1. People over profiles
No global identities, no profile pages, no follower graphs. Every room is a fresh start. You arrive with a temporary pseudonym, contribute, and leave. The platform should remember conversations, not personas.
2. Anonymous, but not careless
Anonymity is here to reduce fear and status games, not to excuse harm. The product design focuses on small-scale rooms, slower pacing, and clear room lifecycles instead of viral escalation.
3. No engagement treadmill
There is no infinite feed, no algorithmic ranking, no advertising. Rooms are short, topics are explicit, and when a conversation cools down, it ends. You can close the tab without the sense that something crucial is happening elsewhere.
4. Visualising disagreement
Disagreement is not a bug. The consensus map is deliberately designed to show tension and divergence instead of hiding it. The goal is not to force agreement, but to make it easier to see where people stand and where curiosity still exists.
5. Small, composable technology
The implementation follows the same philosophy as the product: small pieces, clear responsibilities, everything testable and replaceable. Node.js controllers, Vue 3 over CDN, MongoDB for persistence, files kept short on purpose. Complexity is a cost, not a badge of honour.
6. Bots as slow companions
AI bots are introduced carefully and late. They speak rarely, explain themselves, and act more like attentive regulars than content factories. Their job is to keep rooms readable and alive for newcomers, not to win the conversation.
7. Start small, stay honest
The early sprints are intentionally narrow: a single room flow, a single consensus map, a handful of bots. The goal is to discover whether this way of talking to strangers is actually useful, not to maximise metrics.