[My thanks to “Sagrilarus” of Fortress:AT for the question that stimulated this attempt at classification.]
Phases (sometimes called stages) in a game design are important. These are distinctly different periods of play through the course of a game. They provide at least a perception, if not an actuality, of change, growth, and learning. Phases help the feeling that there's more variety in the game, as well. They help avoid a perception of "sameness" in the gameplay. A game that is "too long" may feel too long because there are not enough phases, not because any specific amount of time has passed. In contrast, many short games have only one phase.
Other entertainments and activities in life have phases. A horse race has phases, movies have the three (or five, or nine) act structure that changes the focus as the movie progresses. Life itself has phases such as early childhood, adolescence, and retirement/old age. In that sense people expect phases in their entertainment and their activities.
What differentiates one phase from another? I don’t think we can closely define that. Much of it must occur in the minds of the player(s). When a game changes from one phase to another the player is thinking about different things, as he or she decides what to do, than he thought about in the preceding phase. Probably the best way to put it is, the phase changes when the immediate (short-term) objective(s) of the players change. I’ll give some examples in a moment.
The longer the game is, the more phases it should have. After all, if a major purpose of phases is to avoid sameness, then the need becomes greater as the game becomes longer.
Some single-episode games that are easily played “best two out of three” have one phase, for example rock-paper-scissors (RPS). Tic-Tac-Toe is another such game, with a maximum of five moves for the "X" player I don't know how sensible it would be to talk about phases. Other very simple games like Candyland and Chutes and Ladders often have one phase. We might be able to characterize short games as "one phase games", although I think we could find fairly short games of more than one phase.
In contrast, Chess can be quite a long game--players are allowed two hours each for 40 moves--so it stands to reason that it needs to have more than one phase. These phases are normally called the opening, the mid-game, and the end-game. The opening phase is a consequence of the severe constraints on movement of pieces at the start of the game, given the standard set up, and of the centuries of study of the best moves to bring pieces into the open and control the center of the board.
Contrast this with Risk, where the opening phase is the placement of armies before the conflict begins, and that placement can vary greatly from one game to another. Even if you use the French setup where the cards are dealt and territories are occupied randomly, you have a setup that varies greatly from one game to another.
Contrast that with many wargames where there is a standard setup, but a player can move every one of his pieces in one turn. As a result the game moves beyond the standard setup very rapidly, as opposed to chess when moving one piece at a time means the opening phase takes 10-20 moves by each player.
And contrast those with games where you have no units, or no maneuver (where geographical location of assets does not matter). Often these games are symmetrical rather than the asymmetricality common in wargames. There can still be an opening phase, but it is not related to maneuver of units.
A second reason for the existence of phases in chess, other than the very constrained initial position of pieces, is that the number of pieces gradually decreases while the area of action remains the same size, thus opening up longer lines of play and new possibilities. There’s a third reason, the piece mix for each player may deviate from the symmetric, from being identical, for example after an exchange of a Knight for a Bishop. Forces can also become imbalanced when one player gains a material advantage, e.g. being a pawn ahead. The mid-game in chess is also a change because players are no longer following the standard openings, but have an immediate objective of gaining positional or (more likely?) material advantage.
The end-game occurs as the number of pieces is much reduced. There is more room to maneuver. Further, the immediate objective becomes checkmate of the opponent’s king, rather than material or positional advantage. Players now try to use a material or positional advantage, if they’ve gained one, to end the game.
Every chess game has an opening and a mid-game, though the latter can be cut short by a quick win. Except when a player stumbles onto a checkmate while still trying to gain positional or material advantage, there will always be an end-game, that is, a phase when players are focusing on checkmate.
What about other games? Play changes in a simple puzzle-game like old PCTetris because the pieces fall faster. At some point there is no further increase in falling speed, and a good player can settle into a cathartic repetition until he or she tires and makes mistakes. We can say there’s the ramping-up phase and then the “maximum fall” phase, a phase only experienced players reach.
Play in RPGs and FPSs changes as player avatars acquire more levels, perks, and loot (especially more and better weapons). The monsters are tougher, the bosses are tougher, the player(s) have many more options. In effect, the rules are modified by the loot, by perks, and by new capabilities gained by leveling up in RPGs. There may also be changes in immediate objective as the story associated with the game develops.
Is the setup a phase? Yes, if players make decisions that affect the outcome, as in American (not French setup) Risk. No, if they don’t, as in chess or checkers.
Many games have no setup phase. Every player begins symmetrically (all players with identical situations and assets), and if he has assets that can be maneuvered, they have not yet been maneuvered into significant positions. Card games are almost always of this type. Chess and most traditional boardgames are also. Turn-based and real-time-strategy video games are symmetrical insofar as each player begins with one unit "somewhere", though the sides are not symmetrical owing to unit differentiation. Most video games are asymmetrical but have a mandated setup.
Historical wargames that might be called "simulations", on the other hand, are almost always asymmetrical (differing situations and assets) in the setup, but sometimes allow players to choose their setup. Games that simulate historical battles are always asymmetrical, but sometimes the setup is mandated by the game, while other times the players can set up pieces as they like. More abstract (non-simulation) wargames are often the opposite. For example Stratego is symmetrical but players can set up their pieces as they like, so the setup becomes the first decision phase of the game. Risk is the same. On the other hand, Diplomacy is asymmetrical but the initial setup is mandated by the game.
Video games involving an avatar are severely asymmetrical, with one character facing numerous opponents. Add the avatar customization opportunities that are so popular in these games and you have thousands if not millions of possible setups.
