About the imprint
An imprint, not a platform.
Independent technical publications, each written by an engineer who works in that language every day, and a single standard of evidence across all of them. Cognate Press puts its name on writing by the person who did the work — and on nothing else.
im·print
/ˈɪm·prɪnt/ · noun · a publisher’s mark
A press mark is shorthand for a standard. A reader who trusts the imprint does not have to re-check every claim from first principles — the name is an assertion that somebody already did. This one asserts exactly one thing, and it is the narrowest claim we could make: the person who wrote this does the work they wrote about.
Why an imprint
Most technical writing on the open web is assembled rather than lived. It restates documentation, arrives at the advice without passing through the mechanism, and stops exactly where the interesting part starts. It is not dishonest; it is just written by someone who read about the thing instead of running it.
Cognate Press was built the other way round. A title opens only when there is one engineer who works in that language every day and will put their name on every issue of it. That constraint is expensive — it is why there are seven publications and not sixty — and it is the only thing the imprint sells. Where the asterisk appears, a single specialist wrote every word, and they do the work they write about.
What it is
- A publisher’s mark on independent technical titles, each on its own domain.
- One language per title, one specialist per title, one byline per title.
- Writing distilled from systems its author built, broke, measured and repaired.
- A standard of evidence a reader is invited to hold the byline to.
What it is not
- Not a platform — there is no way to sign up and publish here.
- Not an aggregator — the imprint hosts nothing and reprints nothing.
- Not an agency — nothing ghostwritten, nothing syndicated, nothing generated.
- Not an ad network — no sponsored posts, no affiliate links, no paid placement.
The mission
Publish the layer under the documentation — measured, attributed and checkable — one language at a time.
Documentation states what a system does. Most articles state what to do about it. The layer in between — why it behaves that way, measured on a stated setup, with the boundary of the result written down — is what every title in this imprint is for. It is the layer that takes production experience to write, and it is the one that cannot be assembled from a search.
The vision follows from it: a reader who works in any language the imprint covers should be able to open its title and find the question they are actually stuck on already answered — with the method shown, so they can disagree.
The standard every title is held to
Five commitments, identical across the imprint. They are not aspirations — each one is a thing a reader can check on any issue and catch us failing at.
Mechanism before advice.
Advice that arrives before the mechanism is a rule you have to trust. Advice that arrives after it is a conclusion you can check, disagree with, and apply to a case the article never mentioned.
Every number carries its method.
A performance claim arrives with the version, the machine, the command and the output it produced — or it says plainly that the thing was not measured. “Faster” without a figure is an opinion, and a benchmark without its method is an opinion with digits.
One byline, and it is a real practitioner.
Every publication carries exactly one name, and that name works in the language it writes about. A title opens when such a person is ready to sign it, and not before.
Nothing published here is paid for.
No sponsored posts, no affiliate links, no vendor writing about its own product under a byline. When a tool is recommended, the reason is in the article and the reason is the only reason.
Corrections are published in place.
A claim that turns out to be wrong is corrected on the page that made it, carrying the date and the evidence that changed it. Nothing is quietly rewritten, because a record that can be edited without a trace is not a record.
How the imprint is put together
The structure follows the promise. If the writing is the product, then nothing central may be allowed to speak over it — so the centre holds as little as it can.
Separate domains, separate builds
Every title is its own site, its own repository and its own deployment. The imprint runs no shared platform underneath them, so a title going quiet takes nothing else with it.
A mark with no brand colour
Historical linguists mark a reconstructed root with an asterisk. The emblem is that asterisk, drawn in three strokes — and it is always set in whatever colour the page around it is already using, so it can sit in a footer the imprint does not control.
Components, not a template
What the titles share is a handful of framework-free blocks — a network bar, a footer grid, a signature — that inherit the host page’s type and colour rather than imposing any. No stylesheet is handed down.
What comes next
The imprint will grow, and it grows one way: a title opens when someone who works in that language every day is ready to put their name on it — not when a keyword looks promising. There is no schedule and no target count. Nothing on this site assumes there are exactly seven, which is why adding one is a data entry rather than a redesign.
The seven names on these publications are publishing bylines. A byline identifies the author of the writing that carries it, and nothing else — the party responsible for this site, and for the processing described in its legal pages, is the imprint, named in the Privacy Policy.
Common questions
The same seven, asked often enough to be worth answering in one place. Each answer is a commitment made somewhere above, restated plainly.
Does Cognate Press accept guest posts or submissions?
No. There is no way to sign up and publish here, and nothing is ghostwritten, syndicated or generated. A title opens when an engineer who works in that language every day is ready to put their name on it, and not before.
Is anything published here sponsored?
No. No sponsored posts, no affiliate links, and no vendor writing about its own product under a byline. When a tool is recommended, the reason is in the article and the reason is the only reason.
Who writes each publication?
Exactly one person. Every publication carries a single byline, and that byline works in the language it writes about — a practitioner who does the work before writing about it, one to a title.
Are the publications hosted on one platform?
No. Every title is its own site, its own repository and its own deployment. The imprint runs no shared platform underneath them, so a title going quiet takes nothing else with it. What the titles share is a handful of framework-free components, not a template.
How are performance claims verified?
A performance claim arrives with the version, the machine, the command and the output it produced — or it says plainly that the thing was not measured. “Faster” without a figure is an opinion, and a benchmark without its method is an opinion with digits.
What happens when an article turns out to be wrong?
It is corrected on the page that made the claim, carrying the date and the evidence that changed it. Nothing is quietly rewritten, because a record that can be edited without a trace is not a record.
Where do I write about a specific article?
To the domain that published it. Each publication answers on its own domain, at contact@ that domain. The imprint’s own address, contact@cognate.press, answers for the network as a whole — rights, press, the newsletter and the legal pages.
Read the titles, read the people behind them, or write to us — the imprint answers for the network, and each publication answers for its own domain.