Posted on

A Model for Decision Making in Games, Part One: Action Selection

Two weeks ago I wrote an article defining worker placement in response to some rather loose use of the term. I thought I might get some disagreement on my definition, but instead I got disagreement on the use of the mechanic itself. Some people apparently hate worker placement because they feel that it restricts their choices and has made previously complex games simple. I disagree, because I think the comments reflected a somewhat superficial understanding of how decisions are actually made in games. Though it’s quite possible that some worker-placement games have fewer meaningful choices than some pre-worker-placement games, I don’t believe that it’s endemic in the category of play, and I’m certain that it’s not a requirement. That’s because worker placement only affects one phase of the decision making process — and not the one that leads to the most voluminous set of options. Hence, this article, the first of three. It’s not about worker placement specifically, but rather about the whole spectrum of decision making in games (and how worker placement fits into that).


When you make a decision in a game, it comes in three parts: what you are doing; how you are doing it; and what the results are. The first two parts of that formula represent an ever-branching tree of options, while the last part involves the mechanics of the game system churning out the results.

Figure 1. The three parts of decision making.

You can also look at these three parts of a decision from a more game-centric point of view. Since games tend to be about actions, the decision points translate to action phases. Together, they comprise the frontend of a game, the player-focused portion.

Figure 2. The three phases of game decisions.

Though every game has these three phases, often one or two phases will be degenerated. Perhaps there’s only one action you can do in a game; perhaps there are no additional choices after you choose your action; and perhaps the results of an action automatically occur. However, a designer should still consider all three phases because they represent the full panoply of game-decision choices. (Spoiler: worker placement is only an element of the first phase; and spoiler: it may often have degenerated second and third phases.)

Phase 1: Action Selection

Action selection is perhaps the driest part of the decision-making process. It’s a level removed from the theme and the backend mechanics of the game and simply involves a player choosing which of several type of actions he wants to take. Thus, it’s pretty impressive that several types of games (action-point games, simultaneous-selection games, card-drafting games, deckbuilding games, and of course worker-placement games) are defined primarily by their action-selection method.

Generally, eurogames have focused the most on the action-selection element of game design. Traditional games and Anglo-American games are more likely to incorporate a degenerated action-selection phase that supports only one action.

The following lists the major types of action selection. Most games are defined by one or more of these action-selection methods.

Option 1a: Basic Action Selection

Basic action selection defines the most traditional elements of action selection. These are the mechanics that are likely to be found in any game, no matter what its heritage, and which are the foundation stones for more complex (usually limited) action selection.

Figure 3. Elements of basic action selection.

The most basic differentiation for action selection defines whether players get to actually select actions or not.

Singular Selection. A degenerated type of action selection, where players have no choice of action type: there’s just a singular option. Examples include Monopoly (roll the dice), Chess (move a piece), and Memoir ’44 (play a card). Obviously, that singular choice then branches out in action execution, but players do not decide among different classes of actions before that.

Rules Selection. The traditional, unsophisticated methodology for action selection. The rules list out several major types of actions that players can take, and players have to remember them. Perhaps the game offers a minor aid, such as a player screen, but it’s not well centralized. Tigris & Euphrates offers a typical example. Players can place a civilization tile; place a catastrophe tile; move, withdraw, or place a leader; or refresh their hands. Call it four actions or six if you differentiate all the leader possibilities. Tigris & Euphrates’ options are listed on player screens that players mostly ignore.

Some games not only allow action selection, but also support the selection of multiple actions every turn.

Action Points. Players are given a set number of action points each turn and can divide them up between multiple actions. In a simple action point system such as Pandemic all actions have the same cost. In a more complex action system such as Tikal (and the rest of Kramer & Kiesling’s Masks series) or Tinners’ Trail different actions have different action-point costs.

Action ResourcesPlayers have a theoretically unlimited number of actions, but they are effectively bounded by the resources that they must spend to take those actions. If some actions don’t cost resources, then they must be explicitly limited. The Settlers of Catan is the classic example of Action Resources: players trade or build until they no longer have resources to do anything useful. Terraforming Mars is a more recent example of an Action Resources game: players play cards and build standard projects by spending money (and sometimes other resources), and also enact blue-card actions, which are limited to once per round (generation).

