Sequencing

Sequencing

Almost everything with more than three parts has an order, and almost nobody decides it. The order arrives by accident: the sequence the material was made in, the sequence it was remembered in, or the sequence a template imposed.

Music, comedy, and live performance solved this a long time ago because the feedback is immediate. A set that peaks too early dies in the room and everyone watching knows it. That produced a working craft for arranging things, and it transfers to any sequence a person moves through in time.

Use the logic. Do not perform the frame. Never talk about DJ sets in the output unless the user asks where this comes from.

What this does and does not do

It works with the material that exists. It reorders, it cuts, it flags gaps.

It never adds material. If the sequence is missing something structural, name the gap and stop. Writing the missing piece is a different job, and a skill that invents content while claiming to fix order will do both badly.

The two mechanisms

Everything in this skill runs on these.

1. Every item needs a job.

In a set, no slot is neutral. Each one opens, builds, peaks, rests, plants, pays off, lands, or closes. An item with no job is not a small problem, it is the problem. Assign a job to every item. Anything left over is a cut candidate, and usually the reason the thing drags.

2. The swap test.

For every adjacent pair, ask whether swapping them loses anything. If the answer is no, the order is not doing work at that seam.

A sequence where most pairs swap freely is not a sequence, it is a list in an arbitrary order, and the audience feels that even when they cannot name it. Report the swappable seams. They are where the reordering happens.

Step 1: Captive or not

Ask this before anything else, because getting it wrong makes every later decision wrong.

Captive. The audience cannot easily leave. A meeting, a live talk, a class, a wedding reception, a demo you are running. You can build. A slower open is affordable and often better, because tension has somewhere to go.

Non-captive. They can leave at any second and it costs them nothing. A deck emailed around, a page, a video, a document, a playlist. You are being abandoned continuously, so the strongest material moves forward and every item has to buy the next one.

The same material has two different correct orders depending on this answer. A pitch deck presented in a room and the same deck sent as a PDF are not the same artifact.

If it is both, sequence for non-captive. It survives the room; the reverse is not true.

Step 2: Map dependencies

Some items cannot move. If B only makes sense after A, that is a constraint, not a preference.

references/shapes.md has the standard running order for each common type, so a deviation reads as a choice rather than an accident. Check the relevant shape before assuming a dependency is real.

List the hard dependencies first, then treat everything else as free. Most sequences have far fewer real dependencies than people assume, and the assumed ones are usually habit.

Dependency beats energy, always. A brilliantly placed payoff that lands before the thing it depends on is not a bold choice, it is a broken sequence.

Step 3: Assign jobs

Give every item a job. Read references/moves.md for what each one requires.

• Open - buys the next item. Does not inform, does not set up. Earns attention
• Build - raises the stakes or deepens the case
• Peak - the big moment. Few of these, and never two adjacent
• Rest - lowers energy on purpose so the next peak has somewhere to rise from
• Plant - places something that pays off later. Must look innocuous now
• Payoff - the callback. The highest-value move available
• Land - the ask, the decision, the thing you need them to do
• Close - the last thing, which is what gets remembered

Items with no job go on the cut list. Do not soften this by inventing a job for them.

Step 4: Draw the energy curve

Plot the sequence and look at the shape. Read references/energy.md.

The failures are visible once drawn:

• Flat. No peaks. Everything the same weight, which reads as long regardless of actual length
• Front-loaded and dropped. Best thing first, then descending. Common in decks built by enthusiasm
• Double peak. Two big moments adjacent. The second one always reads smaller
• No rest. Every item is important, so nothing is
• Peak at the end with nothing before it. A long climb the audience left before finishing

Also check contrast. Two adjacent items in the same mode kill each other, whatever their quality. Two case studies, two charts, two stories with the same shape. Change the mode, not only the topic.

Step 5: Run the swap test

Walk every adjacent pair. Mark each seam as locked (swapping loses something, say what) or free (swapping loses nothing).

Report the free seams honestly, including how many there are. A sequence that is mostly free seams needs a real reorder, not a tweak.

Step 6: Find the walkout

The point where the audience leaves. Every sequence has one and it is findable.

The item they leave on is usually not the culprit. The culprit is the item before it, which failed to buy the next one. Look one step back from where attention dies.

Read references/attention.md. In the walkout mode this step is the whole job.

Step 7: Reorder, and say what changed

Deliver the new order with a one line reason per move. Not a general explanation, a reason per move, because that is what teaches the user to do it themselves next time.

Then state what you cut and why, and what gap you found and did not fill.

Modes

Reorder - the default. Items exist, the order is wrong. All steps.

Build - material exists, no order yet. Steps 1 through 4, then construct.

The walkout - something is losing people and the user does not know where. Steps 1, 5, 6. Output is the marked-up current order with the exact seam where it dies, then the fix. Use assets/walkout.md.

The cut - too much material. Steps 1 through 3, then cut by job. Anything without a job goes first, then the weakest of any duplicated job.

Output format

Deliver in this shape unless the user asks for something else. Prose, no long preamble.

THE READ
Captive or not, and the shape the current order makes. Two or three sentences.

THE WALKOUT
Where it loses them, and the item responsible. Name the seam.

THE NEW ORDER
Numbered. Each item with its job, and a one line reason for anything that moved.

CUT
What goes, and why. Brief.

THE SEAMS
Which adjacent pairs are locked and which are free. Free seams named.

THE GAP
What the sequence needs that does not exist yet. Named, not written.

The worksheet is assets/running-order.md.

Rules of engagement

Never add material. Name gaps, do not fill them.

Never manufacture a reorder. If the order is already right, say it is right and say why. A skill that always finds a fix is not reading, it is performing.

Dependencies beat energy. Every time, without exception.

Cut on job, not on quality. The best item in the room still goes if it has no job, and a mediocre item that provides the only rest in the sequence stays.

Respect what the sequence is for. A funeral service, a legal presentation, a safety briefing, and a curriculum have obligations that outrank engagement. Some orders exist for reasons that are not about attention, and rearranging them for energy is malpractice. Ask what the sequence owes before deciding what it should do.

The last thing is what gets remembered, so decide it on purpose. Ending on Q&A means ending on whatever the last question happened to be, which is a decision to let someone else close for you.