Speech is bidirectional, synchronous streams, so you can process it while it's being emitted. Text chat on the other hand commonly is asynchronous and packet-ba

Speech is bidirectional, synchronous streams, so you can process it while it's being emitted. Text chat on the other hand commonly is asynchronous and packet-based, i.e. you only get completely formed sentences. That means with text there's an idle period between the "hello" packet and the followup. Just seeing the greeting and having to decide whether to wait, respond or ignore it is an annoyance that does not exist in spoken conversation. It's not about the time itself, it's about the uncertain, possibly unbounded latency and the context-switching. In fact text is not just higher-latency, it also has a lower bandwidth. It's really important to use it efficiently. For a bit reduced latency and higher throughput they could implement liveposting as some imageboards do[0]. [0] https://imgur.com/HQKEwXE https://imgur.com/HQKEwXE