1 of 78

Daniel Cook

Company�Spryfox.com

​

Game Design Essays

Lostgarden.com

​

Games

Tyrian, Triple Town, Alphabear, Realm of the Mad God, Beartopia, Steambirds

​

Contact

Twitter: @danctheduck

danc@spryfox.com

2 of 78

Agenda

  • Why to model your game
  • Seeing Interaction Loops
  • Mapping Skill trees
  • Basic Diagnosis

3 of 78

Why to model your game?

Why bother with theory

when winging it is easy!

4 of 78

Why formal tools?

  • Games spend most of their development broken.
  • Good conceptual tools help us find and fix issues faster.

5 of 78

What are formal tools

Simple, pragmatic models help us:

  • Reason about our design.�If the structure is X, Y should occur.
  • Debug designs. If Y doesn’t happen, why not?
  • Build richer structures. �Steel skyscrapers, not mud hovels.

​

​

6 of 78

Formal tool limits

These are tools, not miracles

  • Practice matters. You need to become skilled
  • Aesthetics matter. These tools exist in service to a creative vision. They aren’t going to make a good game for you.

​

7 of 78

The Three Perspectives

Existing lenses for looking at �game design

8 of 78

Design-centric

  • Prototyping
  • Iterative development
  • Playtesting
  • Metrics

9 of 78

Rules-centric

  • Sources, Sinks Resources
  • State machines
  • Procedural generation
  • AI

10 of 78

Player-centric

  • Interaction loops
  • Koster’s Theory Fun
  • Behavioral Economics
  • Play styles �(Bartle, PENS)

​

11 of 78

1. Interaction Loops

What they are

How to see them

12 of 78

What is an Interaction Loop?

A model for understanding how a player learns and uses a specific skill in an interactive game or application.

13 of 78

14 of 78

1. Learn

15 of 78

2. Decide

16 of 78

3. Action

17 of 78

4. Rules

18 of 78

5. Feedback

19 of 78

Universal

  • A primitive of interaction design
  • Exist in every interaction in every game.
  • User Interfaces, Game feel, Boss fights, Tutorials, Metagames!

�Goal: Teach designers to see them.

20 of 78

How they are used

  • Understand how players gain specific skills
  • Balance game feel and pacing
  • Debug issues with any of the above

21 of 78

Example: Jumping in Mario

22 of 78

0. Setup: You get the controller

23 of 78

1. Learn: What do you understand?

  • You recognize buttons! (Schema activated!)

24 of 78

25 of 78

Concept: Schema

Existing mental models of the world

  • Conceptual tools and patterns you know work
  • Validated by some past experience
  • Readily available! Little thinking or effort required.

​

26 of 78

2. Decide. What do you do?

  • Options? Buttons can be pressed!
  • Cost? Pretty low
  • Benefit? You might discover something
  • Other opportunities? Not yet!

27 of 78

3. Action. Perform the action!

  • You “Press A”

28 of 78

4. Rules. The game processes

  • Jump begins in the code
  • Physics systems start calculating how to make Mario jump
  • Time advances

29 of 78

Concept: The Black Box

From the player perspective,

  • Rules and dynamics are a black box.
  • They start as unknown.
  • They can be learned.
  • They can be manipulated to reach goals

This process is called Mastery

30 of 78

5. Feedback. Show success or failure

  • Success: Something happened! You see Mario jump up and then fall down to the ground.
  • Failure: Nothing happened!

31 of 78

And now we loop

32 of 78

33 of 78

1. Learn. Update mental model

You might have learned…

  • Pressing a button make Mario move (Causal Link)
  • Of the buttons, A is the right button (Causal Link)
  • Gravity: You might observe that there is gravity. Things fall in this world (Schema)

​

34 of 78

Concept: Cause and Effect

Players link that Event A causes Event B

  • Association: Players learn this at first fuzzily “When I do something, something happens”
  • Sampling: The more they practice, the more certain they become.
  • Grokking: When they understand the causal link with certainty and can repeatedly cause the effect

​

35 of 78

Skills are Cognitive Tools

When a player builds a causal model, they get super powers:

  • Reason about the world: If I wanted state B, I should do A.
  • Manipulate the world: To get state B, I do A.

​

36 of 78

Grokking skills is delightful

Gaining a tool to get what you want is delightful.

​

You are powerful!

​

A piece of what we call fun is this moment of understanding. See Raph Koster’s Theory of Fun

​

37 of 78

38 of 78

Activity�Seeing Interaction Loops

39 of 78

Setup

  • Split into groups of 3-4 people

40 of 78

  1. Game: Pick a game (Triple Town, Alphabear or other that is familiar to the class)
  2. Skill: Pick a common verb a player does in the game.
  3. Existing Schema: What does the player already know how to do with regards to this action?
  4. Missing Skills: What doesn’t the player already know how to do with regards to this action?
  5. First Action: What does the player do to experiment?
  6. Rules: What game rules occur when the player tries something?
  7. Feedback: What feedback does the player get in response?
    • In the case of success?
    • In the case of failure? Other failure cases?
  8. Learning: What Cause and Effect lessons did they figure out?
  9. Decision: What are costs, benefits and other options they weigh when choosing what to do next?
  10. Next steps: What does the player try next?
  11. Grokking: How many times does the player go through this loop before they feel confident they can do perform the skill?

41 of 78

Review

Each team presents their interaction loop

42 of 78

Questions?

