Corrections and re-checks

Errors are fixed in place, with a date

When a page is wrong, the wrong text is corrected on the page itself and a short note records what changed and the date of the change. The page is not deleted and republished to look as if the error never existed, and a correction is not buried where a reader would not see it. A visible trail is the point.

A correction is made when the desk got something wrong: a figure transcribed incorrectly, a licence statement that does not match the regulator's register, a calculation that does not follow from its own inputs, or a description that misstates what a broker's document says. In each case the corrected version replaces the old one and the note explains the difference.

How Each Correction Is Recorded and Dated on the Page

Each correction is recorded as a dated entry attached to the specific section it changes, not buried in a general changelog. The entry shows the date the correction was made, the sentence or figure as it previously stood, the corrected version, and the source relied on. For example, if the margin example for a 0.10-lot gold position at the applicable leverage cap was published as a different figure and is corrected to about $85.50, the entry names that figure and the date. The date is the day the corrected text goes live, not the day a reader first reported it.

The date format is written out in full, such as 14 March 2025, to avoid the day-month ambiguity that numeric dates create for readers in Kenya and elsewhere. Each entry carries a short identifier so it can be referenced in correspondence, and the identifier is stable even if the page is later reorganised. If a correction is itself corrected, the original entry is not overwritten; a new entry is added and linked to the first. That way the sequence of changes remains readable in order, and a reader can see not just the current text but how it reached that state.

Corrections are dated at the section level and, where a single figure is involved, repeated in the relevant table or example so a reader scanning the number sees the correction without reading the whole log. The reference price of approximately 4275.0 is a snapshot, not a live quote, and any correction to that stated level is dated to the day the snapshot was taken and the day it was corrected. Dates are never backdated to make a page look as though it was right earlier than it was. The log records what happened, in the order it happened, with the dates we can stand behind.

Why No Entry Is Ever Deleted From the Record

No entry is ever deleted because the record is only as trustworthy as its weakest retained entry, and a log that can be pruned is a log that can be gamed. If we could remove an entry, every reader would have to wonder whether the absence of an entry meant no change or a deleted one. Keeping everything removes that doubt. It also means a reader who saw an earlier version of the page can find the entry that explains the difference, rather than being left to assume they misremembered or that the page changed without explanation.

Deletion is not the same as correction. If an entry itself contains an error, such as a wrong date or a misattributed source, a new entry corrects it and the original stays visible with a note pointing to the correction. The same applies to an entry that names the wrong change type: the original mark stays and a later entry reclassifies it. This is more cumbersome than editing in place, and it is the cost of a record that can be checked. The only things removed are accidental duplicates created by a publishing fault, and even then a note records that a duplicate was removed and when.

Retention also protects against a subtler problem: the temptation to tidy history into a story where we were always right. A page about gold trading, with its lot sizes, pip values and margin figures, is exactly the kind of page where a quiet edit could hide a consequential mistake. Keeping every entry, including the ones that show a slow fix or a contested call, makes that impossible. The log is not a marketing asset and is not curated for tone. It is a factual record, and its length is a feature, not a defect.

When a Cited Source Changes Without Notice

When a source changes silently, we treat the page as potentially wrong from the date the source changed, not from the date we noticed. The first step is to capture what the source says now and, where possible, what it said before, using an archived copy or our own saved record. If the earlier text on our page matched the earlier source, the change is logged as an update, because our page was right when published. If it did not match, the change is logged as an error, because the page was wrong at publication regardless of what the source has done since.

Silent source changes are the main reason we keep our own dated copy of every source we rely on for a stated fact. A regulator page that quietly edits a licence entry, or a broker page that removes a funding method, can leave our text stranded without any visible event to point to. Because we hold the dated copy, we can show what we relied on and when. Where we cannot show it, we say so in the entry and mark the claim as unverified rather than pretending the record is complete. That honesty is more useful than a clean-looking log.

If a source changes and then changes back, we do not delete the intervening entry. The log shows the sequence: the source moved, we updated, the source moved back, we updated again. A reader can see that the underlying condition was unstable, which is itself useful information for anyone relying on a regulatory or funding detail. We do not treat a reverted source as evidence that nothing happened, and we do not use the reversion to remove an entry that records a real period of change. The record follows the source, including its reversals.

