Game design analysis of Hanabi and COUP

The following analysis was done originally as an assignment for the Introduction to Game Research course at University of Gothenburg, first section fall 2019.

Game design analysis of Hanabi and COUP

This paper seeks to find the design structures that Hanabi(I will be refering as Hanabi the delux version) and COUP have individually regarding four factors: players engagement, gameplay life time, players social interaction and players feel of achieving while playing. Both Hanabi and COUP are games that I played in the game play sessions of the Introduction of Game Research(DAT385) class, and took notes while playing with my classmates, so most of the descriptions of the games were taken from those experiences.

It is important to mention that in order to make descriptions of the games I am using the definitions for game mechanics (Sicart, 2008), game design patters (Gameplay Design Patterns, n.d.) and aesthetics from the MDA model (Hunicke, LeBlanc, Zubek, 2004). In this analysis I will be referring to mechanics as methods executed by players in order to interact with the game world (Sicart, 2008) which is a definition that makes easier to point mechanics as verbs. Moreover, in order to distinguish patters and mechanics in the paper I decided to use italic font for mechanics , and bold font for patterns.

Games description

In order to analyze and describe each game structures I think it is needed to describe them with their own mechanics , patterns and frameworks, based on the play session that I had in class. Therefore I present the following descriptions for Hanabi and COUP:

Hanabi.

Hanabi is a board game that uses tiles with different forms and numbers, its players have four tiles pointing backwards each one, which makes their value unknown by the owner of the tiles but known for the other players. Moreover, the game gives nine clues and three lives for all players to share with each other. Also, Hanabi makes its participants to divide its actions into turns, so each turn players can execute one of three core mechanics: give clues, use a tile and discard a tile (I based my mechanic interpretation based on Bard descriptions on the paper The Habani Challenge: A new frontier for AI Research) .The give clues mechanics means that a player can give clues to another player so the he can have more information about its own tiles, but one of the available clues becomes unavailable. The use a tile mechanic lets the player to complete an ordered secession of tiles with similar symbols. And the discard a tile mechanic let players to make one unavailable clue to be available again. Moreover, Hanabi ends when the players are out of life(no live tokens left) or there are no tines left to play, then points are counted by the number of tiles that create a sequence with the same symbols.

By the description given in the last parragraph I can conclude that Hanabi is an Imperfect information game, because each player knows limited information of its own tiles, and also by ex formation because each time a player gives a clue it introduces noise to the system. 

Table 1: Hanabi Mechanics

Mechanics	Description
give clues	- Player A gives a clue to Player B, so Player B can have more information about its tiles. This disables one clue token.
use a tile	- Player A uses a tile to complete an ascendant sequence for the same  symbol as the tile to be used.
discard a tile	- Player A discards a tile to enable a previously disabled clue.

COUP.

COUP gameplay works by a minimum of two players, a set of coins and a deck of cards. For a game session each player receives two cards, which are called influences, but they have to keep their card values in secret for the other players. Furthermore, each card has its own actions, effects and counteraction, however the players do not know the values of their opponents and vice versa which means everyone can do the actions that they want, as a consequence the players can object that the others are bluffing , by doing this the rules of the game says that if the player that objects found that another player is effectively bluffing then the bluffer looses a card, but in the opposite case the objector looses a card .With the last beating said I can conclude that this game has a bluffing, a conceal and an ambiguous response pattern since its two core mechanics are bluffing and object, which are the action and the consequence, respectively, of deceive other players about their roles. Therefore, as Hanabi, COUP is an imperfect information game, since during all gameplay most information of its state is hidden from the players.

Table 2:  COUP Mechanics

Mechanic	Description
bluffing	- Player A wants to assassinate one of B’s card, claiming he has a card with assassin value. So Player B bluffs about having a card with Contessa value which can blocks assassination. However, player B can actually have Ambassador which can not block assassination
object	- Player A action is blocked by Player B, since B is saying that it has a card that can block A’s action. So Player A can object about the veracity of B’s statement.

It is worth to mention that the following analysis combines more patterns that appear as a consequence of playing both games, therefore the patters mentioned before are intended to give a general description of the games and the following patterns aid to have a deeper understanding of the engagement , interaction and gameplay time a player can experience.

