Thursday, September 13, 2007

Card Games & Randomness

Recently my Thursday night gaming group has had the opportunity to play some top quality card games, among them Tichu and the German Doppelkopf. Comparing and talking about these games led to interesting discussions of the the element of randomness in card games, which I offer up here for additional thoughts and comments.

The Heart of Randomness

Our conversation got started with one of players saying that Doppelkopf was too random.

The thing is, however, that almost any card game is random. It's a necessary and implicit part of the process. You take 52 cards (or whatever) and divide them up among the players. For a standard 4-player game that represents 635,013,559,600 different possibilities. In other words, sometimes you're going to get a really bad hand and sometimes you're going to get a really good one.

That's life. Or at least a card game.

I know of a couple of games that try and modify this. The classic is Duplicate Bridge where multiple people all play the same hands of cards, and then how well they did is compared. Personally, I find this entirely at odds with how cards work: they're a purposefully arbitrary mechanism meant to divide resources in a random method. To make sure everyone has the same hands is not only tedious, but also suggests you're not using the right component.

Trump, Tricks, Game is a rare game that I saw that had a method to redistribute cards that wasn't random but that I actually respected. After the first round of play you play additional rounds with the cards that you won previously, thus creating an additional level of decisions in the game, where you must not just collect valuable cards, but also cards that will let you do well in future round.

Offsetting Randomness

Given that randomness is a core element of card games, the question then becomes how do you offset it to make the elements of skill more important than the randomness?

The best answer is time. A card game will best offset randomness through continued play of hand after hand. My general assessment is that games that allow for 10-15 hands of play are getting to point where the standard deviation is creeping down, and players have generally had the same amount of luck over a game.

(Though always there will be some games were someone's luck was just unbeatable.)

Regrettably, this is where a lot of Eurogames fall down. They've gotten so into the mindset of short, simple games that you have card games where you might only play one round, or at best a number of rounds equal to the number of players, and this just doesn't do it. If those games feel random, it's probably because they still are.

(And it should further be noted that a lot of Euro card games are light enough that they can't support more than that amount of play, meaning that the high randomness is unfortunately an implicit part of the game that can't be reasonably corrected.)

However the best classic card games have at least three other ways to offset randomness, and these are generally methods that Euro card games could learn from.

The first method is value assessment. This means looking at the hand of cards you were dealt and then determining how good it is. Most games like Bridge and Spades do this through bidding. Poker's betting accomplishes about the same purpose--at least absent bluffing, reading opponents, and other elements that confuse the core value assessment.

The second method is hand refinement. Games like Hearts and Tichu do this by passing and receiving cards from opponents. It's more meaningful in a game like Tichu where you're trying to create various card patterns then in a game like Hearts where you're mainly focusing on Hearts, Spades, and voids. The commercial game Havoc: The Hundred Year's War does it via card-drafting, as does the classic Coloretto.

The third method is card play. Or rather, good card play. If the way to play cards is obvious, you're not going to offset any randomness, but if a good card player can make much more clever use of his cards--even if it's just by quickly completing voids or learning to see how your cards can simultaneously match a few different patterns--then his skill will start to shine through, especially as more rounds of play occur.

Conclusion

Yes, card distribution is random, but good card games overcome that. You have to play more hands, and the game has to be deep enough to support it. However even beyond that I think you'll find that the very best card games offset randomness even more with card assessment, hand refinement, or some combination thereof.

Light Euro card games could learn well from these, and consider how these methods could be built into their own gameplay.

6 comments:

Siow Hwee said...

Pico is a game that you have to swap hands and it has perfect information. It balances the luck.

http://www.boardgamegeek.com/game/606

dan daly said...

Introducing different selections of cards to the deck at different points in the game is another way to make card games less random. Domaine and Twilight Struggle would be 2 examples of this.

huzonfirst said...

Shannon, like many discussions which involve randomness, I think this comes down to the difference between situational luck and resolution luck. In the former, the randomizing action comes first and then the player can react to it. This is the way most card games are and why I gladly embrace the luck inherent in them. Yes, luck determines the cards you get, but the fun comes in figuring out how to play them. There is still some unfairness, but good players will do a much better job of playing a bad hand than beginners will.

This is as opposed to resolution luck, in which the players make their moves and THEN the chance element determines success or failure. This is the way Monopoly, Risk, and most wargames work. The luck in these games bothers me much more, because I can't react to it. So just because two games have luck doesn't mean I'll judge them the same way. The luck in most card games is far preferable to me than in other types of designs.

Shannon Appelcline said...

That's a good way to differentiate them.

dan daly said...

I'd say the situational vs. resolutional luck distinction as described above is pretty hazy.

So if you're playing risk and roll poorly in a battle, you say you can't react to those poor rolls?

You can react by what you do with the remainder of your turn (attack the same place from a different territory, negotiate a peace treaty with someone, change your tactics and go for a different continent, use your free move at the end of the turn in a different manner, keep attacking somewhere else to get a card you need) In short, there could me several ways to respond to poor dice rolls in Risk, and this is only talking about the rest of the turn you made the rolls. You can also respond to them throughout the rest of the game.

Another aspect of the situation that I think is being over looked, is that so called "resolutional luck" allows the player to respond to it BEFORE HAND by stacking the odds in their favor. In a hex and counter wargame, make an attack at 3-1 odds instead of 2-1, in Risk attack with a large army that allows you to roll 3 dice. In this way "resolutional" luck can in fact be mitigated MORE than "situational" luck in many cases. Most games give a player no or very little control over which cards they draw/flip/etc, but players have a decisive input as far as pushing counters or plastic armies around goes.

Frank Branham said...

I'm pretty sure you need 8-10 hands at a minimum to produce reasonable averages. That does depend on the game. Twilight has such completely erratic shifts in scoring that it may be higher--even though the cards are naturally associated with a partnership.

I was very aware of the small number of hands in Dia die Los Muertos. The Gift Exchange card swapping thing is an only partially successful attempt to balance hands based on points scored during the hand.

I've been trying to work out a better way to deal with that problem, but card swapping in the middle of a hand is completely odious to many trick taking players. There still might be a better solution there; I ain't found it.