Editorial Standards
How we decide what to publish, how we check it, how we handle uncertainty, and how we stay independent from the people who pay our bills.
1. Our editorial mission
EarthquakeTracker exists to help a general-public reader understand what is happening seismically around them and around the world, and to make the underlying scientific data legible without flattening the nuance. We are not an academic journal and we are not a disaster-response service; we are a well-edited reference and navigation layer on top of the public USGS earthquake record.
Every editorial decision on this site is guided by three principles, in order: accuracy over engagement, clarity over jargon, and honesty about uncertainty over confident over-reach. When those come into tension — and with rapidly developing earthquake data they sometimes do — we choose in that order.
2. Categories of content
We publish four distinct categories of content, and we apply different standards to each:
- Live scientific data — magnitude, depth, coordinates, ShakeMap intensity, felt-report counts, PAGER alerts, tsunami flags. We render these verbatim from USGS. We do not override, annotate, or reinterpret scientific values.
- Derived summaries — counts, distributions, aggregates, and rule-based event narratives. These are computed by us from the raw feed. Every derivation is described on the methodology page so readers can reproduce them.
- Educational content — articles in the Learn, Faults, History, Prepare, and Retrofit sections. These are written and edited by our team against sourcing standards described below.
- Navigational framing — hero paragraphs, section intros, grid headers, and other connective prose whose only job is to help a reader find something. These are held to factual accuracy but not to the sourcing bar required of educational content.
3. Sourcing standards
Educational content on this site is sourced from primary scientific and governmental records. Our reference hierarchy, in descending order of preference:
- USGS primary documentation and hazard reports: fact sheets, event-specific reports, the National Seismic Hazard Model, and peer-reviewed publications from USGS scientists.
- Peer-reviewed academic journals: Bulletin of the Seismological Society of America, Journal of Geophysical Research, Seismological Research Letters, Geophysical Research Letters. We cite specific papers when a claim depends on a recent research result.
- Authoritative institutions: IRIS, Seismological Society of America, FEMA, Red Cross, Ready.gov, California Earthquake Authority, and equivalent national bodies for international content.
- Established textbooks and reference works: for standard background material that is long-settled science (e.g., how P-waves propagate).
We do not use press releases as primary sources, we do not cite social media, and we do not rely on secondary news reporting for scientific facts. If a claim cannot be supported from the reference hierarchy above, we either leave it out or explicitly frame it as tentative.
We do not scrape any website. All data ingestion is through public APIs with documented terms of service, principally the USGS FDSN event web service. See the methodology page for the full ingestion description.
4. Fact-checking process
For static educational content (Learn, Faults, History, Prepare, Retrofit), each page is fact-checked against at least one primary source before publication. Claims involving specific numbers — historical magnitudes, death tolls, distances, rupture lengths, return periods — are traced to a specific sourced statement. Where sources disagree (e.g., different historical magnitude estimates for pre-instrumental earthquakes), we either choose the most recent USGS estimate and note the disagreement, or we report the range.
Live-data pages are fact-checked indirectly: the data itself comes from USGS, the layer we add (aggregates, counts, narratives) is computed algorithmically from that data, and the rules are published on the methodology page. If a reader spots a discrepancy between our page and the USGS source, we treat USGS as canonical and correct ours.
5. Handling of developing events
In the minutes and hours after a large earthquake, USGS magnitude and location solutions are refined. The first automatic magnitude can shift by 0.2 to 0.5 units and the epicenter by several kilometers before the final W-phase moment tensor stabilizes. We handle this in three ways:
- We always link to the canonical USGS event page so readers can see the latest official values regardless of our cache state.
- We mark events as "Auto" or "Reviewed" based on the USGS review status field, so readers know whether a seismologist has signed off on the current solution.
- We do not produce headlines or push notifications. This site is a reference; it is not a real-time alerting service. Headlines would pressure us to lock in a number early when we should instead be waiting for USGS to stabilize the solution.
For a genuine earthquake emergency, consult official warning channels such as USGS ShakeAlert (where deployed), the National Tsunami Warning Center, or your national meteorological or civil-defense agency. This site is not a replacement for those.
6. AI and automation disclosure
We disclose the use of algorithmic content generation on this site in the interest of transparency:
- Per-event prose narratives (the 6–9 sentence summaries that appear under each earthquake in location-page tables and on the /earthquakes/ hub) are generated by a deterministic, rule-based template system. The template selects sentence variants based on the event's own USGS fields (magnitude, depth, alert level, felt-report count, offshore/onshore heuristic, sequence context) and combines them with varied phrasings. This is not a language model; it is closer to a structured report generator. Every factual claim in these narratives traces to a USGS field or a simple derivation from USGS fields. The template code is version-controlled and reviewed by our team, and no large language model sits between the USGS data and the rendered page.
- Aggregate regional summaries (the "In the past 24 hours the USGS detected X earthquakes..." paragraphs on hub and state pages) are likewise generated by a structured template populated with live counts.
- Educational articles (Learn, Faults, History, Prepare, Retrofit) are human-written and human-edited. Any drafting assistance from generative AI is treated as a starting point only; the published text is reviewed, fact-checked, and rewritten by our editorial team before it appears on the site. We do not publish unedited AI output.
- We do not use AI to modify scientific data values. Magnitudes, depths, coordinates, and other USGS fields are displayed verbatim, never paraphrased or imputed.
If any of this changes materially — for example, if we introduce a language model into the event-narrative pipeline — we will update this section, note the change in the methodology changelog, and publicly date the change.
7. Sensationalism and framing
Earthquakes are sometimes catastrophic. We cover them that way when the facts warrant it — an M7.8 with a red PAGER alert is described as a red PAGER event — but we do not inflate the stakes of routine events for engagement. Specifically:
- We do not use language like "devastating", "monster", or "catastrophic" unless the underlying PAGER/ShakeMap/felt-report data supports it.
- We do not use "predict" or "due for" language. Short-term earthquake prediction is not scientifically possible, and implying otherwise misleads readers. See our earthquake prediction article for the forecasting-vs-prediction distinction.
- Our headlines and meta titles are factual and descriptive. Page titles include magnitudes and locations from live data, not editorial judgments about severity.
8. Independence from commercial interests
EarthquakeTracker is an independent publisher. Our revenue sources are limited to display advertising (served through reputable ad networks such as Google AdSense, with clear visual separation from editorial content) and, in the future, affiliate links on specific preparedness-gear guides (disclosed on those pages as required by the U.S. Federal Trade Commission). No advertiser, affiliate partner, or sponsor has any control over what we publish, and no advertiser or sponsor is shown content in advance of publication.
We do not accept payment for coverage, placement in listings, favorable framing of specific products, or any other form of influence on editorial content. Sponsored or affiliate-linked content, where it appears, is labeled as such at the point of encounter and is excluded from our regular editorial workflow.
9. Corrections, feedback, and contact
If you find an error, please tell us. Our full corrections policy — including how we mark corrected content and our public correction log — is at /corrections/. For immediate reports, email corrections@earthquaketracker.org with the page URL, the claim you believe is in error, and any supporting source. We aim to triage every correction report within two business days.
For general feedback or questions that aren't correction reports, see /contact/.
Related transparency pages
- Methodology — technical detail on data ingestion, caching, computation.
- Corrections policy — error handling and public correction log.
- Cookie policy · Privacy · Terms