Fog drifting across a hillside of dense conifer forest with treetops dissolving into low cloud

Standards

Editorial Standards and Corrections Policy

This is the editorial standards and corrections policy for RationalSurvivor.com. It sets out where our numbers come from, how a draft is checked before it publishes, how often published pages are re-read against their sources, and exactly what happens when we get something wrong. The short version: a claim appears here only if it can be traced to a published agency guideline or to a documented use of the thing itself, and a named human has checked it. Where the evidence is thin, the page says so instead of rounding up to confidence. Rational Survivor is a new publication. We have no back catalogue of corrections to show you and no track record to point at, so this page describes what we promise and how you can hold us to it. If we fall short of it, [tell us](/contact/).

Sourcing Rules: Where Every Claim Comes From

Every factual sentence on this site belongs to one of three categories, and the page makes it obvious which one you are reading.

  • Published guidance. A figure issued by a public agency or a recognised standards body — water planning volumes, boil times, disinfection ratios, food-safety windows, resuscitation rates, carbon monoxide distances. We name the body in the sentence and link to the specific page that carries the number, not to a homepage.
  • Something we used. A result that comes from our own hands: a stove run down to empty, a filter pushed past its rated volume, a shelter built in rain. We say what we did, in what conditions, for how long, and what failed.
  • Judgement. A recommendation that the evidence does not settle — how many spare batteries is sensible, whether a second stove is worth the weight. We label it as judgement and show the reasoning so you can disagree with it.

There is no fourth category. You will not find "studies show" without a study named, a survey percentage with no survey behind it, or a confident number standing in for an honest range.

These are the sources we treat as primary and return to at every review:

Where a manufacturer specification is the only available source — a rated flow rate, a duty cycle, a temperature range — we attribute it to the manufacturer and treat it as a claim, not a measurement.

Agency-Cited Numbers vs. Numbers We Measure

This distinction matters more than any other rule on the page, so we enforce it in the grammar of the sentence itself.

An agency-cited number reads like this: "The EPA directs you to boil water for one minute, or three minutes above 6,500 ft (2,000 m), to make it microbiologically safe." The body is named, the link goes to the document, and the figure is copied from that document at the moment of writing — never from memory, and never from an older page of our own.

A number we measured reads like this: "We ran the stove at a stated ambient temperature, with a stated fuel and pot, until it stopped, and recorded the time." It carries its conditions in the same paragraph, and it carries its sample size even when that sample size is one.

The rules that follow from the split:

  • We never attribute an agency figure to our own testing, and we never dress a single field result as settled consensus.
  • A test result does not overrule a safety guideline. If our experience disagrees with published guidance on something that can hurt you, we publish both, say they disagree, and tell you to follow the guideline.
  • We do not publish a measured figure without the conditions that produced it. A runtime without a temperature, a flow rate without a water source, or a burn time without a fuel is not a result — it is a number with no meaning attached.
  • Arithmetic done on the page — water volumes, watt-hours, calorie targets, fuel burn — is shown in full so you can substitute your household's figures. See how much water to store per person and backup power for a home outage for what that looks like in practice.

Our testing protocol — what we log, how long we run things, and how failures are recorded — is written up separately in How We Test.

How a Draft Becomes a Published Guide

Every guide moves through the same sequence. A draft that fails a stage goes back; it does not publish with a note promising to fix it later.

  • Define the failure. We start from the specific thing that goes wrong — the tap runs dry, the freezer thaws, the fire will not take in rain — not from a product category.
  • Source pass. Every number in the draft is matched to a link before any prose is polished. A figure with no source is either rewritten qualitatively or cut. This is the only hard stop in the process, and it is absolute for anything safety-critical.
  • Arithmetic check. All calculations are re-run independently of the draft, including unit conversions. US-primary with metric in parentheses on first use, checked both ways.
  • Use it. Where the guide describes gear or technique, the gear gets used and the technique gets practised, in conditions we record. Where we have not done that yet, the guide says which parts rest on specifications rather than use.
  • Safety review. Anything medical, water, food-preservation or fuel related goes through the separate check described below.
  • Drill and pass/fail. Guides carry a practice drill with a standard you can actually fail, because that is the editorial line here: skills beat stockpiles. A drill without a pass mark is not a drill. See how to start a fire in wet conditions for the format.
  • Sign-off. A human editor reads the finished page against its sources and is accountable for every claim in it. Publication does not happen any other way.

The page then goes live with a published date and a review date, and enters the cadence below.

The Plain-Language Safety Review

Four subjects get a second, separate pass before publication and at every scheduled review: medical, water treatment, food preservation and fuel or combustion. These are the topics where a wrong number does real damage, so the review is a checklist, not a feeling.

