Verification methodology

Every fact returned by DeepInquiry passes not just corroboration checks but source-independence checks that trace each corroborating source back to distinct evidence — not distinct URLs.

The problem: two sources aren’t always two sources

The industry minimum for “verified” is two independent sources (see the IFCN Code of Principles and Full Fact). Almost nobody defines what “independent” means. That gap creates four well-documented failure modes:

  1. Citogenesis. A false claim added to Wikipedia gets copied by multiple news outlets, each citing the others as “independent.” The Ronnie Hazlehurst case had the BBC, The Guardian, The Times, The Independent, The Stage, and Reuters all copy a fake edit within days.
  2. Wire-service duplication. A single AP or Reuters story is republished by 200+ outlets. Different hosts, same underlying evidence.
  3. Aggregator confusion. Statista and Encyclopedia.com republish government or vendor data. Citing them is really citing whoever they pulled from.
  4. Ownership clustering. WSJ, MarketWatch, and Barron’s are all News Corp. USA Today Network runs 260+ local papers. Same tank, different logo.

Our approach

DeepInquiry treats source counts as a starting point, never as the answer. Every fact in our corpus is graded through a multi-layer verification policy that asks a stricter question than “how many outlets said this?” It asks how much independent evidence stands behind it.

1. Cluster before counting

Every candidate source is grouped by the underlying evidence it traces back to — wire-service origin, publisher ownership, and text similarity of the specific claim. The pipeline then works with clusters, not URLs. Two hundred outlets re-running one AP story is one cluster, not two hundred.

2. Exclude cascade sources

Wikipedia, its known mirrors, and content aggregators are removed from the corroborating set whenever a non-mirror source is available. Language patterns that indicate a page derived its claim from another are flagged as cascade candidates. Later publications that are near-identical to an earlier one are treated as copies, not as independent confirmation — the fingerprint of a citogenesis cascade.

3. Score structural independence

Every surviving cluster pair is evaluated on multiple axes of independence, including ownership, source category (primary vs peer-reviewed vs trade vs other), and geography, and weighted by a prior that favors primary and peer-reviewed sources. A fact is only as independent as its two most-similar remaining pieces of evidence.

4. Escalate exceptional claims

Borrowed from Wikipedia’s WP:V: “any exceptional claim requires multiple high-quality sources.” Every claim is classified at extraction time:

  • Everyday — well-known, non-contested. Standard independence bar applies.
  • Notable — surprising or counter-intuitive. Higher independence bar plus at least one primary source.
  • Exceptional — challenges consensus, involves a public figure, or is time-sensitive. Highest independence bar, primary source required, no citogenesis flags allowed.

5. Grade the output

The API doesn’t hand you a boolean. Every fact ships with a verification tier, an independence score, the number of independent evidence clusters behind it, any citogenesis flags, and the claim tier. You get to decide the bar for your product.

TierWhat it meansAvailable on
GoldMultiple independent clusters, strong independence score, at least one primary or peer-reviewed source, no citogenesis flags.All plans
SilverIndependently corroborated by multiple clusters at the standard independence bar.All plans (default floor)
BronzeCorroborated but with weaker independence — useful for research, discovery, and low-stakes use.Pro and above (opt-in via ?min_tier=bronze)
UnverifiedBelow the Bronze bar. Exposed only for audit and research purposes.Scale and above (audit mode via ?include_unverified=true)

What this policy does not solve

We publish limitations because a fact-verification product that claims to catch everything is one you shouldn’t trust.

  • The primary-only case. A Fed statement, a court ruling, or an SEC filing has one authoritative source by construction. These are treated as authoritative on their own and can reach Gold with a single cluster.
  • Sustained multi-outlet fabrication. The Alan MacMasters toaster hoax was accepted by the BBC, The Scotsman, the UK government, and 12+ books over a decade with no traceable Wikipedia citation. Source independence is a necessary condition for truth, not a sufficient one. The tier is Gold, not Truth.
  • Everyone cites the same primary study. Two news outlets citing the same Nature paper are one primary source, and our pipeline treats them as one cluster. Independent replication of an underlying study is a stronger standard we track separately for a future release.

What we don’t publish here

We publish enough for you to evaluate the policy, cite it in your own product, and hold us accountable to it. We do not publish exact thresholds, cluster keys, the scoring formula, or our proprietary source and mirror lists — the pieces a competitor would need to clone the corpus rather than build alongside it. Enterprise customers with a signed NDA can request a deeper technical brief; email api@deepinquiry.ai.

Sources this policy draws from