Friday, January 9, 2009

A quick post about Feedback Loops

Since I mentioned feedback loops in my last post, before moving on to Part 3, I thought it would be appropriate to include a link to a YouTube video by Scott Nicholson where he includes a discussion of feedback loops by a gentleman named Joris Dormans. In this video, Mr. Dormans uses Power Grid as an example of what he's talking about:

The "Music" of Game Design Part 2

(The diagrams I posted in my last entry were too small to be seen well. So I'm going to offer some clarification/review in this post by using close-ups of the main diagram.)

The essence of the Tri-Level Resource Exchange Model is that you take resources, convert them to intermediate resources, and then convert those intermediate resources into resources that actually feed into the winning condition. So, here I have a diagram of the "Alpha Node" or point of choice where the players choose how they want to go about using their Level 1 resources to acquire Level 2 resources:

After the Alpha Node point of choice, the players then have to decide how they want to convert their Level 2 Resources into Level 3 resources, thus creating a second point of choice or Beta Node:

Now, from the decisions in the Beta Node regarding how to use one's Level 2 resources, come one of three possible types of Level 3 resources:

The VP resources feed into the winning condition:

The "dynamic" Level 3 resources actively feed back into the system:

The feedback loops are of one of three possible types and one of two possible sub-types:

External Game Generated types include situations where the game actively influences what a player can do based on that player's Level 3 resources.

External Player Generated types are where one's Level 3 resources in some way interact with or prevent what other players can directly do to you at the choice points.

Internal Player Generated types are where you simply help yourself in some way that doesn't involve a decision by other players or a "conditional decision" by the game.

Some examples will help illustrate my points:

-A perfect example of an external game generated feedback loop would be the changing of turn order in Power Grid. It is a direct influence from the game exerted on the players based on the amount of Level 3 resources they have.

-An example of an external player generated influence would be situations where you obtained cards or some other sorts of protections or mitigations from the direct attacks of other players.

-An example of an internal player generated influence would by like the "Hacienda" in Puerto Rico. It gives the player a chance to make a choice that can influence one of the interactions in the game - but the player doesn't have to do so.

The two sub-types are either Conditional or Random. With the Hacienda, it is an internal player generated influence but it's subject to randomness. With turn order in Power Grid, it is a game generated influence that is strictly conditional based on a set of criteria.

So, the Alpha and Beta Nodes are not only decision points where player choice is introduced into the system, but they are also points of entry for influences:

So, now that the context of the model has been established, two more assertions I have about what makes for "good" game design will make sense:

-Postive feed back loops for players should be counter-balanced in some way by some other aspect lest a run-away leader problem emerge.

-In situations where External Player Generated influences result in direct loss of resources by one player, having more than two players presents a situation where the game can suffer from triangulation problems (I've written previous posts on this concept).

Edit: One more thing to include with this model before moving on is that Internal Player Generated Feedback influences don't always simply influence the choice nodes. Sometimes the manifestation of this type of feedback influence... the fact that a Level 3 resource doubles as a Level 1 Resource:

Examples of this include how Settlements in Settlers of Catan, Buildings in Puerto Rico, and Buildings in Caylus are the results of Beta Node choices regarding Level 2 resources but they also serve as Level 1 resources in helping create Level 2 resources.

*In part 3, I'll discuss a second model of design called the "Tri-Stage Resource Transition Model". It's a model that is decidedly different from the Tri-Level Exchange model in its form.

Thursday, January 8, 2009

The "Music" of Game Design: Part 1

Board game design and music composition are actually more similar than one might at first think. Both are artistic endeavors. The results of both are judged by others as being "good" or "fun" or as being "bad" or "unpleasant". There are many different types of board games just as there are many different types of music compositions. I assert that, just like how music has fundamental principles behind what tends to make for "good" music and what doesn't, game design also has fundamental principles behind what makes for a "fun" game and what doesn't.

I minored in music while in college and, in my spare time, I'm a drumline instructor for a local high school. In my studies, I've spent some time developing an understanding of many basic principles of music theory not only for personal improvement but also because I occasionally compose percussion features for the ensemble I help instruct (which involves writing parts for mallet percussion as well as drumline "battery" parts). These principles manifest themselves as generalizations that tend to be true of music that people find to be "good" or pleasant. By knowing about these principles, I save myself lots of time. I can identify why certain passages won't work with the rest of the piece before I get very far into composing them.