  • Every figure is re-read from the source that day. Not from memory, not from our own earlier article. That includes boil times — one minute, three minutes above 6,500 ft (2,000 m) — chemical disinfection ratios, which we quote from the EPA's table by bleach concentration rather than as a single remembered drop count, and food-safety windows such as the four hours a closed refrigerator holds food safe, roughly 48 hours for a full freezer and 24 for a half-full one, per FoodSafety.gov.
  • Exclusions sit in the same paragraph as the instruction. Boiling does not remove chemical contamination. A filter rated for bacteria and protozoa does not remove viruses. Bleach loses strength with age and scented or thickened products are not suitable. If the limit is not next to the method, the method is misleading.
  • The failure mode is stated, with the safe direction of error. What happens if you get it wrong, and which way to err when you are unsure.
  • Fuel and combustion instructions carry the distance and the ventilation rule. Generators, camp stoves and charcoal stay outdoors and well away from windows, doors and vents — at least 20 ft (6 m) — and never run indoors or in a garage, per Ready.gov and CDC guidance.
  • Medical content carries a stop line. Every medical page says when to stop improvising and get professional help, and none of it substitutes for hands-on training. Rates and ratios such as chest compressions at 100–120 per minute are quoted from current resuscitation guidance, and our wilderness first aid kit page points you at training rather than pretending reading is enough.
  • It has to be usable when you are cold, tired and holding a headlamp. One action per line, short sentences, measures given in things you own — a teaspoon, a gallon (3.8 L), a wristwatch.
  • If we cannot source an exact figure, we describe it qualitatively. "Until it has been at a rolling boil long enough to meet published guidance" is a worse sentence and a better instruction than a guessed number.

Where guidance says the job needs a licensed professional — gas work, permanent generator wiring, prescription medication — the guide says so and does not offer a workaround.

Review and Update Cadence

Published pages are not finished pages. Each one carries two dates: published, which never changes, and last reviewed, which changes only when a human has actually re-read the page against its sources.

  • Safety and agency-dependent pages — water treatment, food safety and preservation, first aid, carbon monoxide and fuel — are reviewed at least every six months, and immediately whenever we learn that a source document has been revised.
  • Gear pages are reviewed at least every 12 months, and sooner when a product is revised, renamed or discontinued, or when long-term use turns up a failure the original test did not.
  • Skills and planning pages are reviewed at least every 24 months, or whenever the underlying guidance changes. Pages like the 72-hour emergency kit and the 30-day pantry sit on agency guidance, so they follow the six-month cycle for their cited figures even when the surrounding advice is stable.

What the dates mean, precisely:

  • We do not bump a review date to look fresh. If nothing was re-checked, the date does not move. Date-stamping for search rankings is a lie told to readers, and it defeats the purpose of publishing dates at all.
  • A review that changes advice produces a dated note on the page, exactly as a correction does.
  • A review that changes nothing still updates the review date, because "we checked this in the last six months and it still matches the source" is useful information.
  • Pages we no longer stand behind are marked as retired at the top, with the reason, and left in place rather than quietly deleted. A deleted page cannot be corrected, and anyone who acted on it deserves to find out why it is gone.

How to Submit a Correction

If something on this site is wrong, tell us. Corrections take priority over every other message in the inbox.

Email hello@rationalsurvivor.com with "Correction" at the front of the subject line. Include:

  • The page URL, not just the title
  • The exact sentence or figure, quoted
  • What you believe it should say
  • Your source, with a link, document name or page number where one exists
  • Your conditions, if the correction comes from experience — temperature, altitude, water source, fuel, duration, exact model

What happens next:

  • Acknowledgement within two working days, including when the answer is "we are re-checking this." You will not be left in silence while we read.
  • Safety-critical claims are acted on first. If we confirm that a medical, water, food-preservation or fuel figure is wrong, the page is corrected or the passage pulled within 24 hours of confirming it. If verification is going to take longer than that, we remove or hedge the disputed line while we work.
  • Substantive factual errors — a wrong specification, a misread source, bad arithmetic, a superseded guideline — are corrected within five working days, with a dated note.
  • Typographical slips and dead links are fixed silently and continuously.
  • Credit, if you want it. Say in your email whether you would like to be credited by name or initials in the correction note. The default is no name.

We check corrections against the primary source rather than against our own previous article. If the underlying guideline, specification or measurement has changed, the guideline wins and our page changes.

What we cannot act on: a claim that contradicts a cited source without offering a better one, and a request to remove an unfavourable test result. If you disagree with our conclusion rather than our facts, write anyway — but say which it is. The full inbox routing is on the contact page.

How Corrections Are Logged and Dated

Substantive corrections are never made silently. When a page changes, the change is visible on the page.

