A publication's standards are only worth something if they are written down and specific enough to be broken. This page sets out ours.
What we choose to cover
We write about technology that affects what ordinary people buy, play and rely on. Practically, that means a story is in scope if it changes one of:
- What a device or service can do for you
- How much it costs, or how long it will keep working
- How exposed your data or your money is
- How a technology you already use actually functions underneath
We deliberately do not chase every announcement. We are not a wire service, and an article published four hours after a press release usually just restates the press release. We would rather publish something a week later that is worth reading in a year.
We do not cover celebrity news, unrelated lifestyle content, or topics outside our stated sections. A technology site that publishes anything for traffic is not a technology site.
Sourcing
Our rules, in order of preference for any factual claim:
- Primary documents. Specifications, standards texts, filings, source code, official documentation.
- Independent measurement. Testing by a party with no commercial interest in the result, with a published methodology.
- Named expert comment, where the person's relevant expertise is stated.
- Reputable secondary reporting, linked, and identified as second-hand.
Things we treat with explicit caution, and label as such:
- Vendor claims. A manufacturer's performance figure is a claim, not a measurement, and we say "the company says" rather than stating it as fact.
- Benchmarks supplied by the company being benchmarked. Reported as marketing material.
- Projections and market forecasts. Attributed to whoever made them, with the date.
- Anonymous sourcing. We avoid it. Where a claim rests on an unnamed source in another publication, we say so and let you weigh it.
Where we do not know something, we write that we do not know. Where experts disagree, we say they disagree rather than picking whichever answer makes a cleaner sentence.
Numbers
Specific figures are where technology writing most often goes wrong, so we hold them to a stricter standard:
- Every number has a source and a date, or it is marked as an estimate.
- Ranges are given as ranges, not as a single convenient midpoint.
- We state the conditions a performance figure was measured under, because a peak figure in ideal conditions is not a number you will ever see.
- We convert units and currencies explicitly rather than leaving a reader to guess.
- Where a number is disputed, we give the range of published values.
Corrections
We correct errors. We do not quietly edit them away.
How to report one. Email us with the article, the specific claim, and ideally a source. We read everything and we do not take offence.
What we do.
- Factual error affecting the substance of an article: corrected promptly, with a dated correction note at the foot of the article stating what was wrong and what it now says.
- Minor error — a typo, a broken link, a wrong label: fixed without a note.
- A headline that misrepresented the article: changed, with a note, because the headline is part of the claim.
- An article that is wrong beyond repair: retracted rather than rewritten. The page stays up, clearly marked as retracted, with an explanation. Deleting an article you got wrong hides it from the people who read it.
We aim to review reported errors within one working day.
Updating older articles
Technology articles decay. Our approach:
- Every article shows its publication date, and a separate updated date if it has been revised.
- Substantive revisions — a changed recommendation, a new fact that alters the conclusion — get a dated note explaining what changed.
- Routine refreshes to prices, version numbers or broken links update the date without a note.
- We do not silently rewrite an article's conclusion to look as though we always thought that.
- We do not backdate or re-date old articles to make them look fresh. Re-publishing an unchanged article under a new date to game search rankings is a practice we consider dishonest, and we do not do it.
Independence
- Advertisers have no editorial input. No pre-publication review, no veto, no influence on topics. Advert slots are labelled and separated from content.
- We do not publish sponsored content disguised as editorial. If we ever publish sponsored material, it is labelled as advertising above the headline.
- We do not sell links. Requests to insert a link into an existing article in exchange for payment are declined, every time. This is the most common email we receive.
- We do not accept guest posts. Almost every offer is link placement wearing a disguise.
- We do not sign coverage conditions. No approval rights, no agreed talking points, no embargo in exchange for a favourable angle.
- Affiliate commissions do not affect recommendations, and we say when the option we recommend earns us nothing.
The funding arrangements are described in full in our disclaimer.
How we use AI tools
We write about this technology, so we should be straightforward about our own use of it.
What we use it for:
- Research assistance — finding documents, summarising long specifications, checking whether we have missed an angle
- Suggesting structure for a draft
- Copy-editing: grammar, clarity, consistency of style
- Generating the abstract illustrations used as article artwork, which are produced programmatically rather than depicting anything real
What we do not use it for:
- Writing articles. Every article on this site is written by the named author. We do not publish generated text as editorial content.
- Generating facts, figures, quotes or citations. Every number and every source is verified against a primary document by a person. A model is not a source.
- Creating images that depict real people, events or products, where a reader could mistake them for photographs.
- Producing volume. We are not interested in publishing more articles than we can stand behind.
If we ever publish something substantially machine-generated, it will say so on the page.
Authorship
Every article carries a byline and a role, and each article is read by an editor before publication. Bylines are pen names: the writers behind them are real, but the site is run by a small team who prefer to publish under consistent editorial names rather than personal ones. We do not invent credentials, and we do not hide behind a generic "staff" byline to avoid accountability for what a piece says.
If you need to know who is responsible for a specific article — for a correction, a right of reply, or a legal matter — the contact page reaches the editor directly, and a named person will answer.
Comments and reader contributions
Where reader comments are enabled, we moderate for abuse, spam and personal information, not for disagreement. Criticism of our articles is welcome and we would rather have it in public.
Reader tips are gratefully received and independently verified before we publish anything based on them. We do not publish a reader's name or contact details without asking first.
Complaints
If you think we have breached this policy, email us and say which part. We will respond within five working days with either an explanation or a correction.
If you are the subject of an article and believe it is inaccurate or unfair, tell us specifically what is wrong and we will investigate properly. We will not remove accurate reporting because it is unflattering, and we will correct anything genuinely inaccurate without argument.
Changes to this policy
If we change our standards, we will update this page and the date at the top. If we ever loosen one of the commitments above — particularly on sponsored content, paid links or AI-written articles — we will say so plainly rather than editing the wording and hoping nobody notices.