Method
Editorial log
Every published page on this site, who reviewed it, when, and what the review changed.
Scaled content is penalised when it is published without review. Rather than assert that review happens, the record is kept here and the build will not publish a page whose review block is missing or dated before the page was written.
One collection is reviewed differently and the table says so rather than hiding it. Market pages carry no third-party figure — every publishable fact on one is a named authority, a section number, an institution or a link — so each is checked by retrieving the source and searching it, and the URL, the status, the hash and the matched passage are recorded. A field that cannot be established that way is dropped; a page whose required fields cannot be is not published. Everything that carries a figure somebody else measured is still read by a person before it ships.
| Page | Type | Reviewed by | Reviewed | What the review changed |
|---|---|---|---|---|
| Which appointment reminder method works best? | Answer | Answer Production Engine | 2026-08-23 | The draft ranked the methods against each other using the published range, which is precisely the figure the page shows to be unusable. Replaced the ranking with the property that actually separates them and with an evaluation the practice can run on its own patients in a quarter. Moved the telephone rules ahead of the effectiveness argument, since they constrain the choice before effect does. |
| Which business processes should never be automated? | Answer | Answer Production Engine | 2026-08-23 | The draft's fourth category was framed as permanent. It is not — a process nobody has written down is excluded until it is written down, which is a different instruction and a more useful one. Separated it explicitly. Also split the money question, since a blanket rule against automating anything financial costs businesses the safest automation available to them. |
| Which market should a real estate agent target? | Answer | Answer Production Engine | 2026-08-23 | The draft used licence counts as a direct measure of competition. They are not: a licence count includes inactive holders and licences held for adjacent businesses, so a falling count can be read backwards. Added the caveat in the body rather than as an aside, and added the area-size interaction, which is what converts the county-level figures into a decision about a specific number of doors. |
| Which practice areas should a firm advertise? | Answer | Answer Production Engine | 2026-08-23 | The draft implied a general ordering of practice areas by profitability. No admissible source supports one and it would vary by firm and market anyway. Cut it, and made the page about the ranking method and its two adjustments instead. Added the drift argument, which is the actual cost of the current practice and which a single-quarter view cannot show. |
| Which RFQs should a manufacturer decline? | Answer | Answer Production Engine | 2026-08-23 | The draft treated triage as a prioritisation problem, which reproduces the failure it describes: a priority queue still leaves unwinnable RFQs consuming estimator attention. Rewrote it around declining at receipt. Added the existing-customer exception as a separate rule after the general version would have given clearly wrong advice for a growing account. |
| Which services should a practice add capacity for? | Answer | Answer Production Engine | 2026-08-23 | The draft used the two federal series as interchangeable denominators. They are not: one counts a March week at the establishment's primary code and the other counts covered jobs, and the second omits an owner-clinician entirely. Split them into separate paragraphs and stated each exclusion where a reader would otherwise quote the figure unqualified. Also added the intake paragraph, because commissioning capacity and commissioning the intake that fills it are routinely decided apart. |
| Why do AI assistants get facts about your business wrong? | Answer | Answer Production Engine | 2026-08-23 | The draft treated this as a hallucination problem. Most cases are accurate reporting of a stale source, which points at an entirely different and more tractable remedy, so the page was rebuilt around locating and correcting the published record. Added the plain statement that no correction will be issued, since businesses otherwise spend effort seeking one. |
| Why do customers say they never heard back when you replied? | Answer | Answer Production Engine | 2026-08-23 | The draft treated this as a deliverability topic. Email is the largest cause and not the only one, and a page limited to it would have missed the message request folder and the unmonitored automated sender, both of which are common and cheaper to fix. Restructured into four causes with a single audit that covers all of them. |
| Why do quotes go unanswered? | Answer | Answer Production Engine | 2026-08-23 | The draft assumed price was the main cause and organised around objection handling. Reordered by what actually produces silence, which is undecidability, and made the recommended follow-up diagnostic rather than persuasive so that the three causes can be told apart. |
| Why do you rank locally but get no calls? | Answer | Answer Production Engine | 2026-08-23 | The draft treated this as a listing optimisation question. Testing whether calls are actually answered belongs first, because it produces evidence identical to a visibility failure and businesses spend months on listings while the phone is ringing out. Reordered the diagnostic so the free test comes before any change. |
