I think it depends on the type of content. I'm totally with you for content that should be text (e.g. most things related to programming), but some content lends itself to the video form factor and doesn't work as plaintext content.

For example, imagine CinemaSins or Movies with Mikey as plaintext. That would take all of the fun out of it.

Or imagine reading transcripts of Stephen Colbert's monologues. Those would probably be quite tedious since a lot of Colbert's quality is his pantomimic object work.