Finally, basic action selection can be made intuitive.

Menu Selection (or sometimes just: Action Selection). This is just rules selection with a better marketing team. The possible actions have been moved to a player aid or to the board itself, to make sure that they’re always visible. Icons are typically used to make the options even more obvious and accessible. The game aids in Settlers of Catan are a fine example of rules selection made graphical, while the original version of Tinners’ Trail places its options right on the board itself.

The ability to take different sorts of actions and beyond that different numbers of actions is exciting and a big expansion over a game that just requires the same action turn after turn. It’s a large-scale change from classic games without any sort of action selection and a foundational building stone of These Games of Ours.

Option 1b: Limited Action Selection

Basic action selection allows a player to choose what they’re going to do and do it. No muss, no fuss. But over the last few decades clever designers have come up with various ways to ensure that players can’t always do what they wanted to do. This takes the usually staid action-selection phase of the game and turns it into something strategic. Thus far, limited action selection has largely fallen into three categories: hidden action selection, constrained action selection, and conflicted action selection.

Figure 4. Options for limited action selection.

Hidden action selection means that players don’t know what their opponents are doing when they make their choice.

Simultaneous Selection. Players all secretly make an action choice and then simultaneously reveal them, before taking those actions in some set order. Many games introduce conflict into simultaneous selection by punishing players if they make the same choice as someone else, but this isn’t a strict requirement. Basari is an example of a traditional simultaneous-selection game where players only get to take their action freely if they don’t match someone else (and otherwise have to negotiate to see who pays whom for the action). Diplomacy, Wallenstein/Shogun, and Roborally are all examples of Programmed Move simultaneous-selection games, where the limitations come mainly from the surprising interactions of the moves.

Constrained action selection take actions out of the traditional menu and places them on some other component, which in some way limits action choices.

Rondels. The rondel is a mechanism used almost (but not entirely) exclusively by Mac Gerdts. Actions are placed in the slices of a pie chart, and a player moves a few places clockwise on that chart each turn to select an action. This usually gives a player an option of 1-3 actions on a turn and limits most actions to being taken only every two or three turns.  Gerdts’ Antike is a prime example of his simple style of roundel. Examples by other designers include Wolfgang Sentker & Ralf zur Linde’s Finca (which has a variable rondel, and which uses game state to determine distance moved) and Stefan Feld’s Trajan (which uses a Mancala mechanic for rondel movement). This is a very game-y mechanic that has little linkage to game theme, but it nicely constrains choices and introduces tactical and strategic consideration.

Deck ManagementEffectively a rondel system with different components and slightly different constraints. Here, each player has a personal deck of action cards. They play through them one a turn and at some point can spend a turn to get back all their action-cards. Like the roundel, this prevents players from taking the same actions in close proximity to each other, but offers a different sort of player control, as the player can choose when to “waste” a turn, just to get their cards back. Gerdts proved how similar roundels and deck management are by turning to the latter mechanic starting with Concordia. Other examples include Kreta and Assault of the Giants.

Deck Building. A second take on action constraints based on cards. This time, the players purchase their own action-cards turn by turn, using the resources on their cards. This means that sometimes cards are played as non-actions: for their currency. But sometimes they’re also played to have specific effects. It shows have action representatives (such as cards) can actually have multiple effects. Dominion introduced this trend, and there have been a billion games since.

Card Drafting. A third take on action constraints based on cards. Here, the players are given a random set of actions on cards and each take one and pass the rest on the next player, who takes one … Card Drafting is often not lumped in with other action-selection mechanisms because there tends to be a disconnect: the card drafting and the card play tend to be different phases, whereas most action selection is an atomic action. However, when you put the card drafting and play together you definitely have a full-fledged action-selection phase. 7 Wonders is a card drafting game that only feels marginally like action selection because its cards tend to give you resources and points and feel less like actions, but Midgard and its successor Blood Rage more obviously show how card drafting is fundamentally actions selection, because the drafting gives players actions that they later use.

Card Drafting also introduces conflict to action selection, because when one player takes an action-card, the other players can’t. This is the whole basis for the last category of limited-style action-selection methods.

