The format of DConf Online was somewhat problematic from the beginning. I thought people would be used to it by now---watch the talk, then come back to the stream---but apparently the urge to watch both the stream and the talk at the same time is pretty strong. I suppose if I were a viewer, I'd be the same way. So I'm going to remove the temptation next year :-)
My thinking behind the current format was: the speaker has to be available to answer questions in the video's chat box anyway, and that's a poor format for quality answers, so why not record them doing it and give them a chance to verbally expand with more detailed answers? And then I figured I'd fill the gaps when no questions were coming in. Then the viewers could come into the stream with more questions after the talk, and after the conference, I'd release it as a separate Q & A video. (Had I been more versed in the operation of OBS Studio at the time, I might have taken a different approach.)
I'd like to continue releasing Q & A videos after the conference. One option would be to take the Jitsi instance out of the live stream when the talk starts. Then I can talk to the speaker as I have been doing, but the viewers won't have to make a choice.
I'm not sure that's the way to go, though. The thing is, even though I'd prefer everyone go watch the talk while I have my little discussion, the interaction with the stream viewers is part of the fun. Plus it helps keep the discussion going. I would miss that if I took the Jitsi out of the stream.
Adam suggested showing the talk in the stream. I wasn't keen on that at first, but I've since warmed up to it. And now I know how we're doing things next time.
Next time, everything will happen in the livestream. For each talk:
- I'll chat with the speaker for 10 or 15 minutes before the talk, with viewers free to chime in
- I'll switch from the Jitsi to the talk, and we'll watch it together in the stream
- For any on-topic questions (about the talk) during the talk, I'll pause it and let the speaker answer verbally, then I'll resume the talk
- After the talk, we'll keep the speaker around for another 15 or 20 minutes for more Q & A and discussion.
This way, viewers don't have to choose and I can still have my Q & A videos after the conference.
Furthermore, I won't have to schedule the talks as premieres. This way, the schedule won't be limited to YouTube's restrictions (premieres can only be scheduled at :00, :15, :30, and :45 in any given hour). I don't even have to make them public. I can keep them private until after, maybe even release one each day after the conference.
As for when the next time will be, I'm thinking we'll delay until January or February 2024. My choice of November for the first two conferences was based on DConf happening in May every year. Then this year it happened in August. I didn't like the idea of skipping '22 for DConf Online, hence December. But really, that was just too close. We hadn't wrapped up the post-DConf admin before I was already working on the online version. Given that DConf '23 (assuming it happens) is likely to be in August, it makes sense to push DConf Online back a couple months, even if it means we have to skip a number.
So anyway, I wanted to post this here to say thanks to those who provided feedback and suggestions. I had time to mull it over today and I like where it led me.