Conversations are better than a random subset of unending text-message-chat, for several reasons:
We evolved for conversations, and not for text conversations.
In practice, there seems to be some curse laying over text conversations that makes them more neurotic and less productive.
The audience deserves the chance to obtain nonverbal cues (especially of timing of confidence). These cues are often just as informative as the verbal information itself (especially in the case of, for example, “lies”).
A YouTube conversations is a discrete “lump” of a thing, where as text chat has no true beginning or end.
A discrete “lump” of a thing can have an independent, fixed context. If random text is quoted, people won’t know the context.
“Signing” the conversation with your unique voice/appearance encourages people to be on their best behavior.
In a conversation, only one person can speak at a time, and everyone’s attention is fixed on this one person. In text messaging, people will retreat into their own cerebral worlds, and the conversation can often “fork”.