Week 11: Watching Someone Else Play

Everything measured so far has been your own hands on the controls — this week is about what changes when the hands belong to someone else.

3 May 2027

Question#

You’ve spent ten weeks measuring jumps you designed, with expectations you already had going in. What can a measurement taken by someone else, who didn’t design the jump and doesn’t know what you expected, tell you that your own trials can’t?

Prediction (skip if you’d rather just look)#

Before running anything: if you hand someone a preset without telling them what it’s “supposed” to feel like, do you expect their first attempt to look more like your third trial, or your first?

Experiment#

Hand a classmate a preset in the Jump Lab with no explanation beyond the controls. Watch without commenting, coaching, or answering questions about what’s “supposed” to happen — write down only what you actually observe: where they hesitated, what they tried that you wouldn’t have, whether they noticed the metrics at all. Compare your notes to what you expected before they sat down.

Explanation#

A playtest is only useful if it captures what actually happened rather than what you expected to happen. The moment you coach a player toward the outcome you wanted, you’ve replaced their observation with your assumption — which is a more expensive way of measuring your own bias than just writing it down. The discipline is the same reproducibility instinct from week 3, pointed at a harder target: this time the “instrument” is a person, and people vary in ways a deterministic physics engine never will. Recording what you saw, not what you predicted, is what makes the observation worth anything afterward.

Task#

Playtest one preset with a classmate, silently. Write down three concrete observations (not interpretations) of what they did, then one sentence interpreting what those observations suggest about the preset. Keep the observations and the interpretation clearly separated.