ingrained in how I understand a game, and varying them makes it harder to see what's going on. And then you think—well, there's only six different pieces in chess, and it helps there…so you can imagine how important it is to have some sort of connection on a visceral level for thousands of different cards in Magic.
Thanks for the interview, Richard!
Cool interview though I'm a bit confused why it would belong here in RPGG? (But a great interview!)
open the heart
So it can be read by his fans of course
So glad you posted this!
A. B. West
Why aren't you PLAYING a game?
I wish I could hear more from other designers, but Richard Garfield is the guy I think of as the smartest designer of all. I mean in terms of really thinking deeply about design and the implications of choices in design.
open the heart
Yeah, he's a good one
It would be nice to get him to design a MTG cube!