Skip to main content

SLOP2875: Is It Me or Is It the Game?

Measuring game balance without players

SLOP2875 · Semester 1 2027 · one game · nobody plays until week ten

You died to the boss again. Either you are too slow or the boss is too big, and arguing will not settle it. This course finds out by simulation, in one two-minute game, Close Quarters: build the instrument, simulate the players, read the distribution, defend one number. Nobody plays until week ten.

You died to the boss again. Either you are too slow or the boss is too big, and arguing will not settle it. This course finds out by simulation, in one two-minute game, Close Quarters: build the instrument, simulate the players, read the distribution, defend one number. Nobody plays until week ten.

01 · the dial

Start with “is it me”

Take one simulated player and make them slower, 50 ms at a time. Each row below is 51 runs at that reaction time, and a filled dot is a run that beat the game. Your reaction time is the dial. If the game answered it evenly, the rows would shrink evenly. They do not.

Nobody played any of these. Week 5 is about why even steps are not even.

02 · the specimen

What you will do, and who it is for

Twelve weeks measuring one small game.

Specimen
Close Quarters, a two-minute survival run with seven upgrade cards and two bosses. It is live, and linked here on purpose.
Its rules
Eight hundred lines of TypeScript, and those you may read now.
You open it
Week ten. You are asked not to before then.
You leave with
One number: a change to one constant in that game, and a ten-minute case for it that three sceptical classmates could check.
It is for
People who make or mod games and are tired of balance arguments nobody can lose.
You need
To read and edit TypeScript.
You do not need
Statistics beyond a median, or to be good at the game.

03 · the semester

The semester, week by week

Tuesday is the claim, Thursday is the measurement.

WeekLecture, TuesdayTutorial, ThursdayDue, and what to read
123 February 2027A complaint about balance is a predictionslidesThree predictionsRead Mark J. Nelson, Game Metrics Without Players: Strategies for Understanding Game Artifacts
22 March 2027A game that needs a screen cannot be measuredslidesOne headless run
39 March 2027A perfect player describes nobodyslidesA player with reaction timeRead Aaron Isaksen, Exploring Game Space Using Survival Analysis
416 March 2027One run tells you nothingslidesHow many runs
523 March 2027Even steps, uneven difficultyslidesStep one number
630 March 2027A dead option looks like a live oneslidesTake one awayRead Alexander Jaffe, Evaluating Competitive Game Balance with Restricted Play
Teaching break
720 April 2027A gate everyone passes asks no questionslidesSize a gateAssignment 1: One constant, six values due 23 April 2027, 20%
827 April 2027The lazy strategy must loseslidesBreak the game
94 May 2027A metric that cannot fail is noiseslidesFind the ceilingAssignment 2: Nothing broke due 7 May 2027, 25%Read Stefan Freyr Gudmundsson, Human-Like Playtesting with Deep Learning
1011 May 2027Playing finds what simulation never askedslidesPlay itYou play the game for the first time.
1118 May 2027Every boss is easy for somebodyslidesThree kinds of playerRead Shaghayegh Roohi, Predicting Game Difficulty and Churn Without Players
1225 May 2027A change is finished when it can be checkedslidesDefence rehearsalTutorial quizzes due 27 May 2027, 20%Final: Defend a number due 28 May 2027, 35%