Design Structures to keep the players engage with the game

Both, Hanabi and COUP have structures to keep players engaged. Hanabi works by using 4 tiles placed underside each player so each player can guess its tiles by clues given by the other players around, which leads to ambiguous responses (Gameplay Design Patterns, n.d.) from the owner of the tiles. Also Hanabi uses elements of surprise (Björk, Westborg , Sivertsson, 2019) and luck (Gameplay Design Patterns, n.d.), because each time a player decides to throw a tile that he does not really know its value. For instance a player can be thinking that he may has one red one tile on its tiles since other players gave him clues pointing out one red tile and three four yellows, however he will have a surprise if the known red tile results to be a four. In the case of COUP, it uses bluffing as its core mechanic which makes the game interesting because you as a player wont be sure about the information that the others players cards have. Also, as a player makes any action , it can create ambiguous responses to the other players, since they can disagree that other participants are playing as the role of the cards values. One way COUP has to resolve the confusion that a player can have is by confronting(object) the role that a particular player bluffs which reduces the frustration of ignoring the oponent’s cards.

Design structures to make the games end near the stated time

Both games have structures to finish playgame near the stated time which can be described with mechanics and the MDS framework. In the case of Hanabi the way to ensure this is done by using two mechanics: give clues and use tile, which I decided to divide in 4 secondary mechanics (Sicart, 2008) for description sake: give clues, discard tiles, use tiles and take clues, which also evoke dynamics and aesthetics because when a player A gives a clue to a player B it triggers a dynamic which is that the player B would think about the previous information that player A and C already gave to its self in order to use a tile or take a clue. The result of the previous dynamic will create a feeling(aesthetic) of uncertainty for all players involved, since everyone wants to keep the hanabi live tokens and make as much points as they could. And since all players want to remain “alive” the game forces them to take more time taking clues and discarding tiles to open more clues, making it a long play time. Overall with the last example Hanabi has a system intended to make people to play as long as they could.

Likewise, COUP uses a structure based on a beating opponents mechanic which makes the gameplay time to be short(by my experience of the game of playing maybe 15 minutes). COUP has two secondary mechanics that help the players to beat opponents, which are assassinate and coup, both cause the opponent to loose one card. These two mechanics counteract the fact that the game uses bluffing as its core design , making it harder to players to keep cards and stay playing longer.

Design Structures that make the players interact with each other

Hanabi and COUP share the same design structure in order to make social interaction between players, which is the limited foresight (Björk, Westborg, Sivertsson, 2019), because both have mechanics that make the players to reveal limited information on each turn which as a consequence motivate participants to generate social interaction in order to clarify the game state.In the case of COUP, it has an additional structure, to complement the limited foresight, which is the enforcement player anonymity (Gameplay Design Patterns, n.d.) pattern since all players can bluff about their roles(cards) which consequentialy creates an uncertainty feeling(aesthetic). The uncertainty is resolved with the capacity of object the information that opponents expose. The object mechanic enforce players to communicate with each others verbally and physically, for instance a player A can bluff about him to be a Contessa, which can blocks assassination, and player B can object that A is not a Contessa that he is a liar and he is just bluffing. Likewise, Hanabi limits the information showed to the player since the tiles ,that each player has, are hidden at the opposite of the players vision, which led players to make actions based on clues that the other players gave then before. So, the Clues pattern of Hanabi encourage players to socialize while giving hints to another, which creates a feeling of cooperation and a fun bustle atmosphere even though comments from another players can be consider cheating but the game is so engaging with its mechanics that makes it even funnier to play while commenting.   

Design structures that make the players to feel that they are achieving something

For this part of the analysis I decided to see both games as state machines with goal states as aesthetics from the MDA model, since the goal can actually induce players to feel the sense of achieving. Games as state machines have three states: The initial state, the end state and the goal state (Björk, Westborg, Sivertsson, 2019), and for these analysis I decided to divide each game turn in the tree previous mentioned states. Each state cause different feelings to the players. These feelings that I am going to describe are three, belonging to the taxonomy described by Hunique: Sensation, Challenge and Discovery, respectively for each state. For Hanabi, in the initial state the player has to select a tile from the four that he has informed for a sequence of symbols to be completed, which provokes the feeling of Challenge. But for the final state, which the game causes to reveal the tile that the player has previously selected, the game evokes the feeling of Discovery as it is information that is being discovered or revealed and was previously unknown. On the other hand, for the goal state the player can have Sensation because he could feel pleasure in having the right tile to complete the series or the opposite because he might lose points. This last Sensation can also be described through the Goal Achievements design pattern which makes the players to get the fealing that they are completing the sequence in Hanabi.

