Article Structure Denotes the Order of an Argument, Not an Arrangement of Items

Run the procedure for article structure through to the end, and what is left in hand is a list of items. Settle the keyword, check the headings of the top ten results, work out the items required, fill in the fields of the template. The output is a table with the items to be covered laid out in it.

A list is not a structure. A structure is the principle settling the order in which items get arranged, and the list itself contains no grounds for settling that order.

Judgement is therefore required at the stage of rearranging the items you collected. The basis for that judgement, however, appears nowhere in the procedure. Unwritten, and with nothing said beyond refer to the top results, the order of the top results is what generally gets adopted.

There is one principle settling the order. The single claim the piece sets out to make, cut into the steps a reader can take in. Not the importance of the items, and not the frequency of the searches.

Set out first the questions this piece takes on. First, why the work of collecting items from the top ten results does not output a structure. Second, what becomes adjudicable once a heading changes from a category name into a question. Third, what settles the length of an article, given that it is not the size of the subject.

The questions it cannot take on go here too. How many headings to have, how many characters to write under each one, how far to fill in premises — no distribution table comes with this. These move with the number of steps in the claim being handled and with the gap in premises between you and the reader.

Here is where the damage sits. With no grounds for the order on record, the order cannot be verified afterwards. Discovering that a piece is not being read leaves you unable to separate order from content as the cause. Unable to separate them, what gets touched next is expression. Repair the expression and it still goes unread, so you repair the expression again.

And where a template has been supplied, this state becomes harder to see. Fields to fill in produce a sense of completion at the moment they are filled. What the fields ask about is the presence of items, not the grounds for the order.

A note before going on: this is not an argument that the work of collecting items is wasted. What the reader expects is available nowhere else. What gets handled here is what the stage of arranging the collected items requires.

This state persists for one further reason. An article whose order is wrong does not become an unreadable article. The items are all present, so reading to the end delivers the content. The result is a piece that reads but leaves no gist, so order never comes up as a candidate cause.

The Procedure for Structure Begins From Checking the Top Ten

The procedure shared as the way to build article structure has settled into more or less a single form.

Article structure is the blueprint of an article. The assumed reader, the trouble that reader carries, the sequence of headings, what you wish to convey. These are described as work to be settled before the body gets written.

Two reasons get raised for building a structure. First, that quality gets settled by the structure. Second, that a structure makes the work of writing the body faster. Write improvisationally and you hesitate partway and the discussion strays — that is the observation.

The procedure gets set out concretely. Settle the keyword. Settle the assumed reader. Search on that keyword and check the titles and headings of the top ten results. Work out the items required. Assemble the headings.

Templates of structure get set out as well. The representative form places the conclusion first, states reasons, shows concretes, and returns to the conclusion. Readers arrive seeking an answer, so do not place a long preamble — that account gets appended.

A rule of thumb on working time appears too. Finish it in around an hour at most. Do not build it out in excessive detail — that is the advice.

A template for article structure is sometimes supplied. It takes the form of filling in the hierarchy of headings, the content to write under each, and the distribution of characters.

Every part of the procedure is on target. Write without a structure and you hesitate; look at the top results and what readers expect becomes available. Placing the conclusion first is a correct decision for that medium. If an item the reader takes for granted is missing, the article feels incomplete. The work of filling gaps is genuinely necessary.

One shared way of seeing runs underneath the procedure. Structure is organised as the work of collecting items and arranging them. With every required item present, the structure holds. A problem of coverage.

Organised entirely as a matter of coverage, the structure is complete at the moment the items are assembled. The grounds for the order they were arranged in do not go on record.

The same assumption settles the direction of improvement when something did not work. Read it as insufficient coverage and you add items. Where adding items still leaves it unread, what gets suspected next is expression. Order, as an option, was never on record, so it does not even enter the field of candidates.

The rule of thumb on working time works in the same direction. An hour is a reasonable duration for the work of collecting items and arranging them. It does not fit the work of settling the grounds for each ordering decision. What the rule of thumb measures is the work the procedure assumes.