A changed condition is a re-check, not an error

If a broker has since changed a term, a funding route or a margin requirement, the old page was not wrong when it was written. That is a re-check, not a correction: the page is revisited, the figure is updated, and the date it was read is moved forward. The distinction matters because it keeps the correction log honest about what the desk actually got wrong.

This is why every figure carries the date it was read. Gold near 4275.0, a lot of 100 oz, a pip of 0.01 and a margin figure of about $85.50 on 0.10 lots are all snapshots of a stated set of terms. Confirm them with the broker before you rely on them, and tell us if they have moved.

Why the Correction Log Stays Public and Permanent

The log stays public because a reader cannot judge the reliability of a page without seeing how it has changed. Trading content is high-risk, and a wrong lot size or a wrong pip value for gold can lead to a position far larger than intended. If corrections were made quietly, a reader who acted on the earlier text would have no way to know the page had moved. A public log lets anyone check whether the figure they relied on is still the figure we stand behind, and it lets them see how quickly and how honestly we respond when we get something wrong.

Permanence is part of the point. If entries could be removed once they became embarrassing, the log would only show the corrections we were comfortable with, which is close to no log at all. Keeping every entry, including ones that show a slow response or a repeated mistake, gives the record its value. It also creates a useful discipline: because we know an entry will stay, we take more care before publishing and we fix things properly rather than patching the visible sentence and leaving the underlying table wrong.

The log is kept on the same page as the content it describes, not on a separate page that readers have to find. A correction to the margin example for a 0.10-lot gold position at the applicable leverage cap sits next to that example, so a reader who reaches the figure also reaches the note explaining any change to it. This is deliberately inconvenient for us and convenient for the reader, which is the right trade for a page about risk. The log is not marketing material and is not used to advertise responsiveness.

What Qualifies as a Loggable Error on This Site

A loggable error is a factual statement on the page that is demonstrably wrong against the source we cited at publication. That includes a wrong instrument specification, such as stating 1 standard lot of gold as 10 oz instead of 100 oz, or a pip value given as 0.1 instead of 0.01. It also covers a wrong regulator name, a wrong currency symbol, or a funding method listed for Kenya that is not M-Pesa. Typos that do not change meaning, formatting quirks, and differences of opinion about risk framing do not qualify and are not logged as errors.

An error must be traceable to a specific sentence, table cell or figure, not to the page as a whole. If a paragraph says the reference price is approximately 4275.0 and the cited source at the time said a different level, that is a loggable error only if we can show the source we relied on. Broken links, missing pages and outdated screenshots count as errors when they prevent a reader from verifying a stated fact. A reader disagreeing with our editorial angle, or wanting a different broker comparison, is feedback, not an error, and is handled outside the correction log.

Errors of omission count when a required disclosure is missing, such as failing to state that FxPro is licensed by the FCA (UK), CySEC and FSCA but not by the CMA of Kenya. Errors of emphasis do not count: if we say leverage is a cap and a reader thinks we should have said it more strongly, that is a style note. The test is simple: can we point to a source and show the page departs from it? If yes, it is logged. If the only basis is preference, it is not. This keeps the log credible and short enough to be useful.

Who Decides on a Correction and Within What Time

The decision to log a change sits with the editor responsible for the page, not with the person who wrote the original text, so that a writer is not the sole judge of whether their own work was wrong. The editor checks the claim against the cited source and assigns the mark: error, update or clarification. Where the claim concerns regulation, the editor confirms the exact wording, including that FxPro is licensed by the FCA (UK), CySEC and FSCA but not by the CMA of Kenya, before any change goes live. No change is published without that check.

The time target is set by the type of change. A reported error that affects a figure a reader could act on, such as lot size, pip value or margin, is checked and, if confirmed, corrected on the same working day it is verified. Other errors are handled within a short fixed window that the page states, and updates are made when the changed condition is confirmed rather than on a fixed schedule. Clarifications follow the same window as other errors. The target is a commitment about our process, not a promise about any particular outcome, and missed targets are themselves noted in the log.

If the editor and the original writer disagree, the editor's decision stands and the disagreement is recorded in the entry rather than resolved silently. That record is useful: it shows readers that the call was contested and on what grounds, and it stops a later reader from assuming the change was obvious. The editor also decides when a report is not an error and says so in a short reply, so the reader is not left waiting. Decisions are not delegated to anyone without access to the cited sources, because a decision made without the source is a guess with a date on it.