Phases and rule changes
Phases ideally should not include changes in the rules but may include cases where rules that did not matter earlier in the game come to matter later, or where rules are added through acquisition of loot, or cards, or perks, or levels. For example, there may be a rule that limits the number of pieces a player can have, perhaps reflecting supply or maintenance restrictions. This rule may not matter at the beginning of the game but will as players build up their forces.
Ideally the same rules should apply throughout the game, with changes in circumstances leading to changes in phase. Yet sometimes the story or history of the game demands changes in rules. In my game Britannia, which represents 1000 years of British history, the rules are generally the same throughout, but the identity of the offensive nations and defensive nations changes over time owing to invasions and withdrawals. However, the rules are quite different for the Romans at the beginning of the game, and slightly different for the clash of Kings at the end of the game. We have the phase of Roman conquest where submission rules enable British nations to survive the conquest despite the unique power of Roman roads, forts, and legionnaires. This is followed after Roman withdrawal by the phase of Anglo-Saxon invasion and domination, followed by the phase of Viking raids and conquest (the Anglo-Saxons become defenders rather than attackers), followed by the clash of Kings where we have additional reinforcements and cavalry, four phases for a 4 to 5 hour game.
In traditional Risk the phasing is provided by the increase in the number of armies received for turn-in of territory card sets. If you ever play Risk with a low repeating number of armies for card sets, such as 4-6-8-4-6-8, you'll find that it stays in one phase for a very long time. There is less randomness this way, but there is little momentum toward completion. The ever-increasing number of armies received for card sets in the standard (pre-2008) rules provides the momentum to complete the game, although it can still take quite a while. In the 2008 redesign of Risk using mission cards, completion of missions provides the momentum toward completion. I don’t know whether the new style game has many phases or not.
Even a game as poorly-designed as Monopoly has phases. The initial phase is the slow acquisition of properties (slow even when the correct rules, auction when a player chooses not to buy at list price, are used). When players begin to get monopolies they move into the next phase, building houses and ultimately hotels. The last phase is a lot of dice rolling to see who lands on whose built-up properties without being able to pay the piper.
The bottom-of-the-game-design-barrel social network games on Facebook can have phases, in fact phases are important to avoid the extremes of tedious repetition. As players progress in Farmville they can expand their farm, automate it, change their principle crops (or animals, or orchards) as new ones are “unlocked”, and so forth. This provides a feeling of movement and progress in what is essentially a mass-market “game”, working within the rules complexity limits of mass-market games.
Some games don’t have phases, but are episodic. You play several times rather than just once, sometimes with “best two out of three” determining the winner, sometimes with more complex scoring. Video fighting games tend to be of this type, but many traditional 52-card games are the most obvious example.
Typically, these card games do not have phases. You play a hand, the hand is completed, you play another hand, that hand is completed, and so forth, with the game reset to its beginning situation each time, except for the score. In some cases you maintain an accumulating score (or as in poker an amount of chips that varies from player to player). In many cases what happens in previous hands does not affect what happens in later hands. In other cases such as Bridge and poker what has gone before affects each hand, whether through the points and vulnerabilities of Bridge or through the amount of chips/money each player has accumulated (or lost) in poker. Of course, in all of these games players can learn about how others play, and that can affect their own play as time passes.
Flow and learning
Mihaly Csikszentmikalyi’s concept of “the Flow” has been adopted by many (e.g. Raph Koster) as a model for games. (See my explanation in "Why We Play" http://www.gamecareerguide.com/features/625/why_we_.php .). Ideally, a game should become more difficult as players become better at it.
Koster talks about games as learning in a safe environment. Phases mean there’s more to learn in the game. If the phases don’t involve rules changes, all the better, the learning is about how to play well, not about how to deal with new mechanisms of the game. Phases don’t necessarily mean the game becomes harder to play well, but they may still contribute to “the Flow”.
Virtually all games involve repetition, whether it's repetition of turns or something else. The question is whether this repetition can be conducted in varying circumstances which amount to different phases. You can play two rounds with exactly the same rules, yet the results from the first round mean that what goes on in the minds of the players in the second round is rather different. This is most likely to be seen in Eurostyle games with a limited number of rounds in which a lot can happen.
If one round can be, in terms of rules, just like the preceding one, but owing to changes in circumstances it feels different to the players, you’ve effectively increased the variety of the game. And for 21st century gamers, variety is very much “the spice of life.”
Once again, the phase difference is in the mind of the player, and as such it is not something that we can define rigidly. But it usually means that the short-term objective(s) of the players have changed from one phase to the next.
Other reasons for phases
Another reason to have phases in a game design is to mitigate the uncatchable-leader problem. If, after half a game, the player who leads will almost always win, why play the rest of the game? If the game has distinct phases with different gameplay, that can help other players overtake the leader.
Here’s a final, subtle, reason why phases are important. Designers are in some danger of having game fans treat games the way some basketball “fans” treat basketball. These fans only watch the end of a basketball game because they feel that what goes before doesn’t matter to the outcome. They don't recognize that there are phases and variations in basketball that are as interesting as the results. They're only interested in the destination, not in the journey. We see this in video game players who find cheat codes, play only the end of a game, and then say they “beat the game”. Phases help make the journey more interesting, for those willing to experience it.
The point, for game designers, is to find ways to vary their games so that phases, significant changes in what happens in the minds of the player(s), occur. This is likely to make the game more appealing, and more long-lasting. Fortunately, if you're designing a game that lasts more than half an hour or so, it may naturally fall into phases as you work on its other aspects.
In a few days I’ll briefly discuss whether the proverbial "three act structure" that is so often ascribed to films, plays, and novels, is typical in games.