# Time For Data

Now it’s time to get into some numbers. One might even say a lot of numbers. And you and I, we’re going to get all up in their business. Don’t let this scare you, though; Numbers are cool! It’s fun to talk about Models and Goals, like we did above, but eventually you have to start doing some honest research. You can’t just talk about systems, you gotta look at them, touch them. What? How can you touch something so abstract? Through data.

Let me tell you a little story from my early days. I’m working on my first high profile game, let’s call it Duke of Fisticuffs, and I am currently having my entire concept of game design completely shattered — working with talented people will do that. But nothing creates a better foundation for building than complete annihilation. See, from this destruction, I saw the light at the end of the tunnel; I saw the person I wanted to become.

It was only days after my epiphany that the frame data for Street Fighter 4 began to make its appearance on the net. I made it my mission to study, nay scrutinize, that data, and when the game finally came out in America, I would play the game with my laptop sitting beside me, comparing the two. I would not be the designer I am today without those early days spent staring at numbers and wondering. Numbers are important, not scary, for they are storytellers.

Mulder: I’m reading the box scores, Scully. You’d like it. It’s like the Pythagorean Theorem for jocks. It distills all the chaos and action of any game in the history of all baseball games into one tiny, perfect, rectangular sequence of numbers. I can look at this box and I can recreate exactly what happened on some sunny summer day back in 1947. It’s like the numbers talk to me, they comfort me. They tell me that even though lots of things can change some things do remain the same. It’s . . .

Scully: Boring.

Before we get to data, however, let’s review what we’ve learned so far.

# The Story So Far

Crafting systems are a dense, complicated subject, but they are also incredibly simple. The journey started with a general definition of crafting.

Crafting System: A large game system that converts one or more in-game items into some other item or items. It is comprised of three parts: One, an Input Stream of Resources; Two, a system of Manufacturing; and Three, an Output Stream of Goods.

Next, like all tasks, we studied the goals of the system. All crafting systems have one goal, the increase of economic capital, and this capital comes in three forms:

  • Human Capital: you, the person behind the character.
  • Avatar Capital: the intrinsic abilities possessed by your character.
  • Physical Capital: the extrinsic equipment you carry on your character.

Crafting systems, as we have defined them, have a simple 3 step process: Input -> Manufacturing -> Output. Combining this with certain patterns led to a model for crafting systems, and this model allows us to create archetypes:

  • Low Input – Simple Manufacturing – Low Output (LSL)
  • High Input – Simple Manufacturing – Low Output (HSL)
  • Low Input – Complex Manufacturing – High Output (LCH)

These archetypes can be combined with one (or more) of 5 basic intentions:

  • Status: increase your social standing.
  • Combat: add to or enhance your combat options.
  • Access: serve as a gating mechanic.
  • Trade: allow for market specialization.
  • Exploration: serve as a driving force in exploring the world.

We’ve created a model, we’ve stated the goal, and we’ve studied the intent (SCATE). Now it’s time to look at data.

# The Data

I researched a number of different games while writing this, but I focused on eight that I was familiar with.

  • World of Warcraft (vanilla)
  • Dead Rising 2
  • Dead Island
  • Minecraft (version 1.8)
  • Far Cry 3
  • Fallout 3
  • Fallout New Vegas
  • TES: Morrowind

The process started in excel by entering all potential goods and their requisite resources. From that I was able to pull out all of the unique resources (our Input Stream) and calculate the total number of goods that can be created for each resources (our Output Stream). An example:

# Dead Rising 2

In Out Products Type S C A T E Fun
70 1 50 HCL 0 2 0 0 2 3

It’s important to understand what this information is saying. Inputs, as we previously defined, are items that can be inserted into the magic box that is our crafting systems. In Dead Rising 2, for example, it would be all the things you pick up in the game, such as Boxing Gloves, a Flashlight, or a Lizard Mask. “Input: 70” is saying there are 70 items that can be potentially used in the crafting system, far more than any person could reasonably keep in their head. This would be, by our definition, a High Stream.

The output can be a little confusing, but stay with me. Remember: the output stream is defined by the total number of items that can be crafted from a single input. Some items in Dead Rising 2, the Battery for instance, are used in multiple goods, while something like the Baseball Bat can only be used to create one thing. If we were to study the system for just these two items it would look like this:

  • Battery — Output 4
  • Baseball Bat — Output 1

The output is dependent on the input. The listed output for Dead Rising 2 represents the average output of all inputs, which gives us an actual number of 1.4, but I rounded down.

In short, the average item you pick up in Dead Rising 2 is constrained to only one possible good. That is a very good thing, as it turns out, for it improves and meshes with the flow of the game: You generally have “a something” in your inventory, and, through the design of the crafting system, you start asking yourself, “Ok, where the hell can I find X, so that I can make Y.” It gives you purpose.

Next, for all the various crafting systems I indicate (as far as I can tell) the intentions the system was trying to serve: 0 for nothing, 1 for minor, and 2 for major focus. Here you can see that Dead Rising 2 has two major intentions: Combat and Exploration.

Last, but certainly not least, I ranked every crafting system on a 5 point “fun scale”. I realize it seems arbitrary, but it’s just my subjective measure of fun, and I tend to like 5 points scales. In the case of Dead Rising 2 a 3 is actually pretty good (I’m picky).