The Three Change Types and the Mark Each One Carries

Every change is marked as one of three types: error, update or clarification. An error mark means the published text was wrong against its cited source at the time of publication. An update mark means the text was right when published but the underlying condition has since changed, such as a regulatory or leverage change affecting the entity serving Kenya. A clarification mark means the text was not wrong but could reasonably be read in a way we did not intend, and the wording has been tightened. The mark appears at the start of the entry and is repeated in the section where the change was made.

The marks are plain words, not symbols or colour codes, so they survive copying, printing and screen readers. Each mark is followed by a one-line reason and the date, and, for errors, the source that shows the original text was wrong. Clarifications do not require a source, because they do not assert a new fact; they restate an existing one more precisely. Updates require the source of the changed condition, which may be a regulator notice or a broker disclosure. A reader can therefore tell at a glance what kind of change they are looking at and how much weight to give it.

The three marks are not interchangeable and are not chosen to make the record look better. An error is never relabelled a clarification after the fact, and an update is never relabelled an error to appear more responsive. If a mark was itself wrong, a new entry corrects the mark and the original entry stays visible with its original mark, so the sequence shows the reclassification. This is stricter than most correction policies, and it is deliberate: the value of the log depends on the marks being trustworthy, and trust in the marks depends on them not being quietly adjusted.

How a Reader Can Check the History of Any Page

A reader can check a page's history by opening the correction log attached to that page and reading it in date order, newest first. Each entry names the section, the change type, the date and the source relied on, so the reader can go from the current text to the entry that explains it and, where relevant, to the source itself. The log is on the same page as the content, so there is no separate archive to search and no risk of reading a log that has drifted out of step with the text it describes.

For a specific figure, the reader should check three things: the current figure, the most recent entry that mentions it, and the date of that entry. If the figure is a margin example, such as about $85.50 for a 0.10-lot gold position at the applicable leverage cap, the entry will say whether the figure was corrected, updated or clarified, and on what date. If there is no entry, the figure has not changed since publication. That is a check a reader can complete in under a minute, which is the point: a history that takes too long to verify is a history nobody verifies.

Readers can also compare the page against an archived copy taken at an earlier date, and where we hold a dated source copy we say so in the entry so the comparison can be made directly. If a reader believes the log is incomplete, the same reporting route used for errors applies, and the report is itself logged with its outcome. A reader who finds a change that has no entry has found a fault in the log, and that fault is treated as an error in its own right. The history is only useful if it is complete, so gaps are corrected, not explained away.

How to report one

Send the page address, the sentence you believe is wrong, and whatever you have that shows the correct position, to editor@nairobiboardke.com. A broker document, a screenshot of a regulator entry or a link to a published term is enough to start; you do not need to write a long explanation.

Every report is read. If it is an error, it is fixed in place with a dated note. If it is not an error but the page could be clearer, the wording may be tightened without a correction note, because clarity is not the same as accuracy. No response time is promised, and a report that needs a document checked takes longer than one that does not.

Correction Versus Update: Where the Line Sits

A correction fixes something that was wrong when it was published; an update changes something that was right when published but is no longer current. If we wrote that a broker was regulated by a particular body and that was accurate on the date of writing, a later change in licensing is an update, not a correction. If we wrote the wrong regulator name in the first place, that is a correction. The distinction matters because it tells the reader whether the original page was unreliable or simply overtaken by events, and those are different kinds of trust.

Updates are dated the same way as corrections but are labelled as updates and linked to the condition that changed. A change in the leverage cap that applies to the entity serving Kenya, for instance, would be an update because the cap is set externally and can move. A change in the stated pip size for gold from 0.01 to something else would be a correction, because 0.01 is a fixed specification and was either right or wrong at publication. The label is decided by asking what was true on the original publication date, not by asking what is convenient now.

A single edit can contain both. If a page stated a margin example based on an old leverage cap and also misspelled a regulator's name, the margin change is an update and the spelling is a correction, and they are logged as separate entries on the same date. Readers should not have to guess which is which. Every entry states its type in the first line, so the log can be scanned for corrections alone if that is what a reader wants. Nothing is relabelled later to soften the record.