A correction note is appended at the foot of the affected page and carries five things:

  • The date the correction was made, in full, not "recently"
  • What the page said before, quoted, so you can recognise the version you read
  • What it says now
  • What prompted the change — a reader, a revised agency document, our own re-check, a manufacturer's updated specification
  • Whether the correction changes what you should do. This is the important line. If someone following the old version would have stored too little water, mixed a treatment wrongly, trusted a piece of gear further than it deserved, or kept food past a safe window, the note says that plainly rather than calling it a clarification.

We also keep a site-wide correction log so you can see our record in one place rather than hunting through pages. Because Rational Survivor is new, that log is currently empty. We would rather show you an empty log than invent a history of diligence, and it will fill up — every publication that checks its own work eventually corrects something.

Two commitments that go with it:

  • We do not delete a page to make a correction go away, and we do not edit a substantive number without leaving the note.
  • If a correction reaches us and we decide the original stands, we will tell you why, and name the source we are relying on. Disagreement is a legitimate outcome; silence is not.

Our Policy on AI Tools

We use AI tools, and we would rather tell you exactly how than pretend otherwise.

What they are used for:

  • Drafting outlines and restructuring sections
  • Tightening prose that a human wrote
  • Suggesting candidate sources, which a human then opens, reads and verifies
  • Cross-checking consistency — flagging where two pages state different figures for the same thing, or where a unit conversion looks wrong
  • Transcribing and tidying field notes taken in bad weather

What they are never used for:

  • Generating a fact, a figure, a citation or a safety instruction
  • Producing test results, or any sentence describing something we did
  • Writing a page that then publishes without a human reading it against its sources
  • Creating images that could be mistaken for documentary photographs of our own testing. Illustrative artwork, if used, is labelled as illustration.

The accountability rule, stated plainly: a named human editor is responsible for every claim published on this site. Not a model, not a pipeline, not a process. If a number on this site is wrong, a person checked it and that person was wrong, and the correction note reflects that. "The AI wrote it" will never appear on this site as an explanation, because it is not one.

A language model's output is treated exactly like an anonymous tip: possibly useful, never publishable on its own authority, and worthless until traced to the primary sources listed above.

Conflicts of Interest, Money and Gear

How we are funded changes what you should believe, so here is the whole of it.

  • Some links earn commission. Where they do, it is disclosed on the page, and the affiliate disclosure explains the mechanics. A commission never changes a rating, a ranking, or whether a failure gets written up.
  • We do not sell coverage. No sponsored posts, no paid reviews, no paid placement in a buying guide, no paid guest links. Please do not send a rate card.
  • We do not accept payment to remove a finding. Not a negative review, not a failure report, not a correction that a brand dislikes.
  • Samples buy testing, not coverage. A manufacturer sending a product gets a test, or nothing. There is no promise of publication, of a date, or of a rating. Where a sample is retained rather than returned, the page says so.
  • Contributors declare financial interests in the category they are writing about — products sold, training offered, consulting, sponsorship, equity — and those interests are stated on the page. A contributor who sells a product in a category does not write the comparison that ranks it.
  • No brand sees a review before publication. We will check a factual specification with a manufacturer; we do not send drafts for approval.
  • We do not take part in paid commentary, and we will not supply a quote in support of a product or campaign.

The simplest test we apply: if a reader learning how a page was funded would feel misled, the page is wrong, regardless of whether a rule was technically broken. Our gear testing section runs under these rules without exception.

What We Will Not Publish — and How to Escalate

Some things are ruled out at the outline stage, before sourcing even begins.

  • Fear as a motivator. No countdowns, no implied collapse, no urgency language. The realistic hazards — a multi-day outage, a storm or flood, an evacuation order, a water advisory, an injury far from help — are serious enough described accurately.
  • Invented authority. No fabricated credentials, no staff who do not exist, no reader numbers, no awards, no "years of experience" we have not had.
  • Shopping lists dressed as preparedness. A guide that only tells you what to buy has not taught you anything. Skills beat stockpiles, and a page that cannot carry a drill with a pass mark usually should not exist. Start Here sequences the skills that return the most capability for the least money.
  • Advice for scenarios we cannot describe, date or measure.
  • Individual medical, legal or engineering direction. We are a publication, not your clinician or your electrician. See the disclaimer, and treat this site as a supplement to training, never a replacement for it. In an active emergency, follow the National Weather Service and your local authorities rather than any website.

If you think we have broken one of the rules on this page, email hello@rationalsurvivor.com with "Editorial" in the subject line and point at the specific rule and the specific page. We will answer in writing, and we will either fix the page with a dated note or explain why we think it stands. If our answer does not satisfy you, say so and it gets read again by someone who was not part of the original decision.

This policy is itself reviewed every 12 months, and changes to it are dated like any other. Related reading: How We Test, About Rational Survivor, and the full guide index.