The review process

Most of the review on this site is performed by code, and that is deliberate — a gate that runs on every build cannot be forgotten under deadline.

Published by the Land Clearing Authority editorial team · Last revised 2026-07-29 · Editorial policy · Corrections

Most published editorial standards describe intentions. The trouble with an intention is that it can be skipped when a deadline arrives and nobody finds out. So the majority of the review on this site is executable: it runs on every build, it fails the build rather than logging a warning, and it cannot be waived for a single page.

There are currently 26 prohibited-claim patterns, 55 registered sources each carrying its own declared scope, and 95 resource library pages that pass the same structural validation as the rest of the site.

What runs on every build

Automated review gates, what each verifies, and what happens on failure
Check	What it verifies	On failure
Structural validation	Every page must state who it is for, what changes the answer, at least three cost drivers, common mistakes, questions to ask, next steps, and at least three questions answered.	The registry throws at module load. The page cannot exist in a passing build.
Prohibited-claim scan	The rendered text of every exported page is scanned against a published list of sentences this platform will not write.	The build fails, naming the page and the matched phrase.
No published price figures	Any currency figure in the rendered homeowner text is a failure, because a national price for site-priced work would be invented.	The build fails and prints the offending page.
Source scope enforcement	Each citation names a registered source and a claim that source is registered to support. The cross-check runs in both directions.	Citing outside scope throws; a source listing a page that does not cite it also throws.
Near-duplicate detection	Every page is compared against every other page in its pool for editorial overlap.	A pair above the ceiling fails the build and is named.
Thin-content floor	Every substantive page must render a minimum quantity of actual substance, measured across the fields that carry meaning.	A short page fails at registry load and again at QA.
Link graph integrity	Every internal link must resolve to a registered route, and every registered route must have inbound links.	A broken link or an orphan page fails the build.
Schema validity	Structured data is generated from the same arrays the page renders, and only permitted types may appear.	An unpermitted type or an empty required field fails the build.
Revision log integrity	Every route must have a dated revision entry, and every entry must name a real route.	A missing or orphaned entry throws at registry assembly.
Protected market exclusion	No route, page title, or county name may target a market served by a separate regional property.	Route assembly throws rather than warning.

What a person still has to do

Automation catches structure, consistency, and prohibited language. It cannot judge whether a paragraph is true, whether an explanation is the clearest available, or whether a page answers the question a reader actually had. Those stay human judgements, and they are where the editorial time goes.

Deciding whether a claim is supportable at all, before it is written.
Choosing which authority is right for a claim, and what it can honestly support.
Judging whether a page is genuinely distinct from its neighbours or merely differently worded.
Reading a page as a landowner would, and asking whether it resolved anything.
Deciding when a page should be retired rather than updated.

Reviewer roles

Review here is defined by role rather than by named individual, for the reason set out on the editorial team page: no individual has been attached to this content, and naming one who had not reviewed it would be a fabrication. The roles below describe what is actually performed.

Scope review — does this page answer a real decision, and is it distinct from what already exists?
Claim review — is every factual assertion either supportable from a registered source or clearly framed as a judgement?
Boundary review — does the page stay inside what this platform can know, particularly on permits, boundaries, wetlands, protected species, and contractor status?
Local-deference review — where an answer depends on jurisdiction, does the page route the reader to the authority rather than guessing?

When review finds a problem after publication

Corrections are made in place and recorded in the revision log with the date and what changed. A correction is never made silently, and a page that has been corrected keeps the record.

Questions about review

Is every page checked by a subject-matter expert?

No, and claiming so would be false. Pages are checked against registered sources for external claims, checked automatically for prohibited assertions and structural completeness, and read for whether they resolve a decision. Where an answer depends on local rules or professional judgement, the page routes you to the authority instead of substituting for it.

What stops a page making a claim it cannot support?

Two mechanisms. The citation system refuses to render a source outside its registered scope, and the claim scan runs a published list of prohibited assertions against the rendered text of every page on every build. Both fail the build rather than warning.

How often is content reviewed?

Every published page is re-read against its sources at least annually, and immediately whenever a cited authority changes a program or a document moves.

Can I see what changed on a page?

Yes. Every page shows the date of its last recorded revision, and the revision log records what changed. Those dates come from written entries rather than a build clock, so a page that has not changed does not advertise that it has.