Role Selection. Card drafting wasn’t in much use in the eurogame industry when role selection first appeared. Nonetheless, the two categories of play look almost identical when you consider their core mechanics, especially when you look at one of the foundational role-selection games, Citadels. There, one action-card is removed from a collection of eight static  cards, then the rest are given to the players who draft them around the table. The main difference between this and regular card drafting is that the actions are all known and regularized, not a random set of cards. However, in most role-selection games, role cards are set out on the table, then taken one by one, as was the case with Verräter and Meuterer. 

Phase Selection. A very common variant of role selection that gives everyone the option to participate in a role action when it’s selected. In fact, this is so common that many consider it part of role selection’s core definition. (I remain agnostic, other than pointing out that the earliest role-selection games didn’t include this element.) In games like Puerto Rico and Race for the Galaxy, everyone gets to use a slightly lesser version of the action; while in Glory to Rome they participate only if they have a matching action card already available on their board, and in Eminent Domain they do so only if they have a matching action card in their hand.

Worker Placement. And finally, there’s the ever-popular (or not!) mechanic of worker-placement, which is a lot like role selection (or card drafting), but where the actions are instead printed on the board, where they cost a resource to take, and where the set of actions might grow over time. Caylus and Agricola are the two founders of this particular boardgaming dynasty, and there have probably been hundreds since.

Other mechanics such as Auctions can also be used for action-selection, but even moreso than Card Drafting and Deckbuilding, that tends to be a minor element in the overall usage of a backend “core mechanic”. Actions can also be determined by the random draw of cards or tiles or the random roll of dice. There are really countless possibilities. The above listing is simply some of the most common and most explicit ways to choose actions in a decision-making system.

Having not just multiple actions and multiple types of actions but actions with strategic nuance is another foundational building block for eurogames, placed right atop the possibility for basic action selection. By hiding player actions, limiting player actions, or conflicting player actions, a game design can introduce interesting gameplay into the very choice of action, which once upon a time was a simple decision.

It’s no surprise that this style of mechanic, in its many variations, has been a constant obsession of eurogames, starting with simultaneous selection, then moving through role selection, rondels, card drafting, deck building, and deck management until it hit the worker placement that has taken over today’s industry. There will probably be a next mechanic and a next mechanic that also limit actions in different ways.

It’s also no surprise that I’ve also written about these mechanics a lot, including: a discussion on the relationship between rondels and deck management; a three-part ode to “role civilization” games (an introduction, a look at empire games, a look at galactic games), which are a particular branch of role selection; a look at the definition of worker placement; and a whole slew of deckbuilding articles.

Conclusion

Action selection is just the beginning. After a player chooses an action, they then have to make choices about how to apply that action in the action execution phase. Sometimes, those choices might be very big, as in Tigris & Euphrates, where players realistically have dozens of choices for where to place each tile or marker. Sometimes, those choices are degenerated to the point of choicelessness: in Agricola, if a player takes a resource from a resource action-space, he usually has no additional option.

However, a particular action-selection method does not require a particular action-execution method. It’s likely that if one is wide, the other will be narrow, to avoid overwhelming the players, but even that isn’t requirement. Four menu choices could have no execution choices or a million execution choices; a dozen worker-placement spaces could similarly lead to no execution choices or a million of them.

In the next article, I’ll more briefly look at the options of action execution and action resolution, before talking about how this ties in to the core mechanics of a game’s backend. Then I’ll finish up this series in a third article with a number of case studies, to take the theory out for a test spin. I’ll see you back here for that (after my usual article on the last quarter’s games played, as the top of October).


The little graphical icons that accompany this article’s figures are borrowed from a comprehensive set of icons that artist Keith Curtis designed for myself and Christopher Allen to help illustrate our upcoming book on cooperative design, Meeples Together. Except the roundel icon. There apparently aren’t rondel-based co-op games (yet!).

You’ll also see some bonus articles about co-op design here starting next month, when we put Meeples Together out for Kickstarter. You can sign up now for a one-time email when we go live with that Kickstarter.

Liked it? Take a second to support Shannon Appelcline on Patreon!

The original article can be found on the great Mechanics & Meeples