Motion, and how long it felt

SLOP2805 · Week 6 · Studio framing

Does how the bar moves change how long the wait feels?

And — the part that transfers — how would you know?

Where we were, and where we are going

Week 5 left you with six instability figures and no way to choose between them. This week adds the other kind of evidence, and shows you its limits from the inside.

By 17:00 you can:

  • build motion treatments that differ only in easing, over provably equal waits
  • run a class ranking and record it without cleaning it up
  • name three specific reasons your own result is weak

What the published work actually measured

Harrison and colleagues at Carnegie Mellon ran forced-choice comparisons between animations of identical real duration, with participants who did not know the durations matched.

  • the measure was perceived duration
  • reported effects sit in the range of a few per cent of perceived time
  • it is not a claim about trust, comprehension, or whether a number is better

Their summary page is this week’s reading. Read the effect sizes, not the headline.

What twenty classmates can reproduce

The procedure. Not the conditions. Five reasons, named now so the write-up is not hindsight:

  1. Not blind — you built the treatments and know which you expect to win
  2. Order effects — whoever sees the third has already waited a minute
  3. Demand characteristics — they know it is a week about animation
  4. No stakes — nobody is waiting for a file they need
  5. n ≈ 20 against an effect of a few per cent

Worked example — the assertion that makes it valid

Three easings, one duration. The difference is invisible by eye, so assert it:

const TREATMENTS = { linear: t => t,
                     decelerate: t => 1 - (1 - t) ** 2,
                     pulsed: t => t + 0.02 * Math.sin(t * 18 * Math.PI) };

// A treatment 400ms longer than another invalidates the whole exercise.
for (const ease of Object.values(TREATMENTS)) {
  console.assert(Math.abs(ease(1) - 1) < 1e-9, "must finish at exactly 1");
}

Easing changes the shape of the journey, never its length. If ease(1) !== 1 you are comparing durations, not motions.

Your turn — the build, then the run

  1. Three treatments — linear, decelerating, pulsed — over identical waits.
  2. The duration assertion, run and passing.
  3. Rotate in groups of five. Nobody is told the durations match.
  4. The raw ranking table, on the board, unadjusted.
  5. Ten minutes, individually: three weaknesses, before anyone interprets it.

The mistake to expect

Dropping a response.

Someone ranks the pulsed bar fastest “because they weren’t paying attention”. The temptation to exclude them is the exercise.

If you were tempted to drop a response, write the temptation down instead. That sentence is worth more than the cleaned table.

Second mistake: writing “small sample size”. Specific beats generic — “n ≈ 20 against a published effect of a few per cent” is a reason; the other is a phrase.

What this week cannot tell you

Almost anything about perceived duration. That is the intended outcome.

If your three came out in the published order, that is not replication — with this design it is close to a coin landing the way you expected.

What you get is calibration: you now know from the inside what a weak result feels like while you are collecting it.

What leaves the room

  • three motion treatments and a passing duration assertion
  • one raw ranking table
  • three specific weaknesses, in your words

Where it goes: the treatments are audited in week 7 for whether they carried meaning or atmosphere. The weaknesses paragraph is the rehearsal for Assignment 3’s Evaluation and revision, 20 marks, which is scored on whether your conclusions are proportionate to your evidence.