vibe matcher

a social network where the matching algorithm tries to find people you'd actually like — not people who look similar on paper (same school, same job, same interests) but people whose energy, humor, communication style, and values are genuinely compatible with yours. the insight is that most social platforms match on surface features: LinkedIn matches by industry, dating apps by location + photos, Twitter by topic. "vibe" is the thing that determines whether you actually enjoy talking to someone, and it's almost entirely absent from current matching.

the ML angle is interesting because vibes are implicit in behavior, not explicit in profiles. how you write, what you respond to enthusiastically, what you ignore, how you structure sentences — these are all signals. the system would build a profile from natural behavior (messages, reactions, post patterns) and use that to find people with compatible signatures. the obvious concern is that this kind of matching can create bubbles — only connecting you with people who already think like you. the counter-design would intentionally include some "interesting adjacent" matches: people who have a compatible vibe but different context or knowledge, which tends to produce generative conversations.