Collecting the Headings of Top-Ranking Articles Does Not Produce a Structure

At the centre of the procedure sits the check on the headings of top-ranking articles. This work has a part that is effective and a part that is not.

What is effective is learning the items the reader expects. Which questions readers arriving on which keyword carry. The top-ranking articles show the current standard answer to those questions. That is information worth confirming.

What is not effective is deriving a structure from it. Collect the headings of ten top results and you obtain a set of items. That set contains no principle for settling the order.

What settles the order is the claim each writer held. The order required differs by claim. Extract only the items from ten articles and mix them, and you get a heap of items belonging to no claim at all.

Rearranging this heap requires grounds. With no grounds available, the order of the top-ranking article is what generally gets adopted.

A view arises that the order of the top results, being the order actually read, must be correct. It is the correct order for that article, because it was assembled to support that article’s claim. Transplant that order into an article holding a different claim and you have an order with nothing to support.

Furthermore, everybody following the same procedure makes structures converge. Once they converge, the difference between articles survives only on the side of expression. Among articles handling the same items in the same order, the only reason a reader has for choosing one is skill of writing.

There are settings where convergence is not a problem. With explanations of procedures or specifications, where the content itself is fixed, following the standard order reads better. What separates the cases is writing that contains a claim.

From the top results, take the items the reader expects and the answer the top results hold in common. Leave the structure and the phrasing. The former is information about the reader; the latter is judgement belonging to the writer. Looking at the same page, what you carry away changes the nature of the work.

Look at the top results first and your own claim gets drawn towards theirs, because organising your own thinking after reading mixes in what you read. As a sequence, writing your own claim in one line before looking at the top results reduces the contamination.

Write that line first and one possible outcome is finding that the top results and your answer are the same. That is not a loss either. If they are the same, the reason for writing that article has just disappeared, which beats noticing after it is written.

A Structure Is the Order of an Argument, Not an Arrangement of Items

A structure is a single claim decomposed into the order in which a reader can receive it.

The skeleton of an argument has a long-standing organisation. Toulmin, in 1958, divided argument into the claim, the grounds, and the premise carrying the grounds to the claim (Toulmin, The Uses of Argument, Cambridge University Press, 1958). Elements making the reservation and the room for rebuttal explicit get placed alongside them.

There is an organisation from the reader’s side as well. Kintsch and van Dijk, in 1978, divided the comprehension of text into two layers: the processing of individual sentences, and the processing that constructs the gist of the whole (Kintsch & van Dijk, 1978, Psychological Review, 85(5), 363–394). A reader receives sentences in sequence while assembling the gist of the whole at the same time.

This model presents a computational model accounting for the processes of comprehension and production; it is not a report measuring the effect of article structure in practical settings. What can be taken from it is the direction in which the assembly of the gist proceeds while reading.

Take that direction and what order is doing becomes visible. Whether the next step loads onto the gist the reader is midway through assembling. A step that does not load gets placed outside the gist. A step so placed does not enter the gist even when it is read.

That is where Toulmin’s organisation works on structure too. Place the claim first, show the grounds, disclose the premise carrying the grounds to the claim, and limit the cases where it does not apply. You are not arranging items; you are positioning the parts supporting one claim in the order in which they load.

And this positioning admits an adjudication of excess and deficiency. With the claim fixed at one, the number of grounds required is determined. Too few and the claim goes unsupported; too many and information unrelated to the claim mixes in.

Suppose you raise the claim there is no correct number of headings. What you require is its grounds and the premise carrying the grounds to the claim. The item how to write headings is related, and it does not support this claim. Relatedness is not a criterion for necessity.

Whether readers come to read arguments at all is a fair doubt. They do not come to read the form of an argument. What the reader wants is a judgement usable in their own situation. Handing over a judgement requires why it turns out that way. The sequence that hands over why is an argument.

With explanations of procedure, the order of the work takes priority over the argument. Where what to do in which order is the content itself, decomposing a claim becomes nothing more than a preamble. What separates the cases is whether the article hands over a judgement or hands over a task.

