Blog · build notes
Building this site in the open
Updated as the build proceeds.
The site you are reading is being built by the system it describes. The same toolkit that scopes a business problem scoped this one; the same decision discipline that governs a programme governs these pages; and every change is logged as it happens, not afterwards. This post is the running record, kept honestly, including the parts that went wrong.
What is built so far
A person-first landing built on one principle, one decision per screen, so a reader with thirty seconds is never faced with ten choices. Then the whole operating system as four chapters: what it is, what it is made of, what it has read, and how it is actually used. The content is drawn from a single source document, run through an automated sanitisation gate that fails the build if any identifying term survives, so nothing here names a client, an employer or a colleague.
What went wrong, honestly
Midway through, the source document was found truncated by a background file-sync landing mid-edit. The build halted, the loss was scoped, the file was repaired and flagged for cross-check, and the incident became a standing habit: verify a large file still ends where it should after a batch of edits. A system that argues for evidence-first work has to show its own failures, not just its wins.
Now live
The site is published on its own domain, deployed from a single dated package after a full verification walk. The simplest check is still the best one: whether the thing described here exists, works, and contains no identifier the gate bans. The full account, including the engineering record, is on The Build.