Skip to content

The method

Notes are a crutch for missing structure.

Almost everyone who tries to speak without notes tries to do it by remembering harder. That is the wrong problem, and it is why it fails at the first interruption.

Why memorising does not work

A memorised speech is brittle in a specific way: it has one entry point and one path. Lose your place and there is nothing to fall back to, because the structure was never separable from the words. This is why memorisers dread questions. A question is an instruction to leave the path.

Reading from notes fails differently. It works, but it caps you. You cannot look at a room you are reading to, you cannot follow a better thought that arrives mid-sentence, and everyone can tell.

What the good ones are actually doing

Watch someone genuinely fluent without notes and you will notice they return to the same few shapes. Situation, cause, remedy. Their case, the flaw, mine. If we act, if we do not. These are not tricks; they are containers, and the speaker is filling them live from material they already carry.

That separation is the whole thing. Structure is a handful of shapes you can recall under pressure. Substance is a stock of specifics you have already thought about. Hold both and you do not need a script, because you can build one at the speed of speech.

The traditions this draws on

Two recognisable modes sit behind the app's registers. One is the administrator's: short words, hard numbers, plain causation, and a blunt statement of what happens if you are wrong — the style associated with Singapore's Lee Kuan Yew. The other is the diplomat's: historical and geographic framing, analogy that carries real weight, and a calm cadence that uses silence deliberately — closer to how George Yeo speaks. Neither is endorsed by or affiliated with this app; they are simply the clearest illustrations of two ways to be understood.

What the two share matters more than what separates them. One claim, stated early. Evidence you could check. An explicit consequence. Those are the fundamentals, and they are what the rubric scores in every register.

Part one

Eight skeletons

Each is a few beats with a cue attached. Deliberately few: a structure you cannot recall while a room waits is not a structure. Every rep assigns you one, so you drill all eight rather than defaulting to the two you like.

  • PREP

    Point · Reason · Example · Point

    The default. Any question that deserves a straight answer.

  • Problem · Cause · Solution

    What is wrong · why · what you would do

    Anything broken that you are proposing to fix.

  • Two Futures

    The fork · if we do not · if we do · the choice

    When you need to move someone to act.

  • Chronology

    Where we were · are · are going

    Progress reports, anniversaries, anything needing perspective.

  • The Ladder

    Anecdote · principle · implication

    When a small story carries a large idea.

  • Three Pillars

    Thesis on three independent legs

    When one reason is not enough.

  • Steelman · Rebut

    Their best case · the flaw · yours

    Debate, hostile rooms, anywhere you must be seen to be fair.

  • Map · Zoom

    The wide frame · the question · the join

    When context is what makes the answer make sense.

Part two

A bank, on a recall schedule

The bank is your own material: a figure you trust, a story you have told before and know lands, an analogy that makes something click. Ten good items beat a hundred notes, because notes are read and bank items are recalled.

Items are drilled the way vocabulary is drilled. The app shows you the handle and hides the material; you say it out loud from memory before revealing. Get it and the interval roughly doubles. Miss it and it comes back tomorrow. Recognising something on a page is not the same as producing it on your feet, and only one of those is the skill.

Part three

The questions are the exam

Prepared remarks are the easy half. The moment that exposes a speaker is the follow-up: the question aimed at the joint in the argument they were hoping nobody would test.

So every rep ends with two of them, written against what you actually said rather than pulled from a list, and you answer immediately. If you leaned on an assumption you never stated, that is what gets attacked.

Part four

Six things, scored separately

Separating them is the point. A speaker with wonderful material who rambles and a speaker who is beautifully organised but says nothing both score "fine" on a single number, and they need opposite advice.

Thesis
Can your claim be stated in one sentence, and did you state it early?
Spine
Is the skeleton actually visible, with each beat present and in proportion?
Concreteness
Are the claims anchored in named, checkable specifics rather than abstraction?
Economy
Is every sentence carrying weight, or is it hedging and throat-clearing?
Consequence
Does it say what is at stake and what follows, or does it only describe?
Composure
Do you sound in command — deliberate pauses rather than filled ones?

Delivery is handled separately again, and by arithmetic rather than opinion: pace and how much it swings, filler rate, what share of your time was silence, how long the longest silence ran, sentence length, hedges, and how many checkable specifics you produced per minute. Those numbers are then handed to the marker as facts, so composure is graded from evidence instead of guessed at from a transcript.

And then one instruction

However much is wrong with a rep, you get exactly one thing to fix before the next one. A list of six corrections changes nothing. One changes the next rep.

Get early access