| Why does your team not use the CRM? | Answer | Answer Production Engine | 2026-08-23 | The draft recommended training and enforcement. Both treat rational behaviour as a discipline problem and neither addresses the asymmetry that causes it, so the page now works from who bears the cost. Added the parallel-system reading, which converts a compliance annoyance into the most useful requirements document available. |
| Why is cost per click rising in my industry? | Answer | Answer Production Engine | 2026-08-23 | The draft attributed the rise entirely to competition. The click-supply contraction is at least as important and points to a different response, so it was added and given equal weight. Also moved intake repair ahead of the bidding discussion, since it is cheaper and improves channels the bid cannot touch. |
| Acquiring patients through the Maryland health care commission provider directory | Geography | Geo Verification Engine | 2026-08-22 | Verified by machine under ARCHITECTURE.md §2.2 rather than read by a person: every publishable fact on this page was checked against a source retrieved from mgaleg.maryland.gov, www.medchi.org, www.baltimorecity.gov before it was written, and every proposed field that failed its check was dropped rather than published. The record — the URL, the HTTP status and the sha256 of each document read, and the passage each check matched — is at engine/geo/verifications/healthcare/. |
| How specialty trade contractors identify and secure residential projects in Seattle | Geography | Geo Verification Engine | 2026-08-22 | Verified by machine under ARCHITECTURE.md §2.2 rather than read by a person: every publishable fact on this page was checked against a source retrieved from app.leg.wa.gov, www.seattle.gov before it was written, and every proposed field that failed its check was dropped rather than published. The record — the URL, the HTTP status and the sha256 of each document read, and the passage each check matched — is at engine/geo/verifications/home-services/. |
| How industrial manufacturers secure new business in the Houston metro area | Geography | Geo Verification Engine | 2026-08-22 | Verified by machine under ARCHITECTURE.md §2.2 rather than read by a person: every publishable fact on this page was checked against a source retrieved from comptroller.texas.gov, www.tceq.texas.gov, www.hccs.edu before it was written, and every proposed field that failed its check was dropped rather than published. The record — the URL, the HTTP status and the sha256 of each document read, and the passage each check matched — is at engine/geo/verifications/manufacturing/. |
| Acquiring real estate clients through local property data and association networks | Geography | Geo Verification Engine | 2026-08-22 | Verified by machine under ARCHITECTURE.md §2.2 rather than read by a person: every publishable fact on this page was checked against a source retrieved from www.scstatehouse.gov, charlestonrealtors.com, www.charlestoncounty.gov before it was written, and every proposed field that failed its check was dropped rather than published. The record — the URL, the HTTP status and the sha256 of each document read, and the passage each check matched — is at engine/geo/verifications/real-estate/. |
| How brokers acquire clients through local property and disclosure data in Phoenix | Geography | Geo Verification Engine | 2026-08-22 | Verified by machine under ARCHITECTURE.md §2.2 rather than read by a person: every publishable fact on this page was checked against a source retrieved from www.azleg.gov, armls.com, mcassessor.maricopa.gov and others before it was written, and every proposed field that failed its check was dropped rather than published. The record — the URL, the HTTP status and the sha256 of each document read, and the passage each check matched — is at engine/geo/verifications/real-estate/. |
| How Seattle real estate brokers acquire clients through local data and disclosure | Geography | Geo Verification Engine | 2026-08-22 | Verified by machine under ARCHITECTURE.md §2.2 rather than read by a person: every publishable fact on this page was checked against a source retrieved from app.leg.wa.gov, apps.leg.wa.gov, www.seattle.gov before it was written, and every proposed field that failed its check was dropped rather than published. The record — the URL, the HTTP status and the sha256 of each document read, and the passage each check matched — is at engine/geo/verifications/real-estate/. |
| How ambulatory health care providers acquire patients in the Phoenix metro area | Geography | Geo Verification Engine | 2026-08-22 | Verified by machine under ARCHITECTURE.md §2.2 rather than read by a person: every publishable fact on this page was checked against a source retrieved from www.azleg.gov, www.mcmsonline.com before it was written, and every proposed field that failed its check was dropped rather than published. The record — the URL, the HTTP status and the sha256 of each document read, and the passage each check matched — is at engine/geo/verifications/healthcare/. |
