About & Methodology
How CockroachVision works, how evidence is evaluated, and how to appeal.
Mission
CockroachVision is an evidence-based transparency database that tracks coordinated influence campaigns, propaganda networks, and astroturfing operations across social media platforms. Our goal is not to accuse, but to document and provide verifiable evidence so that researchers, journalists, and the public can make their own informed judgments.
How Accounts Are Evaluated
Accounts are flagged and categorized based on observable patterns:
- Suspicious Timing: Multiple accounts posting identical or near-identical content within a short window.
- Hashtag Coordination: Unusual hashtag overlap across accounts that otherwise appear unrelated.
- Content Repetition: Copy-paste or templated content reused across accounts.
- Account Age: Newly created accounts that immediately engage in coordinated activity.
- Cross-Platform Activity: Same content or patterns appearing across different platforms simultaneously.
- Engagement Patterns: Suspicious likes, shares, and comments from networks of accounts.
Status Classifications
Account has been flagged and is being investigated. No conclusion has been reached.
Sufficient evidence has been gathered for analysis. Awaiting final classification.
Multiple independent evidence points confirm coordinated behavior. Evidence is published.
Investigation found insufficient evidence of coordinated activity. Account is cleared.
Evidence Requirements
Every claim on this platform must be backed by at least one of the following:
- Archived screenshots of posts or profiles
- Links to archived content (archive.is, archive.org, etc.)
- Timeline data showing coordinated posting patterns
- Cross-platform activity logs
- Hashtag coordination analysis
Appeal Process
Account owners who believe they have been incorrectly classified may submit an appeal by emailing our team with evidence supporting their claim. Appeals are reviewed within 14 days. If the evidence warrants, the account status will be updated or corrected.
Correction Policy
We are committed to accuracy. If an error is identified — whether through our own review or through an external submission — we will correct the record promptly. Corrections are noted on the relevant account page with a timestamp and explanation.
Neutral Language
We avoid loaded labels such as “government IT cell” or “paid propagandist” unless we have direct, verifiable evidence. Instead, we use neutral, evidence-based classifications: Coordinated Activity Suspected, Under Investigation, High Similarity Network, Evidence Pending Review. This keeps the project focused on transparency rather than accusations.
Privacy
This platform only tracks publicly available information from social media profiles. We do not collect or store private data. Reporter identities are kept confidential. IP addresses of reporters are stored temporarily to prevent abuse and are not shared.