What is evidence?

This repository serves as a public evidence archive supporting a written report that alleges a specific Discord server, run by a content creator, permits and promotes hate speech and other harmful behavior among its members. The maintainer states they collected around forty two thousand messages from the server, spanning multiple channels over roughly a year and a half, and used a mix of keyword searches and AI assisted review to flag messages that appeared to violate common standards around hate speech, followed by a manual check of each flagged item and its surrounding context before including it in the report. The stated purpose of the repository is to let others verify the claims independently rather than take the accompanying report on faith. It includes the underlying dataset of flagged messages along with a system the author describes as tamper detection, which is meant to log and publicly display any attempt to later delete or alter the archived messages, and a companion website where that live log can be viewed. The project is written in TypeScript and does not appear to be a general purpose tool or library. It is a single purpose accountability project built to document and preserve records of alleged online conduct in a way that can be checked by outside parties. The author acknowledges some limitations up front, including that not every flagged message could be independently verified and that some content was left out of the public report for that reason, while noting that leaving something out should not be read as evidence against the claims that were included.