Not a judgment on the decision
It does not say whether the analyzed choice is right or wrong.
Methodological and operational guide
Δ-CSI Manual
The complete guide to the Cognitive Sovereignty Index (delta): the index that measures how intense the adversarial challenge applied to a decision was. What it measures, how it is computed, which guarantees it offers — and which it does not.
- Calculation version
- csi-v1
- Editions
- Italiano · English
- Pages
- 44 (A4)
- Nature
- Self-sufficient on method and calculation: it can be read and verified with no other sources
What the Δ-CSI is not
Three misunderstandings cause nearly every misuse. The manual clears them before explaining any formula.
Not a prediction
It does not estimate the probability that the decision will succeed.
Not a grade for the decision-maker
It does not measure the decision-maker’s competence, nor the quality of the work done before the analysis.
What it measures
The Δ-CSI measures one thing only: the intensity of the challenge exerted by the adversarial process. Five signals feed into it — implicit assumptions (weighted by severity), falsification tests, burden-of-proof questions, cited sources, and the divergence between thesis and counter-scenario — on a 0–100 scale with three fixed bands: marginal, moderate, substantial.
A manual organized by role
Not everyone needs to read all of it. Part I is the common foundation; the other parts are added by role.
| Audience | Path |
|---|---|
| Anyone who must read a score | Part I (self-sufficient) |
| Operator / decision-maker | Part I + Part V |
| Investor / due diligence | Part I + Part III |
| Developer / engine | Part I + Part II + Part V |
| Academic / standards | Part I + Part II + Part III + Part IV |
What it contains
Understanding the Δ-CSI
The meaning of the number and its correct interpretation, with no technical prerequisites.
Technical construction
The calculation formula by formula: components, weights, the anti-inflation rule, divergence strategies, versioning.
Honesty and limits
The defenses against sycophantic output and the declared limits — including the ones still open.
Validity and validation
The methodological foundations and the evidence-accumulation protocol, with a worked calibration example.
Operating procedures
Everyday use in production: launching an analysis, reading a report, recording outcomes.
Reference apparatus
Glossary, three worked numerical examples, complete formula reference, edge cases, changelog.
Part III publicly declares what the metric does not guarantee — including the weaknesses still open. A contradictor that does not declare its own limits is not credible when it declares everyone else’s.