Observation: Noticing What Others Walk Past

1. Set a boundary: pick one place and one time box. Example: a 20-minute window at a busy café entrance.
2. Wide pass (7 minutes): no judgments, no ideas. Record only what you can see, hear, count or quote. Use numbers where possible.
3. Narrow pass (7 minutes): pick the two or three most repeated or most surprising details and observe them closely. Time them, count them, follow one person through the space.
4. Note dump (6 minutes): write 10 raw notes. No solutions yet, no "we should build". Just facts, quotes, and open questions.

Two field rules make the notes usable later:

Rule	Why it matters
Separate fact from interpretation	"Four people stepped back from the counter" is data. "People seemed anxious" is a story you invented.
Capture verbs and numbers, not adjectives	"Twice, someone put their bag on the floor to free a hand" survives; "it was crowded" does not.

Keep it ethical and low-risk: observe public shared spaces, watch behaviour and objects, do not record personal details or identify individuals, and never photograph people's faces or screens. You are collecting patterns, not profiles.

Practice 1 (5 minutes). Pick a spot you already visit daily. Set a 10-minute timer and write 8 fact-only notes. Then mark each note F (fact) or I (interpretation). If more than half are I, rewrite the worst three as countable facts.

Note: Guidance: a strong rewrite moves from "the checkout area is confusing" to "three shoppers in ten minutes turned left, then reversed after about two metres." The second version can be tested. The first cannot.

From Field Notes to Candidate Problem Statements

Raw notes are ore, not metal. The conversion step is where the loop turns convergent, and it has a repeatable shape.

flowchart TD
    A["📝 Raw notes<br/>facts, counts, quotes"] --> B["🔁 Cluster<br/>group by repeated friction"]
    B --> C["🧩 Tension<br/>who wants what, and what blocks it"]
    C --> D["❓ Problem statement<br/>one sentence, testable"]
    D --> E["🔬 Quick check<br/>does the evidence support it?"]

A usable problem statement names a specific person, a specific moment, and an observable gap. Compare:

• Weak: "People find queuing annoying."
• Strong: "First-time visitors at the café entrance cannot tell where the queue starts, so they hover in the doorway and get redirected by staff."

The strong version can already be tested with a five-minute count, which is exactly what you want heading into the capstone. Three starters that keep statements grounded:

1. "When [person] tries to [goal] at [moment], they end up [workaround] because [obstacle]."
2. "Nobody has designed for the moment when [specific awkward thing] happens."
3. "The official process says X, but everyone actually does Y."

Practice 2 (6 minutes). Take your eight notes from Practice 1. Cluster them into at most three themes, then write one problem statement per theme using starter 1. Finally, write the single observation that would prove your statement wrong. If you cannot name that observation, sharpen the statement until you can.

Important: The falsification line is the deliverable here. A problem statement you cannot disprove is a mood, not a finding.

Traps That Flatten Your Data

Field Notes That Survive Contact With Reality

• Do not solve during the sweep. The moment you start designing, you stop seeing. Carry a separate "ideas" column and leave it empty until the note dump.
• Chase the second surprise. The first odd thing you notice is usually the thing everyone else already noticed. Follow it and ask what it hides.
• Count something. One number beats ten adjectives because it can be compared across visits.
• Write notes in the moment, not from memory. Recall quietly rewrites detail into a tidier, less useful story.
• Visit twice. A single sweep captures a moment. Two sweeps at different times separate a pattern from an accident.
• Common mistake: collecting opinions instead of behaviour. Asking people what they want produces polite fiction. Watching what they do produces evidence.
• Rule of thumb: if a note cannot be photographed, counted, timed or quoted, it is probably interpretation wearing a fact's coat.

What You Now Have

You can run a bounded sweep, separate fact from interpretation, cluster notes into tensions, and phrase a testable problem statement with a built-in way to be wrong. That is raw material plus a first convergent pass. Quick check before you move on: an observation that cannot be disproved is _____, and every workaround you spot is _____. A useful problem statement names _____.

Next lesson takes these problem statements and adds a lever: how deliberate limits sharpen a vague idea into something you can actually build and test.

Front: What separates observing from looking? Back: Observing is active and specific (counts, quotes, behaviour); looking is passive confirmation of what you expect.

Front: Why are workarounds valuable observations? Back: They reveal problems people have already solved badly, which makes them worth solving well.

Front: What are the four moves of a 20-minute observation sweep? Back: Set a boundary, wide pass, narrow pass, note dump.

Front: What happens in the wide pass? Back: About seven minutes of recording only what you see, hear, count or quote, with no judgments or ideas.

Front: What happens in the narrow pass? Back: You pick the two or three most repeated or surprising details and observe them closely with counts and timing.

Front: How should fact and interpretation be handled in field notes? Back: Keep them in separate columns; a fact is countable or quotable, an interpretation is a story you added.

Front: Why capture verbs and numbers instead of adjectives? Back: Countable, specific notes survive into analysis; adjectives like crowded collapse into vague impressions.

Front: What three elements make a problem statement testable? Back: A specific person, a specific moment, and an observable gap.

Front: What should accompany every problem statement? Back: The observation that would prove it wrong; if you cannot name one, it is not yet a finding.

Front: Why avoid solving during the observation sweep? Back: Designing shuts down seeing, so ideas go in a separate column and stay empty until the note dump.

Front: Why visit the same place twice? Back: A single sweep captures one moment; a second visit separates a pattern from an accident.

Front: Why is asking people what they want weaker than watching what they do? Back: Stated preferences produce polite fiction; behaviour produces evidence.

Front: What is the rule of thumb for a note that cannot be counted, timed, quoted or photographed? Back: It is probably interpretation dressed up as a fact.

Front: What are the basic ethics of a public-space sweep? Back: Observe shared behaviour and objects in public, never identify individuals or capture faces and screens.