One more thing: this organisation contains a slot for handling reservation — a slot for placing the cases where the claim does not apply as part of the argument. A reservation is not a confession of exceptions but a specification of the claim’s range of application. A claim with no range specified leaves the reader unable to adjudicate whether it applies to them. What cannot be adjudicated does not get carried away.

Assemble in the order of the argument and items you cannot handle emerge — items the reader expects that do not support the claim. Rather than cutting them, touch on them in a paragraph inside a heading, or send them to another article. Put non-supporting items into the main line and the outline of the claim disappears.

A Heading Is Not an Item in a Contents List but a Question to Be Answered

A heading is not a category name for content; it is the question the section answers.

How to write headings is a category name. What settles the number of headings is a question.

Put it in the form of a question and whether the section has finished becomes adjudicable. Answer the question and it is over. With a category name, how much to write before it suffices is not determined.

And put it in the form of a question and the answer fixes at one. Under a category name, several answers can be laid out. Being able to lay them out means it has not been settled what the section claims.

Under the heading how to write emails, the subject line, the body, and the timing can all be put in. Under what settles whether it gets read sits before the subject line, only the answer to that question goes in.

A worry arises that headings all in question form read poorly. The displayed form need not be a question. Whether it gets read is settled before the subject line works as a declarative too. What is required is that the writer has settled, on their own side, the one question that heading answers. The display may take the form of the answer or of the question.

Some sections cannot be decomposed into a question: sections placing a definition, sections sharing a premise. These are not answering a question but building the state in which a question can be raised. They are few in number, and in most cases there are one or two in the earlier part of an article.

The question form also works on the estimate of quantity. The number of steps required to answer one question is roughly visible at the moment the question is raised. A question needing only one step does not require an independent heading.

Conversely, a question needing five or six steps has divided into two questions. The number of headings gets settled as the outcome of that decomposition. It is not a sequence in which the number gets settled first and content gets allocated into it. Enter from allocation and you end up adding related information to satisfy a quantity.

Put it in the form of a question and headings you cannot answer get found — the state in which the question stands and the answer is not on your side. Here, that heading cannot be written. Leave it as a category name and a section with no answer can be filled with related information.

The moment of finding differs as well. Assemble with category names and you notice the absence of an answer partway through writing the body. Assemble in question form and it appears at the structure stage. How much gets rewritten is settled by when you noticed.

Begin From the Claim, Not From an Announcement

The opening this article will explain such and such contains no information.

The reader entered having seen the article’s title. What it is about is something they already know. Repeating that at the opening is a repetition of known information.

On top of which, an announcement contains no claim. Will explain is a forecast that something is about to be said; it does not say anything. What the reader gains in the first few lines comes to zero.

Begin from the claim and material for judgement gets handed over in the first few lines. There is no correct number of headings. What settles it is the number of steps in the claim being handled. With those two sentences, the reader knows what the article will say and can decide whether to read on.

Readers who do not read on emerge. That is not a loss. Had you begun from an announcement, those readers would still have left, only a few paragraphs later.

A worry arises that leading with the conclusion removes the reason to read what follows. The conclusion alone is unusable. Knowing there is no correct number of headings does not settle how many headings your own article should have. Settling it requires why there is no correct number and what settles it instead. That is the body.

The reason to read disappears where the content is complete in the conclusion alone. There, the quantity of the article was itself excessive.

Where the reader does not recognise the problem, sharing that recognition comes first. Readers arriving by search recognise the problem; readers arriving by other routes exist too. The opening for them becomes a description of symptoms rather than an announcement.

An announcement has one further function: lowering the writer’s load at the opening. Will explain can be written with nothing settled about what to write.

An article beginning from an announcement therefore probably had no settled claim at the moment writing began. The first sentence reflects the writer’s state of preparation.

Begin from the claim and a weak claim gets exposed at the opening. Begin from an announcement and a weak claim hides inside the body. The exposure is not a fault; it means something you should have noticed before writing becomes available at the moment writing begins.