Contrast this process with someone who is isn't aware of these principles and who is working purely from a trial and error basis. They try some notes and find out that they aren't working very well. They don't know why but they know they don't sound good. So, they try some more notes and they, again, don't work. They keep going and keep going and keep going. Perhaps many months later, they have something approaching a basic skeleton of a piece that could work but still has many problems with it. For me personally, I can sketch out basic arrangements fairly quickly as I am already aware of basic chord structures, inversions, typical chord progressions within a key, pivot chords, cadences, four-part voice writing "rules", generalities for what makes for good arrangements, etc. Though there are others who are definitely more talented than me in this area, I have a huge leg up on those who are approaching it from a trial and error basis.

The same idea can be applied to game design and its challenges. In both endeavors, a person can go through a lengthy trial and error process and eventually arrive at something decent. Or, if they understand basic principles of theory behind the work they are engaged in, can save themselves a lot of time and compose/design much better works a lot faster than if they were ignorant of those principles.

In one of my game designs recently, I found that the game was "okay" but that it simply lacked that extra dimension that I wanted for it. The trick was, I didn't know what that extra dimension was. I found myself trying to articulate a vague notion that I couldn't, at first, quite verbalize. The same thing can apply to music. If a person doesn't like a piece, if they have the vocabulary, terminology and jargon to call upon, they can express themselves much more precisely than by simply voicing vague notions. So, I set upon the task of verbalizing my notions with respect to my prototype.

What I arrived at was a series of "discoveries" or "conclusions" about specific models of game design that I assert can help one in the process of identifying not only problems in a game design but also what may be lacking or not present yet that could help a game reach the next level. As I arrived at these conclusions, I found that they felt very much like many of the typical principles of composition that I encountered while studying music.

As I said before, there are many types of music and, accordingly, there are many types of game designs. Just as music has its concertos, arias, symphonies, four-part choral arrangements, and percussion features, game design features many types of models that we often describe from the standpoint of one of the game's mechanics (i.e. economic, pick-up-and-deliver, negotiation, etc.). I assert here that one particular model I have been able to identify in many successful games is what I have termed the "Tri-Level Resource Exchange Model". Here is a graphic of it:

For a larger version of this image, click here and then zoom in to see the details.

Though this model is an attempt to capture all of the elements in a game of this type, not all games that employ this model incorporate every element.

Basic Elements
At its core, the "Tri-Level Resource Excange Model" is exactly what its name implies. It is a model where the players use "Level 1" resources to acquire "Level 2" or "intermediate" resources which are then used to acquire "Level 3" resources. The winning condition in games that implement this model are usually based on the acquisition of certain types of Level 3 resources.

It is the presence and use of "intermediate" or "Level 2" resources that creates the intrigue within the system. Having intermediate resources results in there being two interaction "nodes" or points of choice where the players have to make plans and account for "interference" as well. Let's look at the first node:

The essence of this first node is using "Primary" or "Level 1" Resources to acquire usable secondary or "Level 2" resources either through direct production or through exchange. Something that is critical to understand about this model before going on is that certain resources can act on multiple levels but the model still holds true.

The key to the definition of secondary or level 2 resources is that they are merely acquired means to an end. Secondary resources do not pay directly into the winning condition, they simply facility the process of acquiring the resources that do.

Tertiary or level 3 resources are typically of three types:

They are either:

-Static VP's (these types of resources simply serve as points and don't serve any other purpose)
-Dynamic VP's (these types of resources serve as points but also feed back into the system and dynamically impact the processes that are in motion)
-Dynamic Non-VP's (these types of resources purely feed back into the system and affect the process that are in motion but don't serve as any sort of points in and of themselves)

Perhaps a couple of examples would be useful:


Puerto Rico
In the game "Puerto Rico", you use primary resources such as buildings, plantations, and workers to acquire secondary resources or "goods". These goods do not count directly towards the winning condition of the game (i.e. having the most points) but they do act as a means towards acquiring shipping points (a Level 3 Resource which contributes to the winning condition) or towards acquiring money which is then used for purchasing buildings (a Level 3 recource which also contributes to the winning condition).

Notice, however, that buildings act as a level 1 resource and as a level 3 resource. They are a means to acquiring the intermediate resources that lead to more acquisition of the resources that will win the game and they are also part of the resources that will win the game. In this respect, they act on multiple levels. However, shipping points are purely level 3 resources. They do not contribute to the acquisition of more money, more goods, or more buildings. They are simply static VP's.

Buildings, on the other hand, are dynamic VP's. They not only contribute to the winning condition by representing points but they also dynamically feed back into the processes of the game.

In Puerto Rico, there are no truely Dynamic Non-VP's. However, despite the fact that Puerto Rico doesn't utilize one of the three possible types of Level 3 resources, it is still a "Tri-Level Resource Exchange" game.

