Contact — corrections, additions and takedown requests for the archive

What to write about

The archive covers a five-year period that ended in 2016, so most incoming questions fall into a small number of categories. Corrections are the most useful: if a restored dispatch reproduces a figure, a date, a name or a headline differently from the original publication, that is a defect worth reporting, and it will be checked against the archived source rather than argued about.

Restoration requests are the second category. Around 590 dispatches were consolidated rather than rebuilt, on the basis that a stub with a headline and little else is not worth a page of its own. If a specific dispatch matters to you, because you cite it, teach from it, or need it as a source, say which address and why, and the case for restoring it in full will be looked at on the evidence. The bar is whether the page has an identifiable use, not whether it has traffic.

Corrections to archived material

One distinction governs every correction: whether the error is in the reproduction or in the original. A page that misquotes the document it reproduces gets fixed. If the original dispatch was itself wrong, whether a mistaken figure, a forecast that failed or a company name misspelled by the desk in 2012, it stays as published, and where the error is material a note is added around it saying so.

That rule is not pedantry. An archive whose contents can be revised to match later knowledge is not an archive, and a page that changes under a reader is worth nothing to anyone who quotes it on a date. So the answer to "this 2013 price forecast turned out to be badly wrong" is that the page records what was forecast, and the surrounding note says what happened instead.

What to put in a correction

Four things make a correction actionable, and their absence is the usual reason one cannot be settled. The exact address, because a headline alone can match more than one dispatch and the numeric id at the end of every path is what makes them distinct. The passage as it currently stands, quoted rather than described. What it should say instead. And, where the claim is about the original publication rather than about this reproduction, where that can be checked: an archived capture, a contemporaneous report, a filing.

The last of those matters more than it sounds. Every restored page can be compared against the archived original in a few minutes, so a report that identifies the specific text is usually resolved the same way it arrived. A report that a page is "wrong about the 2012 duty change" cannot be, because the first thing that has to be established is which of the two documents is in question: the dispatch as published in 2012, or the explanatory text written for this site around it. Those are held to different standards on purpose, and the difference is set out in about this archive.

What happens after you write

Correspondence is read in the order it arrives and no acknowledgement is sent, so silence is not a decision. A correction that is confirmed against the archived source is applied to the page and the page keeps its original publication date, because the date belongs to the dispatch rather than to the repair. A correction that cannot be confirmed is left alone rather than applied on the balance of probabilities: an archive that edits toward the most recent plausible account is worth less than one that does not move.

A request that cannot be acted on is not therefore ignored. Where a dispatch is asked about and the answer turns out to be documented somewhere on this site rather than in a reply, whether that is the mechanics behind a figure, the reason a particular address consolidates into a hub or the basis on which the twenty-two were chosen, the page that covers it is extended so the next person asking finds it without writing at all. Several sections of the reference layer and of the index exist for that reason.

Images and credits

The photographs on this site are made for it. Several images that once accompanied these dispatches carried third-party watermarks or credit lines from stock libraries, and most of the rest were between 240 and 520 pixels wide, too small to publish at the sizes this design uses. None of them appears here.

If you hold rights in an image and believe it is being served from this domain, write and say which address. The old image paths resolve to the page that carried them, so sites which embedded them do not receive errors, but the image files themselves are not served.

Takedown requests

Requests to remove material are taken seriously and answered on the merits. The archive is a record of published trade journalism, and the presumption is that a published dispatch stays published, but that presumption is not absolute, and requests concerning personal data, an inaccurate statement about an identifiable person, or material whose republication causes concrete harm are treated differently from requests concerning a company's dislike of a 2013 story about it.

When writing about a takedown, name the exact address, quote the passage, and say what the problem is. That produces a decision faster than a general objection, because every restored page can be checked against the archived original in a few minutes if the specific text is identified.

If you arrived looking for a price

This is a dated record rather than a dealing desk or an advisory service: it quotes no live prices and cannot say what gold is worth today. If a current price is what you need, the useful pages here are the ones that explain how the number you are looking for is produced; see how gold is priced, which explains how the number you want is produced and by whom, and contract specifications, which explains why several different current prices exist at once.

For how the two kinds of text on this site are told apart, dated archived documents on one side and current explanatory writing on the other, see about this archive.