SASentiment Analysis Platformworld wire, filtered

Briefing Desk › MTH

Editorial standards for the desk

Neutrality on a desk like this is not the absence of choices. Someone decides which wires are read, which topics a reader follows, which of nine copies of a story becomes the one kept. Each of those is a judgement, and a judgement made quietly is the kind a reader cannot check. The standards below exist so that the judgements are stated in advance, applied the same way to every story, and visible afterwards.

Selection is a choice, so the rule has to be public

The desk does not pretend to survey the world evenly. It reads a broad stream, then keeps what a topic set asks for. That is a rule, and a rule can be written down, tested against a case, and held to. Four commitments follow from it, and every one of them is checkable by a reader who knows the thread.

  • The topic set is applied as written. If your set names shipping and ports, an item about a harbour strike is kept because of the words in it, not because the desk finds strikes interesting this week.
  • No story is kept because it flatters a position. Nor dropped because it embarrasses one. The awkward item is the test of the whole arrangement.
  • No story is dropped for being dull. A quiet customs notice can matter more than a loud press conference, and the mark on the item says what it is, not how it feels.
  • Disagreement is kept as disagreement. Where two editions read an event differently, both readings are named. Averages hide exactly what a careful reader wants to see.

How those commitments are carried out inside the filter is described on the mechanics page; what a reader can change, by contrast, is set out under choosing topics.

What neutrality does not mean

It does not mean that a false claim and a verified figure get the same space. A summary reports what was said and, where the record allows, what is known; it does not present a claim as though the desk had weighed it and found it equal. It also does not mean a reflex to find two sides of a fact. A bridge either reopened or it did not. The even-handed sentence is not the sentence that splits the difference; it is the sentence that says which part is contested and which part is not.

Nor does neutrality require a flat voice. Plain words are not the same as neutral words, and a desk that writes in grey all the way down is not being fair, only unreadable. The aim is a summary a reader can trust line by line, with the pitch of the original writing marked rather than copied.

The tone mark is shown, not implied

Every item arrives with a tone reading, drawn from the language of the item kept: the weight of its adjectives, the direction of its verbs, the way it names people, the distance it keeps from what it describes. The reading is presented on a fixed scale, so two items can be compared and neither is judged in secret.

The bar above runs from a charged, adverse reading at the left, through plain reporting in the middle, to warm or approving language at the right. It is a scale of expression, not of truth. A fiercely written report of a real flood is still a real flood, and the mark says only how hot the prose runs. The tone and framing page explains the bands; the deeper measurement is on measuring tone in running text.

Where a summary ends and opinion begins

A summary answers what happened, who said so, and what is not yet known. An opinion answers what should be done about it. The two can sit near each other without touching, and the marks below are how the desk keeps them apart.

Line in a draftMarkWhy
Officials said the bridge would reopen within a week.SummaryA claim, attributed to a named source, in reported speech
The bridge will reopen within a week.Summary, mark addedAttribution lost; the desk restores it or drops the line
It is a scandal that the bridge was left to rot.OpinionEvaluative word and an implied judgment of blame
Engineers have not yet explained the delay.SummaryA statement about the record, checkable against it

The third line is not forbidden on the site; it belongs in a quoted voice, with the speaker named, or in a framing note that says plainly whose judgement it is. What is never done is to slide it into a summary and let a reader take it for the desk's own finding.

Corrections, and what the desk owes a reader

When a figure in a kept item turns out to be wrong, the thread is corrected rather than quietly re-summarised, and the earlier wording stays visible long enough for anyone who read it to see what changed. The same applies to a tone mark that was set badly: the mark is revised and the revision dated. A desk that only ever shows its finished sentences is asking to be believed rather than checked.

Working press hall with long desks, typewriters and rows of editors at their places
A press hall at the hour copy is filed. The standards above are the desk's answer to the same pressure: many versions, one item kept.

What you can hold the desk to

Four things, in plain terms. A story kept for a topic should name the topic words that brought it in. A tone mark should be explainable by pointing at the sentences that produced it. A summary should not contain a judgement the desk has not attributed. And a correction should be visible rather than folded away. If one of those fails on a thread you follow, the briefing desk would rather hear it than not, and the questions page shows how similar notes have been answered.

Sentiment Analysis Platform is an independent editorial service. It is not connected to, endorsed by, or acting on behalf of any broadcaster, publisher, newsroom, agency, or platform named anywhere on this site. Nothing published here is professional advice of any kind.