The Settlers of Catan
In the game "The Settlers of Catan", we again see the multi-level nature of certain resources. Settlements, Cities, and Roads are all level 1 resources which facilitate the acquisition of level 2 resources (the wood, brick, ore, wheat, and wool cards). These level 2 resources are then used to acquire Level 3 resources which are of one of three types:

-Static VP's The occasional development cards that grant 1 Victory Point serve no other purpose than to simply represent a point for the player.

-Dynamic VP's The creation of more settlements and cities results in more points for the player. However, cities and settlements dynamically affect the number and types of secondary resources the player can acquire. Thus, they dynamically feed back into the system in addition to their serving as points.

-Dynamic Non-VP's The developement cards that allow for acquiring more resources or for moving the robber dynamically feed back into the system but don't serve as a direct source of points.

Some Observations and Assertions:

Part of the enjoyment of music is the creation of tension and then the release of that tension through resolution. In other words, you establish a tonal center, stray from that tonal center, and then return to it in a pleasant way. With respect to games, a friend of mine named Peter aptly phrased part of the "pay-off" that players get from playing a game in that a game grants them the chance to formulate a "plan" and then attempt to execute or carry out that "plan".

In Tri-Level Resource Exchange games, the plan players get to formulate comes from their attempts to acquire various types of intermediate level resources and to then creatively convert those secondary resources into Level 3 resources in the best way possible. If I look at my previous prototype from the standpoint of comparing it with the "Tri-Level Resource Exchange Model", I realize very quickly that there wasn't a second or Beta decision node. Players made choices about acquiring different types of secondary resources but those secondary resources were then simply converted into points at the end without the opportunity for the players to make choices about what to do with the resources they acquired. It was just a straight up acquisition game. Thus, it lacked that second level or second node of choice I've termed the "Beta Node" in my model.

Also, with respect to forming a "plan", there must be a sufficient diversity of second level resources to pursue in the game such that there is a "palette" of choices the players can use to create a unique plan. If there are too few types of secondary resources, then the players don't get the opportunity to exercise some individuality or creativity in the process. If there are too many, the players can pursue parallel courses and never get in each other's way.

The Power of "5"

A pattern I've noticed among several games that implement this model (as well as among other games that don't use this tri-level exchange model) is the number "5" with respect to the types of resources that players can go after in the game. It seems to serve as a natural number in that having 5 types of secodary resources allows for sufficient creativity and diversity in pursing a plan while not providing so many options such that the players don't get in each other's way. Here are some examples from well known games - including some that don't use the Tri-Level Resource Exchange Model:

-Puerto Rico uses 5 types of goods (corn, indigo, sugar, tobacco, coffee)

-Caylus uses 5 types of goods (food, wood, stone, cloth, gold)

-Settlers of Catan uses 5 types of goods (wood, brick, ore, wheat, wool)

-RA uses 5 types of tiles for in-game scoring (God, Pharaoh, River, Civilization, Gold)

-The Princes of Florence uses 5 factors that contribute to points scored via works (Buildings, Landscapes, Freedoms, Jesters, and other Work Cards)

-Power Grid uses 5 types of sources for powering the power plants (coal, oil, garbage, nuclear and "wind" - for the free power plants)

-Lost Cities uses 5 colors that players attempt to score points on.

-China uses 5 types of colored cards for playing (red, orange, yellow, green, and purple)

-Niagara uses 5 types of colored gems (purple, white, brown, blue, and pink)

-Transamerica uses 5 different colored cities you are trying to connect each round

Now, I'm not saying that a designer has to incorporate 5 different types of secondary resources into his game for it to work. I'm merely observing that this number comes up a lot and that it's worth paying attention to as it works for a number of games that have proven successful in their appeal.

Some games use 4:

-Saint Petersburg uses 4 types of cards (workers, buildings, aristocrats, and upgrades)

-Tigris and Euphrates uses 4 types of tiles (red, black, green, and blue)

-Loot uses 4 pirate colors (orange, blue purple, green)

-Carcassonne allows for 4 ways to score using meeples (thief, knight, farmer, monk)

-Blokus uses 4 colors (red, yellow, green, blue)

...while others use 6:

-Ingenious uses 6 colors (red, blue, yellow, green, orange, and purple)

-Ticket to Ride uses 6 colors (red, orange, yellow, blue, green, and pink)

...but 5 seems to be the most common number I encounter in these games. Thus, if I'm trying to create a game that uses mutliple types of resources, I'm probably going to start out with five as a safe starting point and then alter it based on feed back and playtesting.

*Part 2 to follow. In it, I will complete my discussion of the Tri-Level Resource Exchange Model by going into the "Feedback Loops" one finds in games of this kind. I will also introduce and discuss another model of game design called the "Tri-Stage Resource Transition Model" - which is slightly different in its construction.