# Relationship Text Analysis Methodology ## Grey Mirror Full-Thread Evidence Framework **Version:** 2026.08 **Document Type:** Technical Methodology **Purpose:** Explain how Grey Mirror transforms message exports into relationship analysis reports --- ## Executive Summary Grey Mirror is a full-thread relationship text analysis system that processes exported message histories to identify observable communication patterns. This document describes the methodology, evidence standards, and technical approach used to transform raw message data into relationship insights. ### Key Principles 1. **Full-Thread Analysis**: Preserve message sequence, timestamps, and participant attribution 2. **Evidence-Linked Claims**: Every important finding connects to supporting message windows 3. **Confidence Bands**: Uncertainty is disclosed rather than hidden 4. **Privacy-First**: Account-scoped processing with published retention and deletion policies --- ## 1. Supported Exports ### 1.1 Supported Export Formats Grey Mirror accepts message exports from common platforms: - **iMessage**: `.txt` export from Mac Messages app - **WhatsApp**: `.txt` export (UTF-8, tab-delimited) - **Facebook Messenger**: `.json` or message history export - **Telegram**: `.json` export - **Signal**: Plain text export - **SMS/Android**: `.txt` or CSV export from Google Messages - **Generic formats**: CSV, JSON, HTML with message structures ### 1.2 Reading the Export Grey Mirror reads the complete export as one ordered history. Message order, timing and who sent what are preserved, and malformed or damaged rows are flagged rather than silently dropped. When it is not clear who is who, Grey Mirror asks you to confirm before any finding is attributed to either person. ### 1.3 Data Integrity Import handling follows the data-integrity principles of NIST mobile forensics guidelines (SP 800-101 Rev. 1). See [NIST Guidelines](https://www.nist.gov/itl/sed/nist-special-publication-800-101-rev-1) for detailed principles. --- ## 2. Analysis Metrics ### 2.1 Core Metric Categories Grey Mirror measures observable communication patterns across these dimensions: #### Timing & Response - **Response-time drift**: Change in reply speed over time - **Silence gaps**: Duration of non-response periods - **Turn-taking**: Message exchange rhythm #### Effort & Reciprocity - **Initiation balance**: Who starts conversations - **Response rate**: Questions answered, statements acknowledged - **Planning language**: Future-oriented statements #### Conflict & Repair - **Repair attempts**: Frequency and quality of repair messages - **Repair success rate**: Whether repair attempts land - **Conflict escalation**: How arguments intensify or de-escalate #### Emotional Patterns - **Emotional reciprocity**: Vulnerability met with similar depth - **Positivity reciprocity**: Warmth returned in kind - **Emotional tone drift**: Changes in warmth over time #### Structural Patterns - **Effort balance**: Initiations, follow-ups, questions, plans - **Topic stability**: Whether conversation focus changes - **Loop detection**: Recurring communication patterns ### 2.2 What Each Metric Shows Each metric compares both people, shows how the pattern moved across the history, and is read against the pair's own baseline rather than an outside norm. A confidence label reflects how much evidence supports it. ### 2.3 Evidence Linking Important findings include: - **Message-level evidence**: Specific line references (when privacy allows) - **Window-level evidence**: Time ranges where patterns appear - **Metric-level evidence**: Counts, deltas, and confidence - **Cross-metric agreement**: Whether multiple signals align --- ## 3. Confidence & Evidence Standards ### 3.1 Confidence Bands Grey Mirror uses confidence labels to communicate reliability: | Label | Meaning | User Interpretation | |-------|---------|---------------------| | High confidence | Multiple evidence sources agree | Useful signal, still not a verdict | | Moderate confidence | Some support but context incomplete | Read with caveats | | Low confidence | Signal is thin, ambiguous, or limited by the export | Do not over-weight this claim | | Insufficient evidence | Report cannot support the claim | Treat as absent, not negative | ### 3.2 Evidence Quality Factors Confidence is reduced when: - Timestamps are missing or unreliable - Participant mapping is ambiguous - Export has visible gaps or deleted messages - Sample size is too small for reliable patterns - Content is heavily redacted or censored ### 3.3 Legal Evidence Alignment Our evidence approach aligns with Federal Rules of Evidence principles: - **Rule 401** (Relevance): Evidence must relate to the pattern being analyzed - **Rule 402** (Admissibility): Evidence must be reliable and relevant - **Rule 403** (Exclusion): Prejudicial interpretations are avoided See [Federal Rules of Evidence](https://www.law.cornell.edu/rules/fre) for detailed legal standards. --- ## 4. How Patterns Are Read ### 4.1 Patterns Across Time Long histories are compared against themselves: how the relationship looked early on, how it developed, and where it stands now. That comparison is what reveals trajectory: are patterns changing over time, and when did the change begin? ### 4.2 Pattern Detection Patterns are identified through: 1. **Recurrence**: Does the behavior repeat across windows? 2. **Timing**: Is the pattern consistent in time and context? 3. **Participant asymmetry**: Is one person driving the pattern? 