And the place of exposure being the opening carries an advantage. The opening is a single paragraph, so the cost of rewriting is nearly zero. Notice the same thing after writing eight sections and the cost is eight sections.

A Structure That Will Not Settle Usually Has Two Claims Inside It

The state the structure will not come together is not short of items. Two or more claims are inside it.

With two claims in one article, the order does not settle, because whichever goes first leaves the other’s grounds hanging.

Put an instance beside it. Structure is the order of an argument and a heading is a question are related, and they are different claims. Raise both as claims and the article becomes two articles advancing alternately.

There are two ways to resolve it: raise one as the claim and lower the other into the position of its grounds, or split the article. Lowering settles the order. Splitting settles it too. Only the option of leaving both standing as claims keeps the unsettled state going.

The adjudication is easy. Write the article’s content in one sentence, and if and or also enters that sentence, there are two or more claims.

A view arises that an article handling several points is the more useful one. The quantity of information rises. The quantity received does not. What a reader can carry away is, in most cases, one thing. Write two and what gets carried away is one of the two, or an impression that is neither.

Several points are sometimes required to support a claim. That is different from having several claims. Points are subordinate to a claim; a claim is not subordinate. The adjudication is whether removing that point collapses the claim.

There is another form that assists the adjudication: writing out, for each heading, which claim it supports. Two or more claim names appearing means two articles are inside this one.

And headings emerge for which you cannot write out a supported claim. Those are items that are merely related, and they are candidates for removal from the main line. What decides removal is not the strength of the relation but whether it supports.

Depending on the subject, two claims sometimes have to sit side by side — where the boundary between the two is itself the subject. Here the claim becomes the single claim the two are different things.

Narrowing carries resistance. Both are things you want to say, so lowering one looks like a loss. In fact the lowered side does not disappear; it moves to the position of grounds, or to another article. What disappears is only the option of raising both as claims at once.

Narrow the claim to one and the number of articles you can write increases, because the claim you cut becomes the centre of another article. The state in which a structure will not come together is frequently the state of having two articles rather than one.

Length Is Settled by the Distance From the Reader

What settles the length of an article is not the size of the subject.

What settles it is the gap in premises between writer and reader. A large gap requires many steps to fill. A small gap requires few.

Convey the same claim to somebody working in the same territory and a few lines suffice, because much is shared. To somebody outside the territory, you end up building from the premises.

This structure holds across forms of writing in general. Long sales writing is long for the same reason. The shallower the relationship, the more there is that must be confirmed.

Length is therefore not a target. It is the quantity that results from having filled the gap in premises. Wrap up before filling it and the piece is short; add after filling it and the piece is padded.

A doubt arises about the notion that longer articles get better evaluations. The search side has stated plainly that it does not treat character count as a target. The observed correlation is accounted for in the sequence articles with thorough treatment turn out long. Reverse it and you settle the length first and add content afterwards.

The gap in premises differs by reader. One article cannot fill every reader’s gap. What gets settled is which position of reader to match. That is a judgement belonging to the setting of the target.

This structure has a practical use. When an article runs longer than assumed, the cause can be separated in two: a subject with a large gap in premises, or the same thing said repeatedly.

The separation becomes visible by compressing each step into one sentence. If the compressed sentences line up into an argument, the length is a consequence of the number of steps. If the same content appears twice, it is repetition.

This compression is separate from cutting. Before deciding whether to cut, you adjudicate what the length actually is. Cut without adjudicating and what falls is the steps rather than the repetition.

There is also a limit on how far you can measure the gap from your side. What the reader already knows is not directly visible. What is visible is the level of explanation standardly shared in that territory. You take that level as a floor and build from there.

Do the work of filling the gap thoroughly and the article gets long. And a long article is redundant for readers who already hold the premises. The two do not coexist, so one of them gets chosen. Make the choice explicit and readers outside the target can leave early, which saves expenditure on both sides.

A Template Is a Skeleton, Not the Correct Order

A template lowers the load of assembly and does not guarantee the correctness of the order.

