...it's that I (are anybody else doing this for that matter) need a director. Somebody not in front of the camera, who can point things out objectively. My wife is supportive, which doesn't help in improving quality. Colleagues are either really enthusiastic about the concept and tend not to notice the video itself, or think it's a waste of time for "real scientists".
Thinking about pros first. You can use visuality to emphasise a point and show graphs. If you happen to be a good performer you could score a lot of points that way and get a more diverse audience. A lot of people watch Youtube but aren't too interested in Neuropsychiatr Dis Treat.
The cons: you can't skim. You can read/skim a lot of abstracts in the time it takes to watch one video. Also, you loose a lot of credibility if your scientific abstract comes off as a bad home movie.
I agree with your points. Especially the final one is a real hazard. You come across a lot of bad videos, which damage the author/presenter. But then again: so do presentations at conferences (man, do I have bad stories of those) and posters. For example: people who are disappointed they didn't get an oral presentation and just print out the hand outs and stick it to the poster wall. That just won't do. Even if people put in the time and effort, they can be so bad as to be laughable.
The YouTube is just mine, the journal doesn't post there. They only publish it on their own website, so the audience is limited, but targeted.