Ok, now that we’ve got this example under our belt we can look at the full list of data:

Game In Out Produced Unique Type S C A T E Fun
Wow (Tailor) 5 47 234 234 LSH 1 1 0 1 2 2
Wow (Leather) 5 64 322 322 LSH 1 1 0 1 2 2
Wow (Blacksmith) 10 40 399 399 LSH 1 1 0 1 2 2
Wow (Alchemy) 28 7 194 194 HSL 1 2 0 1 2 3
Dead Rising 2 70 1 51 50 HCL 0 2 0 0 2 3
Dead Island 116 17 431 69 HSH 0 2 1 0 2 1
Far Cry 3 28 2 53 53 HSL 0 2 2 0 2 4
Fallout 3 28 1 7 7 HSL 0 2 0 0 2 2
Fallout New Vegas 180 3 209 160 HSL 0 2 0 0 2 4
Minecraft 79 3 162 160 HCL 2 2 2 1 2 5
TES: Morrowind (Alchemy) 105 179 78 78 HCH 0 2 1 0 2 4

In this full version I also included the number of unique products. In the case of Dead Island, for example, you can produce over 400 items, but most of these are duplicates. This is all very fascinating, but it is feels we are missing something critical — you can feel it too, no doubt. I have completely overlooked the kinds of goods you are crafting.

# Systems Within Systems

A look into the kinds of goods we are crafting demands we delve a little bit deeper. This requires some new terminology, which we shall take from my article on economics and video games. I defined three kinds of items in a synthetic economy:

Consumable Product: Products that are destroyed after being used for the first time. Such as a health potion.

Dependable Product: Products that persist after being used, but they depreciate in value over time. The quicker something depreciates in value the more it begins to act like a Consumable.

Permeable Product: Products that are never consumed and never depreciate in value over time. Most items in a video game fall into this category.

For fun, and without understanding the import, I labelled the goods in a few crafting systems by their product type. It became clear, however, that the success of certain systems hinged on understanding the kind of products your system was creating (which, when you think about it, should be stupendously obvious). This observation tends to reveal systems within systems, where the outward appearance can be one way, but on the inside it acts in other ways. We shall look at two: Far Cry 3 and Minecraft.

# Far Cry 3

Far Cry 3 is a clear example of this. Everything you make with animal parts (with the exception of the fire arrows) is a Permeable. You are making some new thing that that will permanently be with you. Everything made with the plants, however, is a Consumable.

Let’s treat the Far Cry 3 crafting system as if it is two independent systems: the things you can make with plants, Consumables; and the things you can make with animal parts, Permeables.

Far Cry 3 Input Output Products Type S C A T E
Consumables (Syringes) 5 4 13 LSL 0 2 1 0 1
Permeables (Equipment) 22 2 40 HSL 0 1 2 0 2

In this case, we are learning a lot more about the system by breaking it up this way. Syringes are primarily about combat, while the equipment is primarily about access and exploration. Once I made this discovery, I started to pay attention to how I was playing the game, and it was interesting to notice how important this was to the flow of the experience.

I never cared about finding plants, but I was usually concerned with my current stockpile of health syringes. Because there were so few plants to worry about (low input stream) it was all very organic: I would simply grab the plants I needed as I was traveling around. The equipment, however, forced me to make meaningful choices about where I travelled. I’d find myself asking, “Man I really need a bigger money pouch, what do I need for that?” This would force me to create a little mini mission for myself. This is an example of a crafting system that perfectly fits into the context of the entire package. This concept of a “search” versus “find” mentality for resources is even more apparent when looking at Minecraft.

# Minecraft

Breaking up Minecraft’s goods into Permeables, Dependables, and Consumables results in a similar illumination: There are systems hiding within systems. Let’s see what the numbers say:

Minecraft Input Output Products Type S C A T E
Consumables 18 1 12 HCL 0 1 0 1 1
Dependables 8 9 44 LCH 0 2 2 1 2
Permeables 67 3 106 HCL 2 0 0 2 0

This was interesting, but I think the more appropriate view of this data comes when you take a closer look into the dependables. Consider that Minecraft is really built around 5 “base” resources: Sticks, Wood Planks, Cobblestone, Iron Ingots, and Diamond Gems. If we separate the crafting system and look at just items that involve these base components we see something totally different than if you looked at the system as a whole. Let’s compare the two.

Minecraft Input Output Products Type
All Items 79 3 162 HCL
Base Resources 5 21 75 LCH

Each of these base goods can make, on average, 21 different things; and, what’s more, result in over 75 different goods — that’s almost half of all possible goods in the entire game! This heavily contrasts with the view of the game as a whole, but, like Far Cry 3, this is important to the game’s flow.

Part of what makes Minecraft work is that the first time you build a wooden tool, due to the nature of the crafting system, you instantly grasp the potential implications (I can make this same tool with stone, or iron). It draws you to go out searching for new resources that you intuitively know exist. Knowledge of iron and diamond are not even required, for it would be natural for you to expect that something out there will slot into the system. This kind of intuitive expansion is nearly impossible when a game uses obtuse recipes, nor would it be feasibly manageable in a game with more than than 5 “base” items — the standard limit on our memory.

In both of these cases (Minecraft and Far Cry 3) we have discovered systems within systems, where the complete image belies sub-archetypes that are driven not only by different kinds of goods, but also by different intentions. But of course, this all leads us to the critical question: how do we use this analysis to construct good crafting systems?