In the case of COUP, I identify the initial state as the fact that the player takes an action(action as a concept that the game has, not as mechanic since I expose them before, but as features working with the mechanics) for convenience either to have more points or to eliminate a player(assassin role), which evokes Challenge so that others do not object. The final state, on the other hand, the action taken which leads to the possibility that another player objects to the previous action, which in turn evokes feelings of Discovery for the other players and at the same time a feeling of Challenge because if the players who object are wrong, they would lose a card. Consequently, the goal state is the fact that the player gets away when he used bluffing creates a Sensation with the pleasure of having bluffed to outwit others.

Discussion: ussing patters to describe worse case escenario(many players as possible in a certain time)

This section tries to understand the games described before in the worst possible scenario, that is, to understand the state of the player when the games have lengthened more than necessary and with many players, in which the players would no longer experience pleasure. Also take the opportunity to try to find the limitations of the structures described above.

For Hanabi, I will define that the worst case would be the one in which there are five players and the game is reaching sixty minutes of playtime. I have chosen this scenario because, according to the experience that I had with the game, the players begin to have more anxiety to finish the game after twenty minutes, so one hour should evoke feelings of discomfort among the players. I have also decided that the stage has five players so that there is social tension between them. Therefore my observations of the gameplay pointed me out that the moment that a player takes too long to make his movements an excessive paralysis analysis can be evoked, which triggers feelings of boredom and frustration to his teammates, which are factors that the MDA model does not fully take into account as it rather defines one Aesthetic as a way to feel fun. It is true that these emotions are not pleasant, but they typical appear in a game session, so it may be that the MDA model is not totally adequate to define a complete gaming experience.

Likewise, for COUP the worst possible scenario I defined it as six players playing it for an extended time of 15 minutes, and as a thesis I assessed the possibility that the players no longer felt pleasure in playing it. With this the sense of Challenge(From Hunicke taxonomy of Fun) would disappear from the game, because, by observation, I had concluded that the COUP playing time is inversely proportional to the number of players, and if it is so fast, the sense of playing it could become boring since it would be so fast that it would not exist a desire to win. However, after extensive games I have seen that both my teammates and I, really wanted to continue playing, which made me think the question of "why we were having fun in such an extreme scenario?". The answer to this question could be found in the description of social interaction (in the previous analysis above), since being COUP based on bluffing patterns and enforcement player anonymity makes players to not be tired or bored of the game because curiosity emerges in the players, which I have not found described in the patterns I have read but it would be an interesting way to describe games.

Conclusion

The board games analyzed in this paper have mechanics and design patterns that help them to keep players engaged, keep them interacting with each other, to have constant feeling of achieving goals, and also to keep the gameplay session to end in a certain time. The mechanics were described as verbs players can execute while playing , following patterns that the games have. These patters helped the paper to describe players engagement, interaction and gameplay time. However, the feeling to achieve something could be described mixing the idea of a game as a state machine in which the goal state is showing an aesthetic from the MDA model. Moreover, the design pattern structures and the MDA model are effective in describing the responses of the aforementioned analyzes, however, it is possible that not every structure helps designers to analyze games in a deeper way, as this is the case of analyzing them in the worst case scenario.

References

Björk, Lundgren, Holopainen. (2003). Game Design Patterns. Proceedings of the 2003 DiGRA International Conference: Level Up, 2. Retrieved from DiGRA:  http://www.digra.org/digital-library/publications/game-design-patterns/

Björk, Westborg , Sivertsson. (2 de August de 2019). Introduction to Game Research. Lecture 1.1. Gothenburg.

Björk, Westborg, Sivertsson. (2 de Sep de 2019). Introduction to Game Research. Lecute 1.2. Gothenburg.

Gameplay Design Patterns. (s.f.). Obtenido de Category:Patterns: http://virt10.itu.chalmers.se/index.php?title=Category:Patterns&pageuntil=First-Person+Views#mw-pages

Hunicke, LeBlanc, Zubek. (2004). MDA: A Formal Approach to Game Design and Game Research. Retrieved from CiteSeerX: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.79.4561

Nolan Bard, Jakob N. Foerster, Sarath Chandar, Neil Burch, Marc Lanctot, H. Francis Song, Emilio Parisotto, Vincent Dumoulin, Subhodeep Moitra, Edward Hughes, Iain Dunning, Shibl Mourad, Hugo Larochelle, Marc G. Bellemare, Michael Bowling. (2019). The Hanabi Challenge: A New Frontier for AI Research. Cornel University, Machine Learning.

Sicart. (2008, December). Defining Game Mechanics. Retrieved from Game Studies: http://gamestudies.org/0802/articles/sicart