Methodology

How we evaluate sources and track stories (without giving away the secret sauce)

Our Core Principles

1. Track record matters more than brand. We care about what sources do, not just what they're called.

2. Time is truth's filter. Stories that hold up over hours and days are more reliable than breaking reports.

3. Multiple sources validate claims. One source could be wrong; ten sources rarely are.

4. Behavior tells the truth. Corrections, retractions, and original reporting reveal credibility.

How We Evaluate Sources

Track Record Over Time

We watch how sources behave across hundreds or thousands of articles. Do they break stories first? Do they issue corrections when wrong? Do they retract false claims?

This behavioral data builds a pattern over time. A source with a consistent track record earns trust, regardless of brand recognition.

Multiple Factors Combined

No single metric determines trust. We look at multiple signals:

• How often they publish original reporting vs. copying others
• How frequently they need to issue corrections
• Whether they break stories before competitors
• If they've published stories that were later proven false
• How long they've been operating with good behavior

We combine these signals mathematically, but the exact formula is proprietary. The key is: behavior patterns matter more than any single article.

Brand Reputation Still Counts

We're not naive. Established outlets like Reuters or The New York Times have earned their reputation through decades of work. But brand isn't everything—it's one factor among many. A well-behaved newer source can rise; a careless established source can fall.

How We Track Stories

Continuous Monitoring

We check news sources every 15 minutes, 24/7. When articles are published, we timestamp them precisely. This lets us see who published first and how stories spread over time.

Automatic Grouping

Articles about the same event get grouped together automatically based on content similarity—not just keywords. This prevents pollution where unrelated stories share similar words.

Time-Based Trust

We adjust confidence based on story maturity:

↓ Breaking news (first few hours) gets lower confidence—facts are still emerging
→ Developing stories (hours old, multiple sources) get moderate confidence
↑ Confirmed stories (day+ old, many sources) get high confidence

Multi-Source Validation

A story reported by one source is less reliable than one reported by ten independent sources. We track source count and diversity, adjusting confidence accordingly. Even a highly-trusted source publishing alone gets flagged as "unconfirmed" until others verify.

Detecting False Stories

Pattern Recognition

Stories that are likely false or unimportant follow predictable patterns:

• Quick abandonment: Multiple sources published it, then stopped covering it within hours
• Low-tier only: Only unreliable sources covered it; established outlets ignored it
• Retractions: Some sources that published later retracted or heavily corrected
• Original removed: The first publisher deleted their article

When we detect these patterns, we flag the story as potentially false or "dead"—worth viewing with extreme skepticism.

What We Don't Do

❌ We don't use AI to "fact-check." We track patterns and let you judge.
❌ We don't claim to know absolute truth. We show you the evidence and let you decide.
❌ We don't censor sources. Even sources with poor track records are shown, just with appropriate warnings.
❌ We don't share exact formulas. But we're transparent about the principles behind them.

Our Limitations

Being honest about what we can't do:

•
New sources need time. Sources with limited publishing history don't have reliable patterns yet.
•
Exclusive scoops get flagged. When a trusted source breaks a story alone, we initially mark it "unconfirmed" until others verify—even if it's true.
•
Coordinated falsehoods can slip through. If many sources publish the same false claim, consensus isn't always truth.
•
Coverage is selective. We monitor around 900 active RSS feeds from established news outlets worldwide. We prioritize quality and consistency over breadth.

Why Not Share Exact Formulas?

Two reasons:

1. Gaming the system. If we publish exact formulas, bad actors can optimize their behavior to appear trustworthy while still misleading readers. Keeping some details private makes manipulation harder.

2. Competitive edge. We've spent significant time developing these methods. While we're committed to transparency about principles, the specific implementation is part of what makes TruthTime unique.

That said, we're exploring open-sourcing our algorithms in the future, once the system is mature and we've figured out how to balance transparency with manipulation-resistance.

Our Transparency Commitment

We're committed to being as transparent as possible without enabling manipulation. If you have questions about our methodology, we're happy to discuss principles and approach.