About Deyeken
A curated registry of credibility analyses.
Deyeken is a credibility-first platform for analyzing digital content. A macOS application generates structured credibility analyses and publishes them here. The website is not a social network or a blog — it is a curated registry of app-generated analyses, authored by registered, identifiable users.
Its value is trust. Author identity and app-generated analysis — not free-form user input — are what make Deyeken different.
The name. “Deyeken” is broad Scots for “d’ye ken?” — “do you know?” From the Scots verb ken, to know. It’s the question the platform helps you answer: before you trust a piece of content, d’ye ken how credible it really is?
Why it matters
Trust collapsed. The tools didn’t keep up.
Public trust in media has fallen from roughly 72% in the 1970s to about 28% today, even as the cost of publishing dropped to near zero and newsrooms shed tens of thousands of fact-checking roles. The result is a flood of unverified content that no individual can navigate alone.
Existing tools rate whole publishers instead of individual articles, flag claims without verifying them, or hand down opaque true/false verdicts with no reasoning — and most miss cherry-picking, where every stated fact is true but the context that matters is left out. Deyeken is built to close that gap.
What it does
How Deyeken assesses credibility
Rather than a true/false label, Deyeken extracts every claim, separates fact from opinion, and gathers independent evidence on each — then explains its reasoning so you can interrogate and share it.
Claim-level analysis
Breaks an article into individual assertions and verifies each one independently against external evidence — catching unsupported statistics and inconsistencies that document-level scoring misses.
Cherry-picking detection
Flags content where every stated fact is true but critical context is left out. The Completeness axis weighs what is missing, not just what is present.
Fair, balanced & objective
Actively gathers both supporting and opposing evidence, and judges facts and opinions by different standards. No axis penalises a viewpoint — only the quality and completeness of the evidence behind it.
Transparent reasoning
Every score comes with a plain explanation and linked evidence, so you can see why — not just what — and check each source yourself.
Gets smarter over time
A shared credibility network: every analysis adds to a growing base of publisher and source assessments, so repeat content is faster and more consistent for everyone.
Works everywhere, one click
One analysis engine, built to run across browser, mobile, desktop, and a Web API — so analysing any content is a single click, wherever you are.
The Method
Three weighted axes
Every analysis is scored across three axes. The weights reflect what most determines whether an argument holds up to scrutiny.
20%
Source Credibility
Publisher authority, author expertise, and outlet track record — assessed before the analysis itself.
45%
Evidence Alignment
Every claim verified against external sources. Supporting and opposing evidence gathered; cherry-picking detected.
35%
Completeness
Critical context included. Omissions and gaps in the evidentiary record identified and flagged.
Reading a score
Bands, not decimals
Scores commit to a categorical band. We never report false precision with decimals, and we never hedge with words like "somewhat" or "fairly." The score describes the credibility of the evidentiary record — it is not a claim about absolute truth.
Performance
Minutes, not days — and it compounds
Most analyses finish in a few minutes for well under a dollar. A shared credibility cache means that once a publisher or source has been assessed, the result is reused across every analysis — so repeat content, the bulk of mainstream news, gets faster and cheaper for everyone over time.
~2–11 min
Per analysis
Shorter news articles are quicker; long technical documents take longer.
~$0.10–$0.85
Per analysis
Approximate — and expected to fall as the system is optimized.
85%+
Cache hit rate
Typical for mainstream news — reused source and publisher assessments cut the cost and time of repeat content.
Figures are approximate and vary with a document’s length and the depth of analysis chosen.
Who it’s for
One engine, many uses
A single analysis engine is built to serve every reader and team that needs to know whether content holds up — today through the macOS app and this registry, and expanding outward from there.
Readers
Check a news story, social post, or research paper before you trust it or share it — with the reasoning laid out so you stay the judge.
Publishers & newsrooms
Verify stories in editorial review before publication, and — by design — analyse once to show a credibility assessment to every reader.
Enterprise & education
Corporate due diligence, academic research platforms, and media-literacy tools for schools and universities.
See the analyses
Browse the registry, or create an account to download the macOS app.