Dither

Podcast 2 — In conversation

Speaker A: So I was watching my project manager yesterday spend literally ten minutes deciding whether to use blue or teal for the header colour. Just completely dithering. In a meeting. With the client sitting there.

Speaker B: That is painfully relatable. The classic dither. There is nothing quite like watching someone dither in real time when a decision simply needs to be made.

Speaker A: Exactly. And the word dither is just so accurate for it, isn't it? It is not angry indecision — it is flustered, uncertain, kind of endearing but also maddening.

Speaker B: Right. You could say: 'Please stop dithering and just pick a colour.' That sentence would not work with hesitate. Hesitate is too neutral. Dither has that edge of mild exasperation built in.

Speaker A: What about the noun form? I sometimes hear people say 'he was in a right dither'. Is that common?

Speaker B: Very common in British English, yes. 'In a dither' means in a state of agitated confusion. You could say: 'She arrived at the presentation in a complete dither because she had forgotten her slides.' It describes flustered panic as much as indecision.

Speaker A: So there is a slight difference between dithering — the act of not deciding — and being in a dither, which is more about general anxiety?

Speaker B: Exactly. The verb form focuses on the indecision itself, while the noun phrase captures a broader sense of fluster. Both are informal, both are distinctly British.

Speaker A: What about common mistakes with the word? People sometimes confuse dither with dawdle, I think.

Speaker B: Good catch. Dawdle is about moving slowly or taking too long over an action. Dithering is specifically about being unable to choose between options. You can dawdle on your way to a decision, but you dither at the decision point itself.

Speaker A: So if someone is spending ages walking to the coffee machine, they are dawdling. If they are standing in front of it unable to choose between latte and flat white, they are dithering?

Speaker B: Perfect. You have got it. Dither is the word for that exact moment of frozen choice. Precise, warm, and wonderfully British.

Podcast 3 — Prompt engineering

INST: Today we are looking at dither in the context of developer prompts. The word dither is particularly useful in prompts because it describes a specific behavioural pattern — indecision between states — which translates directly into UI states, conditional logic, and user experience design.

STU: That is an interesting angle. I had not thought of a vocabulary word as a logic concept before.

INST: Exactly. Let us start with a UI scenario. Suppose you are building a decision workflow and you want to prevent users from dithering indefinitely:

Prompt: Design a React modal component called DecisionPrompt that prevents user dithering by enforcing a 30-second countdown timer. Display the remaining time prominently, auto-select the default option when time expires, and highlight the recommended choice with a pulsing green border. Include cancel and confirm buttons.

INST: The word dithering in the prompt tells the AI this component solves a specific UX problem: users who delay too long. The output will focus on time pressure, default fallbacks, and visual hierarchy — all because one word set the context.

STU: So dithering becomes the design brief in one word. What about a database scenario?

INST: Here is a schema prompt for tracking decision delays:

Prompt: Design a PostgreSQL schema for a workflow approval system that logs dithering behaviour. Include a decisions table with columns: decision_id, user_id, created_at, first_viewed_at, submitted_at, and a generated column dither_seconds as the interval in seconds between first_viewed_at and submitted_at. Add an index on dither_seconds for analytics queries.

INST: The generated column dither_seconds is named after the concept. Any analyst querying this table immediately understands what the column measures. The word dither makes the schema self-documenting without needing a comment.

STU: That is a really clever naming convention. What about an API endpoint?

INST: Here is one for a Node.js backend:

Prompt: Build a Node.js Express middleware called ditherGuard that intercepts form submissions and checks whether the user took more than 120 seconds to submit after first opening the form. If so, log the event to a dither_log table in PostgreSQL with user_id, form_name, dither_duration, and timestamp. Return the submission normally regardless.

INST: The middleware name ditherGuard communicates its purpose immediately. Any developer reading the codebase knows this middleware is specifically tracking hesitation behaviour, not general latency or errors.

STU: What about an application that helps users stop dithering themselves?

INST: Excellent thinking. Here is a full application prompt:

Prompt: Build a full-stack web application called NoDither using React and Node.js. The app presents users with one decision per day — a question with two or three options and a 60-second countdown. Users cannot skip or revisit. Track all decisions in PostgreSQL with a dither_score per user, calculated as average decision time across the last 30 sessions. Display a weekly dither trend chart on the dashboard.

INST: Every feature in this application flows from the word dither. The timer, the no-skip rule, the scoring system, the trend chart — all of it is a direct translation of the concept into product design.

STU: And for an analytics tool?

INST: Try this Python prompt:

Prompt: Write a Python script called dither_analyser.py that reads a CSV of user session events with columns event_type, user_id, and timestamp. Identify dithering sessions where a user triggered a decision_viewed event but took more than 90 seconds to trigger decision_submitted. Output a summary with total_sessions, dithering_sessions, dither_rate_percent, and top_10_dithering_users sorted by average dither time.

INST: The column names, function names, and output keys all carry the word dither. This consistency makes the script immediately understandable and ensures any follow-up prompt from a teammate will produce compatible naming.

STU: I see how one word, used consistently, can anchor an entire codebase around a concept.

INST: That is the core lesson. In prompt engineering, precise vocabulary is not decoration — it is specification. Dither tells the AI exactly what behaviour, what logic branch, and what naming convention to use throughout the entire output.