4. **Cross-metric agreement**: Do multiple signals support the same conclusion? ### 4.3 Context-Aware Interpretation The same pattern can mean different things in different relationship contexts: - **Romantic**: Commitment, affection, intimacy signals - **Platonic**: Mutual interest, boundary respect, shared activities - **Family**: Obligation, care, support dynamics - **Professional**: Collaboration, respect, task completion --- ## 5. Limitations & Safety ### 5.1 What Grey Mirror Cannot Know - Private intent or feelings not expressed in text - Off-platform events (in-person conversations, phone calls) - Deleted messages or content never in the export - Sarcasm, private jokes, or cultural context - Clinical diagnoses or mental health conditions - Legal facts or admissibility determinations ### 5.2 Out-of-Scope Uses Grey Mirror should NOT be used for: - Clinical diagnosis or mental health labeling - Legal conclusions or evidence handling advice - Emergency response or abuse-safety planning - Proof of love, cheating, intent, or compatibility - Stalking, harassment, or coercive decision-making ### 5.3 Bias & Error Sources Potential error sources include: - Parser mistakes in reading export formats - Wrong participant mapping - Language ambiguity or cultural nuance - Sentiment misreads in context - Uneven data quality across platforms - Metric thresholds that don't fit specific contexts --- ## 6. Privacy & Data Handling ### 6.1 Processing Scope - **Account-scoped**: Processing happens within user account boundaries - **Job-scoped**: Each analysis is isolated to a specific upload - **Participant-scoped**: Report access respects participant consent designations ### 6.2 Retention & Deletion - **Published policy**: Retention periods are publicly documented - **User control**: Users can request deletion of jobs and reports - **No training**: Private messages are not used for model training ### 6.3 Access Controls - **Preview access**: Completely free analysis of your full relationship timeline, with evidence-linked findings from your own conversation. No credit card required. Full reports are optional. Four evidence-linked findings are included. - **Full access**: Free analysis includes four evidence-linked findings, with no card required. A complete report costs $17.99 USD once, or uses one credit from Grey Mirror Plus at $9.99 USD/month. Each paid month provides one credit. Unused credits carry forward; unlocked reports remain available after cancellation. - **Participant access**: Optional sharing with specific participants --- ## 7. Relationship Science Foundations ### 7.1 Academic Influences Grey Mirror's metrics draw from established relationship science research: - **Gottman Institute**: Research on repair attempts, conflict escalation, and communication patterns - **Digital communication studies**: Research on texting patterns and relationship quality - **Attachment theory**: Research on anxious/avoidant patterns in communication See [Gottman Institute Research](https://www.gottman.com/about/research/couples/) for foundational relationship science. ### 7.2 Digital Forensics Principles Evidence preservation follows digital forensics best practices: - Chain of custody from export to analysis - Data integrity verification - Participant attribution tracking - Timestamp preservation and normalization --- ## 8. Capacity - **History length**: built for complete histories, including exports with hundreds of thousands of messages - **Time**: most reports finish within minutes; very large histories take longer rather than being thinned - **Import quality**: results depend on how completely the source app exports the conversation - **Coverage**: some metrics need enough history before they can be reported reliably --- ## 9. Version History ### Version 2026.08 - Added methodology foundations section with authoritative references - Enhanced evidence standards with legal evidence alignment - Improved confidence band documentation - Added relationship science research citations ### Version 2026.05 - Initial public methodology release - Core metric definitions - Evidence linking framework - Privacy and deletion controls --- ## 10. References ### Technical Standards - NIST SP 800-101 Rev. 1: Guidelines on Mobile Device Forensics - NIST SP 800-86: Guide to Integrating Forensic Techniques into Incident Response - Federal Rules of Evidence Rules 401, 402, 403 ### Relationship Science - Gottman Institute Research on Couples and Relationship Patterns - Research on attachment styles in digital communication - Studies on texting patterns and relationship quality ### Digital Communication - Academic research on response timing in text messaging - Studies on emotional reciprocity in digital relationships - Research on conflict resolution via text --- **Document Status:** Public **Classification:** Technical Methodology **Last Updated:** August 2026 **Next Review:** February 2027 For questions about this methodology, contact: methodology@justlay.me ## Current pricing Start with a completely free analysis of your entire text history, with no card required. Your free preview includes 4 evidence-linked findings. An optional full report is $17.99 USD once: your complete relationship story across the full timeline, cinematic narrative, Memory Lane, evidence, metrics, custom charts, and available results from Grey Mirror's 30-analyzer inventory. Grey Mirror Plus is $9.99/month for 1 full report credit each paid month. Unused credits carry forward; keep unlocked reports permanently. Authoritative pricing: https://thegreymirror.com/grey-mirror/relationship-report-pricing