TRANSPARENT BY DESIGNMETHODOLOGY / DRAFT 01

The REALITAS standard

Question the claim.
Inspect the method.

A verdict is only as useful as the reasoning you can examine.

PROTOTYPE Six sample investigations across two fields. Research and scores are illustrative, not published findings.

What does the score mean?

The proposed REALITAS Score summarizes how strongly the available evidence supports a precisely defined claim. It is an editorial evidence-support index from 0 to 100.

It is not a percentage probability that a claim is true. A score of 23 does not mean a 23% chance. It is not a scientific measurement, a vote count, or a rating of a religion’s followers.

All numbers in this version are arbitrary design samples. They were selected to demonstrate the interface and have not been calculated from research. The provisional verdict labels demonstrate presentation only.

The investigation, in full.

  1. Define the claim. State the scope and what would count as support or a challenge.
  2. Establish context. Separate historical, empirical, textual, and philosophical questions.
  3. Build the strongest case for. Represent serious defenses in a form their proponents could recognize.
  4. Build the strongest case against. Test objections and alternatives with the same care.
  5. Evaluate the sources. Examine provenance, expertise, independence, relevance, and limitations.
  6. Explain the analysis. Make assumptions and unresolved questions visible.
  7. Publish a provisional assessment. Distinguish the support for a claim from confidence in the assessment.
  8. Say what could change it. State the evidence that would trigger reassessment.
  9. Record revisions. Explain what changed, when, and why.

Not every question fits a number.

A philosophical or moral claim may depend on premises that cannot be measured empirically. The future system must allow “Not scored” and qualitative analysis instead of manufacturing precision.

One method, many subjects.

REALITAS applies the same steps to religious, historical, scientific, technological, and social claims. Predictions about the future, such as warnings about artificial intelligence, need extra care: they rest on extrapolation and expert judgment rather than a completed record, and an assessment must say so.

One investigation, three languages.

Each investigation is one research record. Its evidence, sources, and score are shared across English, Portuguese, and Spanish; only the editorial text is translated. When a translation is not available yet, the original version is shown with a clear notice.

Before scores can be published

The research version needs an explicit rubric, rules for missing evidence, domain-specific criteria, documented weights, reviewer calibration, uncertainty reporting, and worked examples. Until then, no sample number should be treated as a conclusion.

Counter-evidence needs its direction defined: stronger counter-evidence should weaken support for a claim. The sample breakdown is a layout demonstration, not an average or a working formula.

Evidence and funding stay separate.

Advertisements and future affiliate recommendations will be labeled. Sponsors will not purchase verdicts. Community votes should surface useful contributions, not determine whether a claim is true.