Any questions people might have about interaction loops

43 of 78

Mapping Skill Trees

What they are

How to see them

Architectures

44 of 78

Interaction Loops are Legos

  • You can build new skills out of mastered skills
  • Current skills were once built out of even simpler skills

45 of 78

Compound Skills

46 of 78

Compound Skills

  • Skills are composed of other skills
  • By first learning lower level skills, player unlock the ability to master more complex higher level skills

47 of 78

Skill Trees

You can map related interaction loops as a tree diagram.

Tap A

Dpad

See Goomba

Jump

Walk

Air Control

Kill Goomba

Hold A

Jump Position

Variable Height Jump

48 of 78

Skill Trees

  • Simple: You can go all the way down to really basic action fast, frequent actions like pressing buttons
  • Complex: And all the way up to compound slow infrequent strategies.

49 of 78

Concept: Frequency

50 of 78

Concept: Frequency

  • Game Feel: Loops that occur <200ms
  • Moment-to-Moment: 1-5 seconds
  • Minute
  • Session
  • Daily
  • Weekly

51 of 78

Activity�Building a Skill Tree

52 of 78

  1. Game: Pick a game you are familiar with
  2. Skill: Pick a common skill in the game
  3. Build the start of the tree:
    • What skills does the player already need how to perform in order to do this this skill?
    • What skills are needed for those skills?
  4. Build the end of the tree: What compound skills can the player perform once they’ve mastered this skill?
  5. Fill in the tree
    • Are there any high frequency skills you are missing? Game feel for example. Where would they fit?
    • Are there any low frequency skills missing? Strategies for example. Where would they fit?

53 of 78

Basic Diagnosis

Common problems

How to see them

54 of 78

What’s the point again?

Skill tree = a map of your game.

You can diagnose problems with it.

  • What skill is the problem?
  • What step in the interaction loop is broken?

55 of 78

Step 1: Identify broken skill

Tap A

Dpad

See Goomba

Jump

Walk

Air Control

Kill Goomba

Hold A

Jump Position

Variable Height Jump

56 of 78

Step 2: Zoom in on skill

57 of 78

Failure in Skill Tree Architecture

Common issues

  • Incomplete map: You missed an important lower level skill.
  • Pacing: You didn’t give time or enough iterations to learn lower level skills
  • Order: You are failing at teaching a lower level skill

​

58 of 78

Failure to Master Skill due to Feedback

Common issues

  • Missing: There wasn’t any feedback!
  • Ignored: Feedback was ignored
  • Feedback didn’t suggest a cause and effect link
  • Noise: There was lots going on so I couldn’t focus on the high priority feedback.

​

59 of 78

Failure to Master Skills due to Action

  • Player is unable to execute action.
  • Players struggles to replicate the action

�These are often Rules issues. ��It is rarely the player’s fault.

​

60 of 78

Failure to Master Skills due to Rules

Common issues

  • Rules were hard to convert into simple cause and effect models
  • Rules were overly dynamic so I didn’t get to see the same thing enough times to see a pattern

61 of 78

Failure to Master Skills due to Learning

Common issues

  • Player focused on the wrong schema
  • Player didn’t have enough time to process
  • Player didn’t have enough opportunities.

​

62 of 78

Burnout

You learned the skill! But stopped! Ahhh!

  • Happens at the Decide step inside the interaction loop
  • Common issues
    • Cost is too high
    • Benefit is too low. Not useful in future strategies
    • There were better options

63 of 78

Burnout

Tap A

Dpad

See Goomba

Jump

Walk

Air Control

Kill Goomba

Hold A

Jump Position

Variable Height Jump

64 of 78

Example: Triple Town

  • A puzzle game where you place a tile and if there are 3 of a kind, they merge into a single higher tier tile.
  • Issue: Player think the game is matching randomly.

65 of 78

What is the skill tree?

Place Object

Merge 3

Place Adjacent

66 of 78

What skill is the problem?

  • Not placing. People do that.
  • People accidentally 3 merge into 1.
  • But they don’t know why.
  • Look at skill tree
  • Likely something wrong with the Place Same Type Adjacent skill.

67 of 78

What step in the loop is the problem?

  • They are missing the causal link.
  • Usually a feedback issue.

​

68 of 78

Experiment: Feedback

  • Brainstorm ways of clearly showing causation
  • Add an animation of pieces merging together.
    • When you place same pieces adjacent…
    • Start merge
    • Transformation animation

​

69 of 78

Result?

  • People are much better at understanding 3 merge into 1.
  • But they are surprised every time it happens.
  • And they still aren’t planning

​

70 of 78

What if our skill tree is wrong?

  • There’s another implicit skill here.
  • “Seeing which items will match if you were to place the piece in the cell.”

71 of 78

Revise the skill tree!

Place Object

Merge 3

Place Adjacent

Seeing which pieces will merge

Plan optimal move

72 of 78

Revisit the loop

  • We have zero feedback on this new skill!
  • So of course players won’t learn.

73 of 78

Experiment: Feedback

  • As you move your held item across squares, similar item pulse slightly towards it.
  • They are signaling how they will merge before you make the action.
  • Enables planning

74 of 78

Result?

  • Players stop thinking the game is random
  • They can plan before they make a move.
  • They start playing more strategically.

​

75 of 78

Activity�Identifying Problems

76 of 78

  1. Game: Download Alphabear
  2. Play: Until you make a pretty big bear
  3. Skill Tree: Build a rough skill tree to get to the point where you are confused. ��(if you understand everything, you are smarter than 99% of all players) �
  4. What skill did you fail to learn?
  5. What portion of the skill loop is the problem?�(Learning, Decision, Action, Rules, Feedback)
  6. What is your proposed solution?

77 of 78

Review

Each team presents their case study

78 of 78

Questions?

Any questions people might have about skill trees.