Scoring Methodology
Why We Publish Our Methodology
Every story on Brightcast is selected by an algorithm, and we believe you should be able to see how that algorithm thinks. This page documents the Brightcast Intelligence Score (BIS) — the scoring system behind our curation — and how daily scores roll up into the Global Hope Index. For our editorial standards, attribution, and corrections process, see the Editorial Policy.
The Brightcast Intelligence Score (BIS)
BIS is a 0–100 score computed for every candidate article. It is the simple sum of three components, each assessed by an AI analysis of the full article text:
Hope (0–40 points)
The most heavily weighted component: does this story represent genuine, constructive progress? Hope is scored across four sub-factors: novelty (is this a new development, not recycled news?), scalability (can the progress it describes grow or be replicated?), emotional resonance (does it credibly inspire rather than manipulate?), and evidence (is the positive claim backed by concrete facts?).
Reach (0–30 points)
How many lives does this story touch, and for how long? Sub-factors: beneficiaries (how many people are affected), geographic scope (local to global), temporal durability (a one-day event vs. a lasting change), and ripple effects (second-order benefits beyond the immediate story).
Verification (0–30 points)
How trustworthy is the reporting? Sub-factors: sources cited (are claims attributed?), source tier (the credibility track record of the publisher), specificity (names, numbers, and dates vs. vague claims), and consensus (is the story corroborated by other reporting?).
Publication Gates
A high total score alone is not enough. Before publication, every article must also pass hard gates:
- Hope minimum: articles scoring below the minimum Hope threshold are rejected regardless of their Reach and Verification scores — credible and big is not the same as hopeful.
- Obituary and tragedy filter: death notices and memorial content are automatically rejected, even when they celebrate achievements.
- Consumer-content cap: product launches, deals, and promotional content are score-capped below the publication threshold so advertising cannot masquerade as good news.
- Duplicate detection: stories substantially identical to something we published in the previous 48 hours are rejected.
These gates are algorithmic and cannot be paid or petitioned around. The pipeline that applies them is described in How It Works.
From BIS to the Global Hope Index
The Global Hope Index aggregates the BIS scores of everything we publish into a daily measure of constructive news, broken down by category and by country. Because every underlying article is scored on the same 0–100 scale, the index is comparable day to day and across topics. You can explore it on the Hope Index pages or read the Weekly Hope Report.
Limitations & Honest Caveats
- BIS is an AI-assisted judgment, not an objective measurement. Two reasonable people (or two model runs) can disagree at the margins; the gates and sub-factor structure exist to keep that disagreement small.
- The index measures the positivity of reported news, which is not the same as the state of the world. Coverage gaps in source media propagate to us.
- Scores are calibrated over time. When we improve the scoring system, we version the change rather than silently rescoring history.
Use the Data
Hope Index trends, category insights, and anonymized reading-pattern datasets are free to download and reuse under CC BY 4.0 on our Open Data page. If you cite the Hope Index in research or reporting, attribute “Brightcast Global Hope Index” and link back so readers can inspect this methodology.
Questions
Methodology questions and critiques are welcome at [email protected].