Put every fragment on one page in its rough form, label each one as a claim, a question or a reaction, and then work on a single claim rather than sorting the whole set. The sorting that feels like the job can wait, because what you need first is one sentence you could be wrong about, and the other fragments stay in view until it exists.
That order matters because scattered thoughts usually arrive in a form that resists sorting. A voice note from a walk, three lines typed at midnight and a comment someone made in a meeting do not share a shape, so grouping them asks for a view of the whole idea that you do not have yet. The method below reaches that view by a different route, and every example in it is invented to show the steps rather than to report on anyone.
Why tidying the notes first tends to stall
Three first moves are common when thoughts feel scattered, and each one stalls for a reason worth naming. Sorting the fragments into categories comes first for most people, and it fails because the categories depend on knowing what the idea is about, which is the thing you are trying to find out. Polishing the wording of the fragment that reads best is the second move, and it produces a smooth sentence whose point you still cannot state. Waiting for the idea to settle on its own is the third. It treats a lack of structure as a lack of time. The table below sets each move beside where it stalls and the move I would make instead, and those readings come from reasoning about how the moves behave rather than from any measurement.
| Common first move | Where it tends to stall | The move that replaces it |
|---|---|---|
| Sort every fragment into named categories | The categories need a view of the whole idea that you do not have yet | Label each fragment by what it can do, as a claim, a question or a reaction |
| Polish the wording of the clearest fragment | A smooth sentence can still have no point that you could be wrong about | Compress one claim to a single sentence and write down what the compression dropped |
| Wait for the idea to settle before writing anything | A lack of structure gets treated as a lack of time, and the notes keep growing | Spend ten minutes capturing everything, then pick one claim to start from |
Three common first moves on scattered thoughts, where each one tends to stall and the move that replaces it.
Put every fragment on one page before you judge any of it
The first step is a capture pass of about ten minutes in which you get every fragment onto one page without editing. Speak the ones that live in your head and type the ones that live in other places, and copy in the scraps from messages, margins and earlier notes. A fragment belongs on the page if it has crossed your mind about this subject in the last few weeks, even when it looks like a complaint or a half-formed worry. Editing at this stage removes the material you will need in the next step, because the awkward fragments are often the ones carrying the uncertainty that matters.
To make the steps concrete, imagine a designer who has been meaning to redo their portfolio site for a year. Their capture pass produces fourteen fragments: a voice note saying the current site hides the client work that pays, a line about wanting fewer case studies, a worry that nobody reads case studies anyway, a note that two recent enquiries came through a single talk rather than the site, and ten more in a similar state. Nothing about that list is organised. That is the correct condition for the page at the end of a capture pass.
Label each fragment as a claim, a question or a reaction
The second step is the only sorting the method asks for, and it sorts by what a fragment can do rather than by what it is about. A claim is a statement that could turn out to be false, such as the line that nobody reads case studies. A question is a gap you could go and fill, such as which of the last ten enquiries mentioned the site at all. A reaction records how you feel about the project, such as the wish for fewer case studies, and it carries information about you without carrying anything you could check. A repeat restates an earlier fragment in different words, and marking it as one shows you which points your mind keeps returning to.
| Kind of fragment | Count of fourteen | An invented example from the capture pass |
|---|---|---|
| Claims | 4 | Nobody reads case studies before they get in touch |
| Questions | 5 | How many of the last ten enquiries mentioned the site at all |
| Repeats | 3 | The site hides the client work that pays, said three ways |
| Reactions | 2 | I want far fewer case studies on the new site |
The invented counts behind the chart, with one example fragment of each kind from the portfolio redesign.
In the invented set, four of the fourteen fragments were claims and five were questions, so nine of them could do something once labelled. Three were repeats of the point about hidden client work, which tells the designer where their attention keeps landing, and two were reactions that record a preference without offering anything to check. The labelling pass on fourteen fragments takes a few minutes. The picture it leaves differs from the one the unsorted page gave, where every fragment looked equally urgent and equally vague.
Start from one claim and write down what you leave behind
The third step is to choose one claim and compress it to a single sentence that could be wrong. Pick the claim whose truth would change the most of the other fragments, since that is the one whose test pays for itself. For the designer, the claim about nobody reading case studies decides whether fewer case studies would help, whether the hidden client work matters and whether the talk outperformed the site, so it is the one to start from. Written as a sentence that could be wrong, it becomes a statement that the people who got in touch in the last year did not read a case study first, and that sentence points at ten people who could be asked.
Compression always discards some of the material. The discarded part is where the uncertainty worth keeping lives. Alongside the sentence, write a short record of what the compression dropped, such as the suspicion that case studies matter for larger clients even if they did not matter for the last ten, and the question of whether the talk was a one-off. That record is the reason the method does not flatten the idea into something tidier than you believe. A sentence you could be wrong about, with its leftovers listed beside it, is a starting point you can act on without pretending to be more certain than you are. If you want to take that sentence further, the earlier note on turning a vague idea into something you can test covers the step from a claim to a small check with a stopping result.
Keep the other fragments where you can see them
The fragments you did not start from go on a dated list that you keep visible, and this is what separates the method from choosing a favourite and forgetting the rest. Questions on the list are candidates for the next starting point once the first claim has been checked. Reactions stay as a record of what you wanted before the evidence arrived. They are worth rereading later to see whether the evidence moved you. Repeats can be struck through, with a note of what they repeated, because they have already done their work by showing where your attention gathers.
The list has a return rule, which is that you come back to it only when the current claim has been checked or dropped. Without the rule, the list becomes a second scattered page. The cycle then begins again from a larger pile. With it, the list behaves like a queue that your own results advance, and a fragment that still looks important after three returns has earned a turn as the starting point. One that looks smaller every time can be dropped with a date beside it, which turns a nagging note into a decision you made.
What to do tonight and what remains unknown
Tonight, open one blank page and spend ten minutes putting every fragment about the subject on it, spoken or typed, without correcting any of them. Then mark each fragment with a single letter for claim, question, reaction or repeat, choose the claim whose truth would change the most of the others, and write it as one sentence that could be wrong with its leftovers listed underneath. Put the remaining fragments on a dated list and close the page for the night, so that tomorrow begins with one sentence to check rather than a page to tidy.
Several things remain unknown, including whether the four labels suit every kind of project. I have no measurement of how often a capture pass produces a claim worth starting from, and the fourteen fragments above are invented, so their proportions show the method rather than a typical case. Some sets of thoughts may yield only questions and reactions, in which case the starting point is the question whose answer would create the first claim. Your own labelled page is the only evidence that bears on your own thoughts, and it may show that the scatter was hiding one claim you have been avoiding.