Deaths today this week: verified reports and corrections

Every item a records desk handles is in one of a small number of states, and most of the confusion readers feel about death reporting comes from not being able to see which state an item is in. A report that has been heard looks, from the outside, exactly like one that has been confirmed. The desk knows the difference. The reader usually does not.

This article sets out those states, how an item moves between them over the course of a week, and what "verified" and "corrected" mean at each stage. No person is named here and no death is reported.

What to take away

• An item passes through heard, held, confirmed, published, and sometimes corrected or withdrawn. Only the published state is visible, and only after confirmation.
• Most items never leave the held state. That is the system working.
• A correction changes a published entry and leaves the change visible. A withdrawal removes an entry that should never have been published, and says so.

The states

Heard. A report has reached the desk from somewhere: a wire, a post, a message, a reader. The desk writes down what was heard, from where, and when. Nothing is published. Nothing is assumed.

Held. The report has been looked at and cannot yet be confirmed. It stays in the file with its source and date. Most items that are heard on any day are held at the end of it, and a great many are still held a week later. Holding is not doubt; it is the absence of a direct source.

Confirmed. A source with direct knowledge has established the fact and the identity: a family or someone speaking for them, a funeral home, an institutional statement, or a register entry made under a jurisdiction's own statutory process, of the kind described in the official guidance on registering a death. Two independent routes are better than one. At this point the desk has something it can stand behind.

Published. The confirmed item has been entered on a page, with its sources and the dates on which they were checked. This is the only state a reader sees.

Corrected. A published entry has been found to contain an error in a field: a spelling, a date, an affiliation, a survivor's name. The entry is amended, the amendment is dated, and the original state is noted so the change can be seen.

Withdrawn. A published entry should not have been published at all, most seriously because the person is alive or because the death belongs to someone else. The entry is removed from the list, a dated note is left saying that an entry was withdrawn and why, and the note stays.

How items move during a week

Over any seven days the traffic between states has a shape, and knowing it explains why a weekly page looks the way it does.

Early in the week the desk hears a great deal and confirms very little. Reports are fresh, the people who could confirm them are occupied, and almost everything goes into held.

Over the following days, held items start to resolve. A family statement appears, a funeral notice runs, an institution says something. Each such event moves one item to confirmed, and from there to published, dated on the day it happened. Items from the previous week resolve alongside items from this one, which is why a weekly page always contains entries relating to deaths from before its window.

Some held items resolve the other way. The person posts something. A representative issues a denial. The original report is traced to an account with no standing. These items are closed in the file with a note, and they never reach a page. Nothing about them is published, because publishing a denial of a report the desk never repeated only spreads the report.

At the end of the week, the held file is still full. Most of it will stay full. That is the normal condition and it is not a backlog.

For what a weekly window can and cannot contain, the parent page on deaths this week sets it out.

What "verified" means, by state

Readers ask whether a report is verified as though it were a single property. It is a stage, and the honest answer depends on where the item is.

An item that has been heard is not verified in any sense. An item that is held has been looked at and found insufficient, which is a finding but not a verification. An item that is confirmed has been established by a direct source, and that is what verification means. A published item was confirmed before it was published, and its sources are on the page.

When a desk says it "cannot verify" a report, it means the item is held. It does not mean the report is false, and it does not mean it is true. It means the desk does not yet have what it needs, and it will wait.

Corrections and withdrawals are different acts

A correction fixes a field in an entry that belongs on the page. A withdrawal removes an entry that does not. The distinction matters because the two are handled differently and mean different things to a reader.

A corrected entry stays where it is, with the corrected field, a date, and a note of what was changed. The person did die; the record was imperfect; the record is now better. Readers who saw the earlier version can see what changed.

A withdrawn entry leaves the list. What remains is a dated note that an entry was published and withdrawn, and the reason in general terms. This is the most serious event a desk can have, because it usually means a living person was listed among the dead, or a death was attached to the wrong person. The note stays so that anyone who copied the original can find the withdrawal, and so that the desk's own record shows what happened.

Neither is done silently, and that is not a house preference: standing commitments such as the Associated Press statement of news values and principles undertake to acknowledge errors openly rather than to edit them away. Silent correction leaves the original circulating with no trace that it was ever questioned. Silent withdrawal is worse, because it erases the only evidence that the desk acknowledged the error.

Keeping the file

The file that holds all of this is the real product of a records desk, and it is worth more than the page. Every item in it has a source, a date heard, a current state, and a note of what would change that state. When a page is questioned, the file answers. When an entry is corrected, the file shows why. When a report resurfaces months later, the file shows it was heard before and what became of it.

A desk that keeps only the page, and not the file, cannot explain itself, and cannot correct itself properly.

The states described here begin on the first day, and deaths today explains why almost nothing leaves the heard state on that day. The same lag pushes late-year confirmations into the following year's record, which deaths by year takes up. For why the cause field stays empty through every state, see cause of death guides.

Common questions

A report has been circulating for days and is not on the weekly page. Has it been rejected?

Probably not. It is most likely held, which means the desk has no direct source yet. Rejected items are closed in the file and never published; held items wait.

If the desk knows a report is false, why not say so?

Because publishing a denial repeats the report to people who had not seen it. Where the desk never published the claim, the correct response is to not publish it. Where a person or their representative has publicly denied it, that denial is theirs to make, and it can be pointed to.

What is the difference between a correction and a withdrawal?

A correction fixes a detail in an entry that belongs on the page. A withdrawal removes an entry that should never have been there. Both leave a dated, visible note.

How long can an item stay held?

Indefinitely. Some deaths are never publicly confirmed, and the item stays in the file with its last known state.

Why does a weekly page include deaths from before the week?

Because the page is organized by the date of confirmation, and confirmations lag deaths by days or longer. An entry dated within the week relates to a death confirmed within it, wherever the death itself falls.