Safety & Community Playbook
How we enforce a creative, respectful, and safe matching sanctuary.
1. Core Philosophy: Why Mythic Match Exists
Mythic Match is a dark fantasy, social, fun, and safe matchmaking experience. It is deliberately slow-paced, user interactions are limited, and all content is strictly moderated.
Anti-Tinder by Design
Unlike typical dating apps that focus on real-world photos, rapid swiping, validation loops, and hookup culture, Mythic Match is built as a creative sanctuary. It is a space where real people can connect through mythological personas, roleplay, and narrative resonance.
Because we target a wide age group, we enforce a strict zero-tolerance policy for:
- Sexual innuendos, hookup requests, or sexually explicit lore.
- Harassment, bullying, or stalking.
- Predatory behavior, grooming, or child exploitation of any kind.
- Bad actors, trolls, and perverts looking to abuse open chat features.
To enforce this, we have engineered out the primary vector for online harassment and grooming: unrestricted free-text private messaging.
2. Bounded Communication Protocols
To maintain a fun, respectful, and safe space, all communication in Mythic Match is strictly structured and symbolic. There is no way for a user to type and send free-text messages to another user.
A. Pre-Written Symbolic Cards
Once a match (or 'resonance') is established, entities can communicate only through a curated deck of pre-written cards. These cards are crafted in a high-fantasy, atmospheric style. They allow users to express interest, set boundaries, invite collaboration, or slow things down.
- “The mist remembers you.”
- “Our realms may align.”
- “A ritual alliance seems possible.”
- “The omens are favorable.”
- “Your storms do not frighten me.”
- “Your signal feels steady.”
- “I would share a quiet omen.”
- “May your path stay protected.”
- “I welcome a slow ritual.”
- “Your lore deserves patience.”
- “I honor your boundary.”
- “I will keep respectful distance.”
- “This signal thread can rest.”
B. Free-Form Emoji Messages
- The Rule: Emoji messages must contain only valid emoji characters (e.g., 🦄🌈🌈⭐🕧).
- The Filter: Any standard alphanumeric text, punctuation, or hidden characters will be rejected automatically by the client and server.
- The Benefit: Emojis allow for playful, non-verbal storytelling, emotional check-ins, and shared fantasy themes (like casting a spell using 🪄✨🔮) while making it impossible to transmit slurs, phone numbers, external social handles, or inappropriate text.
C. Card Reactions
Users can tap a sent card in their thread to toggle subtle, positive emoji reactions (e.g., 👍, ❤️, 😮, 😢) to show acknowledgement or delight without starting a new conversation thread.
3. The AI Moderation Engine
Every text input in Mythic Match goes through a layered safety pipeline powered by a fail-closed implementation of Google's Gemini AI.
A. Safety Thresholds
Our Gemini prompts are configured with zero-tolerance safety settings across four primary harm categories:
- Harassment: Block insults, aggressive language, and targeted intimidation.
- Hate Speech: Block derogatory statements, slurs, and discriminatory language.
- Sexually Explicit: Block physical/sexual descriptors, innuendos, and references to hookups or dating apps.
- Dangerous Content: Block self-harm, violence, doxxing, and illegal activities.
B. Fail-Closed Design
In production, the configuration parameter MODERATION_FAIL_OPEN is set to 0. This means if the Gemini API is down, times out, or encounters a rate limit, the profile creation or lore update fails closed. No unmoderated text is ever written to the database.
C. Custom Taxonomy Moderation
When players enter custom realms (e.g., 'Nebula Sprite') or custom affinities, these strings are saved to a separate queue for review by app operators. They do not appear in the global dropdown lists for other players until they have been audited, preventing offensive words from leaking into the shared taxonomy.
4. Reporting & Operator Protocols
If an entity generates a profile that bypasses automated safety checks (e.g., highly stylized roleplay that borders on harassment or includes hidden patterns), players have immediate recourse.
A. Reporting a Profile
From the Discover deck, players can report any candidate. The reporting UI requires selecting a reason:
- Harassment or Bullying
- Sexually Explicit / Dating Focus
- Real-World PII (Names, Phone Numbers, Social Media links)
- Inappropriate generated image (Borders, watermarks, text, or multiple faces)
- Off-topic / Non-fantasy theme (Trolling)
B. Report Containment & Deck Advancement
- Immediate Block: Submitting a report automatically hides the reported creature from the reporter's deck forever.
- Deck Progression: The swipe pool is rebuilt instantly, advancing the queue so the user is never stuck looking at the reported profile.
- Operator Review Queue: Reports are logged directly to the operator dashboard, where the admin can view the creature's full profile details and take remediation actions.
C. Actionable Operator Remedies
- Quarantine: Hide the offending profile from all Discover views instantly.
- Regenerate Lore/Avatar: Force-overwrite bad inputs with default placeholder cards if the player entered inappropriate text that bypassed the model.
- Purge: Permanently delete the profile and record its Creature ID hash to prevent re-registration under the same profile footprint.