01
Score contract
A number is a model
Every future score must expose variables, source population, transformations, weights, missingness, sensitivity, uncertainty, and version.
- Data dictionary
- Normalization
- Weights
- Missing-data policy
Published facts remain sourced; every unfinished model, scenario, or recommendation is explicitly labeled.
How facts and examples are separated ↗BlackMaxxing Scores
Models you can inspect.
A unified architecture for City, Career, Product, Employer, and College ROI scoring with versioned inputs and visible uncertainty.
Every score remains provisional and unpublished. The interfaces show intended methodology and genuine inputs where available—not rankings.
02
Personalization
One ranking cannot fit everyone
Users may adjust priorities while the system distinguishes personal preference from the base evidence model.
- Base evidence
- User priorities
- Sensitivity view
- No sponsor influence
Authenticity contract
Structure can be complete before claims are real.
Where genuine evidence exists, the site publishes it with source, vintage, identifier, and limitation. Where a marketplace, review, recommendation, partnership, price, benefit, event, or score lacks real-world support, the interface remains a clearly labeled example.
Audit genuine datasets ↗Read the commercial firewall ↗