Methodology and trust notes

This MVP uses structured demo data and prewritten plain-language summaries to model how a source-first civic-information product could work. In a live version, summaries would be generated from public records, then reviewed against source citations and editorial rules.

Summaries are designed to describe what is documented, not to predict outcomes or recommend a vote.

Each candidate, measure, and evidence block includes attached source objects. In this demo those sources point to local text files that stand in for public records like filings, hearing notes, and budget documents.

The interface keeps source counts visible and offers drawers or side panels so users can inspect source context without losing their place.

Ballot Clarity is informational, not advisory. The product should not tell users who to support, rank candidates, or hide tradeoffs behind a single score.

The compare page is intentionally tabular and neutral. It shows differences in record and funding without turning elections into a leaderboard.

A future alignment module could compare district concerns with public records and stated positions. It would need transparent methodology, clear weighting rules, and strong explanations of uncertainty.

That feature is not live in this MVP. It is labeled as experimental wherever it appears so users do not mistake it for a finished recommendation system.

Public data is incomplete, uneven across race types, and often delayed. Challengers usually have fewer public records than incumbents, and local races can depend heavily on meeting minutes or questionnaires.

This MVP makes those limits explicit through demo labels, “What we know” and “What we do not know” sections, and repeated reminders to consult original records.