| What does an employee actually cost, over and above the wage? | Benchmark | Siddharth Sharma | 2026-08-21 | First publication, drafted from the engine reconciliation recon-employer-labour-cost-load-2026-08-21. Three calls made in drafting and worth naming. The audited values are reproduced exactly as the engine holds them, so two of them carry no percent sign, because the publisher printed those two as table cells under a percent column heading and in no sentence anywhere. The sample column for the three statutory rates says that there is no sample rather than that none was stated, because a rate set in law is not an estimate and reporting it as an undisclosed sample would be the wrong criticism. The small-establishment row states the whole private-industry sample and says plainly that the release does not break it down by band. The two open editorial decisions the reconciliation records are left open and printed on the page rather than resolved. Updated the same day: two of those open questions had been settled by the time the day ended, and a page that goes on calling a settled question open is a page nobody can trust about the ones still open. The statutory-share question was decided — a floor is computable for a named payroll and is not publishable as a share — and the multiplier trail was walked to its origin, which is now its own page. Both entries say what changed rather than being deleted. Nothing else moved: no row, no value, no definition, no verdict. |
| Where the loaded-labour multiplier actually comes from | Provenance | Siddharth Sharma | 2026-08-21 | First publication, from a trail walked on the day. Four calls made in drafting. The terminal source is classed as practitioner rather than academic: the author is a senior lecturer and the document is on his university's server, but it is a monthly advice column and not a study, and the class is about what a document can structurally support rather than about where it is hosted. The origin's date is recorded as undated rather than guessed at from its copyright line, which begins decades before the tax figures inside the article. The point figure appears only inside the audited columns, never in the prose, because the site's number gate forbids an unsourced numeral in copy and is right to — this is a figure the page is examining, not one it asserts. And the vendor page that attributes the range to the statistical agency appears once, in the trail, for what it attributes the figure to; its own worked-example point figure is described in the prose rather than repeated as a fourth circulating row, so no publisher is counted twice. |
| What does one employee cost you in federal payroll tax? | Calculator | Siddharth Sharma | 2026-08-21 | First publication, built on the three IRS statutory rates the research engine audited for reconciliation recon-employer-labour-cost-load-2026-08-21 and recorded in the migration 0049_two_documents_were_read_to_the_bottom. Three calls worth naming. The three rates enter the sourced data appendix as §18.7 — the first rows in that file to come through the evidence layer rather than from a reading of a published report — and each fact's value carries the base its rate stops at, because a constant that mirrors the rate while losing the base computes a wrong number out of two correct-looking halves. The instrument cites the Social Security topic page, which is the largest component and carries two of the three rates; the federal unemployment topic page is named on the page and in the assumptions rather than in the source line. And the last step states a share of the entered wage, which no publisher states and this page does not claim any publisher states — it exists only because a wage was supplied, which is exactly the open editorial decision the benchmark page records and this page does not settle. |
| What is a law firm's intake conversion rate? | Benchmark | Siddharth Sharma | 2026-08-08 | First publication. Rewrote the short answer to lead with the absence of a usable benchmark rather than with the range, since the range is the misleading part. Added the unanswered-enquiry row after deciding it belonged in the table as a constraint rather than in the prose as an aside. |
| Where the five-minute rule actually comes from | Provenance | Siddharth Sharma | 2026-08-08 | First publication. Moved the attribution finding ahead of the drift table, since the misattribution is the more common error. Removed an asserted sample size for the earlier study and recorded it as an open question instead, because the figure could not be confirmed against the study itself. |
| Is a fast automated reply better than a slower human one? | Answer | Siddharth Sharma | 2026-08-06 | Reworked the opening after the first draft read as a defence of automation in general, which is not what the question asks and not a position this site holds. Added the closing paragraph putting the automated reply back in proportion — it is the cheapest and most copyable part of a response system, and a page that oversells it undermines the argument for the rest. |
What the build refuses to publish.
Four checks run before any page in this system is generated, and each one stops the build rather than producing a warning nobody reads.
A page must name, in writing, what is on it that could not appear unchanged on a sibling page — and no two pages may give the same answer. Independently of that, every page is reduced to a fingerprint with its own subject and place names masked out; if two fingerprints match, or come close, both pages are named and the build stops. That is the check that catches a page whose only distinguishing feature was the city in the heading.
A page must also carry at least two datapoints that are specific to its own subject rather than to its category, and if the same figure is claimed as specific on two different pages, it was specific to neither. Pages ship in cohorts with a size cap so indexation can be observed between them, and every page carries the review block that produces the table above.
All four are properties of the schema, not of anyone remembering. Reviewed by Siddharth Sharma, except the 7 market page(s) above, which are checked against their sources and carry the record of it.
GATES
A gate enforced by a person reading pages stops being enforced somewhere around page 60.