Observatory methodology ยท v1

What the Observatory can and cannot say.

A transparent headline-level measurement layer for cognitive-pressure signals in public information environments. It does not diagnose authors, audiences or outlets with cognitive biases.

Unit of analysis

One headline plus source metadata. Article bodies are not copied or analyzed in v0.1.

Interpretation boundary

Report observable lexical and structural signals. Do not infer a cognitive bias, intent, truthfulness, political position or causal effect from one signal.

Sampling boundary

Every snapshot must state its provider, query, date window, language, sample size and sampling limitations. Curated pilot snapshots must never be presented as representative media samples.

Gain framing cue

Counts words that foreground opportunity, growth, improvement or creation.

Current lexicon: opportunity, opportunities, gain, gains, growth, grow, create, creates, creating, hiring, boost, benefit, benefits, bright, improve, improves, success

Loss framing cue

Counts words that foreground losses, cuts, decline, threat or damage.

Current lexicon: loss, losses, cut, cuts, layoff, layoffs, decline, declines, threat, threatens, risk, risks, grim, shortage, shortages, displace, displacement, redundancy, jobless, backlash

Uncertainty cue

Counts modal or questioning language that keeps a claim open rather than presenting it as settled.

Current lexicon: may, might, could, can, unclear, uncertain, whether, really, possible, possibly, appears, suggests, so far, outlook

Certainty cue

Counts language that presents an outcome or relationship with stronger certainty.

Current lexicon: will, must, proves, proven, definitely, certainly, inevitable, always, never, leads, causes

Salience cue

Counts vivid or high-intensity words that can make a headline more attention-grabbing.

Current lexicon: grim, bright, frantic, surge, crisis, massive, dramatic, race, fear, fears, shock, historic, record, backlash, scrappy

Authority cue

Counts references that foreground institutions, experts, studies, reports or named authority statements.

Current lexicon: says, said, study, studies, report, reports, research, researchers, expert, experts, poll, fed, government, minister, scientists

Social proof cue

Counts language emphasizing popularity, majorities, widespread behavior or crowd adoption.

Current lexicon: majority, most, many, millions, popular, viral, widespread, everyone, surge, racing, rush

Numerical emphasis

Counts explicit numerals, percentages, currencies and number-like tokens in the headline.

Rule: Regex-based count; no claim is made about whether the number is accurate or misleading.

Question headline

Marks headlines ending in a question mark. A question can signal uncertainty, curiosity or rhetorical framing, but the observable fact is only the form.

Rule: Boolean: trimmed headline ends with '?'.

Near-duplicate repetition

Groups headlines that share a high proportion of content words after stop-word removal.

Rule: Pairwise Jaccard similarity >= 0.62. This detects similar wording, not shared truth or independent confirmation.

Derived views

Source infrastructure

Back to Observatory