The function a template performs is lowering the load of assembly. Hold the sequence claim, reason, concrete, conclusion, and you avoid thinking out what to write from zero every time. This benefit is real.

The function a template does not perform is guaranteeing that the order is correct. A template generalises a sequence that functions in many settings; it is not the optimal sequence for a particular claim.

With a claim the reader already holds an opposing view on, placing the claim first gets it refused. Here what goes first is the observation you can share. The sequence becomes sharing an observation and then advancing to the claim.

The judgement to depart from a template is available only where you know what the template is doing. Without that, the two options are following the template or writing without order.

The view that following a template avoids going badly wrong carries a condition. It does not go wrong. Nor does it hit. A template is an instrument guaranteeing average results and nothing more. That is not a defect; that is the kind of instrument it is. Raising the floor and raising the ceiling are different operations.

At an unpractised stage, following the template produces better results, because departing with no material for judgement leaves the departure unexplainable. The judgement to depart becomes available only after several pieces written with the template.

What departing from a template requires is not a substitute order but grounds for the order. If why place the observation first here can be answered in one sentence, departing is fine. If it cannot, there are no grounds for departing.

This adjudication applies to the template itself. For passages that follow the template, examine whether you can answer why that order. If you can, you are not following the template; you merely happen to coincide with it.

Coinciding and following cannot be distinguished from outside. Only the writer can distinguish them, and whether they have is what appears the next time a subject arrives that the template does not fit.

It appears not in the form of being unable to write. It appears in the form of writing successfully to fit the template. The claim shrinks to the size the template accommodates, and the shrinking does not appear on the page. The point that dropped out went unwritten, so rereading will not find it.

Keep using a template and you stop raising claims that do not fit inside it. A structure that does not fit claim, reason, concrete — the structure showing a boundary between two things, for instance — is awkward to handle inside the template. Awkward structures gradually stop being written.

Whether the Order Has Grounds Can Be Measured Before You Write

Three things can be measured before the body gets written.

The first confirmation is to write the article’s content in one sentence.

If you can, there is one claim. If and or also enters, there are two or more. The check takes a few minutes, and it removes one of the causes of rebuilding a structure.

The second confirmation is to rewrite the headings in question form.

A heading you cannot rewrite is a category name. A heading left as a category name has not settled what would finish it. Having rewritten it, try answering that question in one line. A heading you cannot answer is a heading you cannot write.

The third confirmation is to swap the order of the headings and see whether it still holds.

If it still holds after swapping, that order has no grounds. An order without grounds is a list of items. If it stops holding once swapped, the earlier heading is a premise of the later one, and the argument is connected.

Coverage is not what any of the three examines. Coverage is a necessary condition for a structure, not the structure itself.

There is something the three do not adjudicate: whether the claim has value for the reader. An article with the claim fixed at one, with grounds for the order, and still already known to the reader, is entirely possible. That adjudication cannot be made on the side of the structure. What can be made, once, before writing, is confirming that your claim is not the same as the answer the top-ranking articles already give. If it is the same, no assembly of the structure will produce anything new.

How many headings to have, how many characters under each, how far to fill in premises. No distribution table comes with this. If length is settled by the gap in premises, a distribution cannot be written without looking at that gap. Handing over a table would be settling the quantity without examining the gap.

Run the procedure to the end and what comes out is a list of items. A list contains no principle for settling order. A reader assembles the gist while reading, so order is settled by whether the next step loads; whether it loads is settled by the claim being fixed at one; and quantity is settled by the gap in premises with the reader. The three are not separate rules. They are separate faces of the single fact that order comes out of the claim.

The criterion for the structure you assembled comes down, in the end, to one line. Can you answer, in one sentence, why it has to be this order? The places where you cannot are the places still sitting there as a list.

How to raise the headings themselves is treated in how to write headings, and how to write the opening few lines in how to write a lead paragraph. Where what to raise as a claim will not settle in the first place, when you do not know what to write about sits earlier in the sequence. The question of why it has to be this order applies not only inside an article but to the order of the writing work itself. The side of where to begin is held by verbalisation as a whole.