Corroboration — 20% weight

How many genuinely independent sources confirm the core claims? Sources that all derive from the same original report do not count as multiple corroborating sources.

1. Outlet identity of all cited / linked sources
2. Common ownership or editorial relationships between cited sources
3. Timeline of coverage across outlets (who published first, who cited whom)

Attribution — 18% weight

How traceable is the sourcing chain? Are claims attributed to named sources? Is the attribution chain visible or obscured?

1. Count of named vs. unnamed sources
2. Attribution chain depth (is the named source citing another unnamed source?)
3. Presence and accessibility of primary source links

Speculation Handling — 15% weight

What proportion of content is inferential, and how explicitly is speculation marked? Has speculation been upgraded to fact as the story traveled?

1. Hedging language density calibrated to claim consequence level
2. Explicit speculation marking in text
3. Claim upgrade detection (speculative in earlier versions, appears as fact in current)

Source Formation — 17% weight

What is the quality and independence of cited sources? Do they have structural conflicts of interest? Are they appropriately qualified for the claims they are cited on?

1. Independence assessment of cited sources
2. Conflict of interest detection
3. Source qualification relative to claim domain

Reader Formation — 12% weight

Does the content build or degrade the reader's evaluative capacity? Does it present multiple perspectives, or does it pre-digest conclusions?

1. Perspective diversity in framing
2. Evidence presentation vs. conclusion assertion ratio
3. Reader agency preservation in rhetorical structure

Engagement Distortion — 10% weight

Does framing exploit emotional reactivity rather than inform? Is the headline accurately representative of the body?

1. Headline vs. body claim alignment score
2. Emotional amplification language density relative to informational language density
3. Outrage-optimized framing pattern detection

Visibility — 8% weight

Are the conditions under which the content was produced transparent? Is funding disclosed? Are conflicts of interest visible?

1. Production conditions transparency
2. Funding disclosure presence and completeness
3. Conflict of interest visibility

Potential Capture

Definition

Structural conditions present that create possibility of capture.

Triggers

Ownership relationship, funding relationship, or personnel relationship creating conditions for capture.

Meaning

Does not mean source is unreliable — means conditions for unreliability are present.

Probable Capture

Definition

Behavioral patterns consistent with capture are observable.

Triggers

Pattern detection across a corpus: systematic absence of inconvenient information, directional skew in source selection, correction rate below baseline.

Meaning

Identified through pattern detection, not editorial judgment about individual pieces.

Verified Capture

Definition

Independently verified conflicts of interest documented and confirmed.

Triggers

Verifiable funding disclosure, documented coordination with covered party, formal correction record demonstrating systematic inaccuracy.

Meaning

Requires external verification — not inferred from pattern alone.

Browser Extension

Status: Live. Audience: Individual readers.

A transparency layer over the information environments you already inhabit. Three-layer progressive disclosure: ambient badge, deep dive panel, inline score map.

Lucent Thread Education

Status: Coming Soon. Audience: Students and teachers.

Converts scoring signals into Socratic learning prompts. Students develop the capacity to see what the system sees. Teachers get comprehension maps by dimension.

Source Index

Status: Live. Audience: Researchers, platforms, publishers.

A continuously updated database of source credibility trajectories. Not static ratings — behavioral trajectories derived from observed content patterns over time.

The Formation Problem

The information environment does not need to be full of lies to be harmful. It only needs to move faster than the reader's capacity to evaluate it. Speculative claims circulate before they can be verified. Corrections arrive late and travel less far. Sources with hidden conflicts of interest present themselves as independent.

The dominant responses — fact-checking organizations, platform moderation, political bias ratings — share a common flaw: they position themselves as authorities over truth rather than as tools that empower readers.

A Different Response

Lucent Thread does not tell readers what to think. It surfaces signals that allow readers to evaluate what they encounter. A transparency tool that renders verdicts commits the same formation failure it was designed to fix — it substitutes the tool's judgment for the reader's.

Lucent Thread surfaces measurements. The reader decides.

Gauges, Not Grades

Each dimension produces a score from 0.0 to 1.0. Scores are presented as gauge fill levels — not grades, not verdicts. A gauge implies a measurement of current conditions. A grade implies a judge who has rendered a verdict; a gauge implies an instrument reading the environment. The reader interprets. The tool measures.