Editorial
What We Know is an independent project created by Theo Koorehpaz. It uses automated discovery, source collection, and AI-assisted synthesis to produce concise, source-linked briefings that help readers compare reporting around the same story.
What We Know creates original, source-linked reporting summaries that help readers compare how reputable outlets describe the same story. We link readers to original reporting, minimise quotation, and correct or remove material when warranted.
How stories are made
The pipeline identifies a candidate, gathers related reporting, removes obvious duplicates and noise, then produces a core account from the captured material. It separately compares each useful source with that account. The public meter groups those comparisons as Corroborates, Adds context, or Conflicts.
What the comparison means
Corroborates means a source materially supports a central part of the core account. Adds context means it contributes relevant detail, framing, uncertainty, or a distinct angle without materially changing that account. Conflicts means it directly contradicts or materially qualifies a central claim. These labels describe the relationship between the included reporting and the briefing; they are not a measure of objective truth, editorial quality, or political position.
What we show
Each story presents What We Know, Why It Matters, and What Remains Unclear alongside the source comparison and links to the available original reporting. We use short excerpts for context and do not intend the service to replace the original work of publishers.
Source use and rights
We aim to use only the material needed to identify, compare, and link to reporting: source names, links, factual references, and short attributed excerpts where context requires them. We do not publish full articles, seek to bypass paywalls or access controls, or present third-party reporting as our own.
Publisher names and links are used to identify the reporting reviewed. Their presence does not imply a partnership, endorsement, or affiliation. Rights in third-party material remain with their respective owners.
Limits
Automation can miss important reporting, misread nuance, or reproduce errors in source material. A group of sources may not be independent, and source counts are not a guarantee of accuracy. Readers should consult the linked sources, especially where a story is developing or a claim matters to them.
Conflicts and corrections
We aim to surface meaningful differences without manufacturing disagreement for balance. Report factual errors, missing material context, attribution concerns, or copyright concerns through the corrections process. Where a concern is substantiated, we may correct, clarify, update, archive, or remove material.