Trust
Data methodology
Fractals turns scattered local information into structured, readable updates for 6 community sites. This page explains where that information comes from, how it is produced and labeled, what it can't do yet, and how to get something corrected.
Public sources
Most of what appears on Fractals community sites starts with information that is already public, including:
- Government meeting agendas, packets, minutes and recorded meeting video
- Official city, county, school and agency websites, notices and calendars
- Public datasets, such as weather, prices, court and jail records, and economic indicators, where a market has them
- Public posts and announcements from local organizations, businesses and agencies
- Submissions from residents, businesses and organizations, such as events, announcements and obituaries
Fractals does not own government records or other publishers' reporting. We link to them, summarize them and organize them.
Source provenance
Each story, event or data point is recorded alongside the source it came from, such as a meeting video, an agenda packet, a dataset or a public post. Where it's practical, we name or link that source in what we publish. Our curated Community Graph example shows what this looks like for one decision, with every relationship tied to a public record.
Timestamps and freshness
Stories and updates carry the date they were published. When an item describes something that happened earlier, like a meeting, we say when that event took place. Metrics describe the period they cover, and that period can lag behind today because many public datasets are released weekly, monthly or less often. We monitor sources for staleness internally, but a source can still go quiet without our noticing right away.
Automation and human review
Fractals relies heavily on automation, including AI, to collect sources, draft summaries, build charts and lay out print editions. We use verified sources, data, AI, and human judgment where needed to explain what changed.
Not everything is reviewed by a person before it publishes, and not every item is independently verified. Human review goes first to sensitive topics and to anything automation flags as uncertain. If something looks wrong, please tell us (see below).
Corrections
When we get something wrong, we want to fix it. Email [email protected] with the link, what's wrong and, if you can, a source. You can also use the contact page on the community site itself. Once we've confirmed an error, we correct the item.
Limitations
- Coverage varies by market. The number and type of sources and metrics depend on what each community publishes. Communities can track up to about 1,000 metrics, and many track fewer.
- Public records can be wrong or late. When a source is wrong, a summary of it can be wrong too.
- Automation makes mistakes. AI-assisted drafts can misread documents or video, which is one reason sensitive items get human review.
- The Community Graph is still being built. Cross-entity and cross-community queries are in development. Examples on this site that connect entities were curated by hand and are labeled that way.
- This is not legal, financial or official advice. For official records, use the issuing agency's own documents.
Privacy
We focus on public-interest information about public bodies, public decisions, organizations and community events. Names of private individuals generally appear only when they are already in a public record or a public announcement, or when someone submitted them. For how we handle data about visitors and subscribers, read our privacy policy. To ask about information concerning you, email [email protected].
Last updated October 2026.