What sources does Iron Leaks use?

Four tiers, in order: position stands and guidelines from named bodies such as the ISSN and ACSM, peer-reviewed trials and meta-analyses on PubMed, primary data sources like USDA FoodData Central, and practitioner judgment clearly labeled as opinion.

Does Iron Leaks cite other fitness blogs?

Never. Other blogs, content mills, forum posts and AI-generated summaries sit outside the source hierarchy entirely and cannot support any claim on the site, including background claims that look uncontroversial.

What happens if a number cannot be verified?

The sentence is cut rather than hedged. A figure reaches publication only if it traces to a primary source that was opened and read, with the citation placed inline next to the claim it supports.

How often are articles reviewed?

Priority pages are re-checked on roughly a 30-day cadence and every article carries a visible last-reviewed date. A review re-opens the cited sources and checks whether newer research has changed the recommendation.

How do I report a correction?

Email editor@cernovichleaks.com with the page URL, the exact sentence, and the source that contradicts it. Corrections are made in the article with the change noted, never edited out silently.

How We Research: Sources, Reviews and Corrections

The short answer

Iron Leaks ranks sources in four tiers: position stands from named bodies, then peer-reviewed trials and meta-analyses, then primary data sets like USDA FoodData Central, then practitioner opinion labeled as opinion. Other blogs are never cited. Any number that cannot be traced to a source that was actually read gets cut.

Every guide on this site links here, because the method is the product. An article is only worth the sourcing behind it, and most fitness content has none: in a review of flagship guides at the largest US training sites, several carried zero citations of any kind. Iron Leaks runs 5 to 7 inline citations per 1,000 words, primary sources first.

What counts as a source

Sources are ranked, and the rank decides what an article is allowed to say. A position stand can support a recommendation. A single trial can support a finding. A food database can support a number on a label. Practitioner experience can support an approach, and only when it is labeled as one person’s judgment.

The Iron Leaks source hierarchy, strongest tier first
Tier	Source type	Can support	Cannot support
1	Position stands and guidelines from named bodies: ISSN, ACSM, health.gov Physical Activity Guidelines	General recommendations and target ranges stated as consensus	Claims about an individual reader, or anything outside the population studied
2	Peer-reviewed trials and meta-analyses indexed on PubMed or PubMed Central	Specific effect sizes, dose-response relationships, direction of an effect	Certainty from one small trial, or a result generalized past its subjects
3	Primary data sources: USDA FoodData Central, manufacturer spec sheets, product labels	Nutrient amounts, weights, dimensions, load ratings, listed figures	Any claim about a physiological outcome
4	Practitioner sources, including the editor’s own coaching judgment	Programming choices and sequencing, always labeled as opinion	Anything presented as a research finding or a fact
None	Other blogs, content mills, forum posts, AI-generated summaries	Nothing	Nothing. These are never cited

How does a number get into an article?

A number reaches publication only if it traces to a primary source that was fetched and read in full. Not a secondhand summary, not an abstract skimmed for a figure, not a number remembered from somewhere. The source is opened, the sentence containing the value is located, and the citation goes inline next to the claim.

If that trace fails, the sentence is cut. Not softened. Not hedged into vagueness. Cut. A guide that says “studies show” without a link is doing the same work as a guide that made the number up, and the reader has no way to tell them apart. The named-attribution style used across the site exists for the same reason: the ISSN position stand puts daily protein for exercising people at 1.4 to 2.0 g/kg of body weight is a checkable sentence, and “experts recommend plenty of protein” is not.

Ranges are reported as ranges. Where a source gives 1.4 to 2.0 g/kg, the article does not collapse it into one tidy figure; the width of the range is itself information.

How is uncertainty handled?

Where the evidence does not support a confident answer, the article says so and explains what is missing. That is a deliberate trade: a page admitting a gap is less satisfying to read and more useful to act on.

Population limits are stated explicitly. The PROT-AGE position paper on protein intake, for example, studied adults over 65, so an article citing it says who it was written for rather than quietly applying it to a 45-year-old. A meta-analysis of trained lifters does not automatically describe untrained ones.

How often is content reviewed?

Priority pages are re-checked on roughly a 30-day cadence, and every article carries a visible “Last reviewed” date near the top. The review re-opens the cited sources, confirms the linked URLs still resolve, and checks whether a newer position stand or meta-analysis has changed the picture.

The date is honest. It moves when the article was genuinely re-read, not on a schedule to look fresh.

How do corrections work?

Corrections are made in the article, with the change noted, not silently. If a figure was wrong, the article says what it previously said and why it changed. Quietly deleting a bad number is worse than the original error, because it hides that the page was ever unreliable.

To report one, email [email protected] with three things: the URL, the exact sentence, and the source that contradicts it. That third item is what makes a correction fast. Corrections get the fastest response of any message the site receives; the contact page lists what else the inbox accepts.

What Iron Leaks will not do

It will not invent a statistic, a study, a quote or a rating. It will not invent a personal anecdote, a client, or a test that was not run. First-hand here means a paper that was read or a spec sheet that was checked, and nothing else. It will not give medical advice, diagnose anything, or write a rehab plan for an injury.

It will also not take money from any company whose products it might discuss. There are no affiliate links, no ads and no sponsored placements anywhere on the site, as covered on the about page. That is not a moral posture. It is what makes this page enforceable, because a source hierarchy is worthless on a site with a commercial reason to ignore it.

How We Research