Assignment 1: One constant, six values
Step one number in the specimen's rules evenly across six values and show what difficulty did in response.
Due: 23 April 2027 · Weight: 20%
The brief#
Turn one dial evenly. Show that the game did not respond evenly, and say why.
This is the week 5 tutorial done properly. The tutorial let you be rough; here the seeds are fixed, the run count is stated, and your prediction is written down before you measure, where everyone can see how wrong it was. Being wrong is not marked down. Not having predicted anything is.
A strong response finds the cause. “The curve falls off at 0.20” is a measurement. “The curve falls off at 0.20 because past that speed enemies reach the player before the first upgrade arrives” is an explanation, and it is usually one more simulation away.
What the marks look like#
| Band | What the two pages do |
|---|---|
| HD | Find where the curve bends, name what in the rules causes it, and run the one extra simulation that would show the cause is wrong. Report wins and median together, and say where either stopped answering. |
| D to Cr | Six clean values on the same seeds, the bend in the right place, a cause offered and not tested. |
| P | A table with no prediction dated before it, or a story read into a row of noise. |
What a strong and a weak one look like#
Strong, from the course’s own fork. Starting hearts, 1 to 6: wins of 51 go 1, 2, 12, 23, 31, 37. Median survival goes 48.8 s, 85.0 s, 101.6 s, 117.2 s, 120.0 s, 120.0 s. The bend is between 2 and 3 hearts, and from 5 hearts the median has hit the end of the run and says nothing more, so the last two rows are read from wins alone.
Weak, also from the course’s own fork. Bolt damage, 13 to 23: wins go 12, 9, 12, 14, 13, 11, with the conclusion “the game is best balanced at 19”. The same player on a second set of seeds moves by 5 wins. Every value in that row is inside 5 of every other. The honest reading is that this range has no bend in it, and the submission should have widened the range or changed constant.
What you submit#
A fork of the specimen with your stepping script in it, and two pages: the prediction, the table, the plot and the paragraph. No slides.
What it feeds#
Assignment 2 starts from one value on this curve. Choose a constant you are willing to live with until May.
The spec
What has to be in it. The marker goes down this list before reading anything else, and every line is something a classmate could check without asking you.
- one constant from the rules, not reaction time and not the boss's time limit
- your predicted curve, dated before your first run
- wins out of 51 and median survival at six evenly spaced values, from fixed seeds
- a paragraph on where the curve bends and what in the rules causes it
How it is marked
| Criterion | Weight |
|---|---|
| The prediction, written before anything was measured | 20% |
| The measurement: six even values, 51 runs each, the same seeds throughout | 40% |
| Reading the curve, where it bends and why | 40% |