On Aug 16, 2013, at 6:39 PM, John C Klensin <john-ietf@xxxxxxx> wrote: > IIR, we've tried audio input. It works really well for > conference-sized meetings (e.g., a dozen or two dozen people > around a table) with a few remote participants. It works really > well for a larger group (50 or 100 or more) and one or two > remote participants. I've even co-chaired IETF WG meetings > remotely that way (with a lot of help and sympathy from the > other co-chair or someone else taking an in-room leadership > role). > > But, try it for several remote participants and a large room > full of people, allow for audio delays in both directions, and > about the last thing one needs is a bunch of disembodied voices > coming out of the in-room audio system at times that are not > really coordinated with what is going on in the room. Now it > can all certainly be made to work: it takes a bit of > coordination on a chat (or equivalent) channel, requests to get > in or out of the queue that are monitored from within the room, > and someone managing those queues along with the mic lines. Yup, it definitely takes those things. Been there, done that, got the IETF t-shirt. :) I think we might be able to do it, using the jabber scribes for those "coordination" actions. Maybe. It depends on the number of remote active participants and quality of scribes. The jabber scribes would have to act like the operator-assisting person in big conferences with remote participants. (the old "we've got a question from Jane Doe, go ahead Jane" type thing) For the WGs I go to (RAI area mostly), we have good scribes and not a large number of remote people who actually participate (as opposed to monitor). We've had some exceptions, but my impression is the things the remote people wanted to say in those cases were usually said by someone locally anyway so they're more of a +1 thing. I.e., if there are lots of local attendees, you usually get someone saying what you were going to say anyway. Not that someone remote shouldn't say it as well, because it does matter if you hear the same thing being repeated. But at least it's not so much "interaction" needed for hearing that. But yes if there are a dozen remote active participants, and a 100 people locally in the room, it's chaos. It's not chaos because the remote participants don't get mic time - it's chaos because they *do* get mic time. The delay in letting them know it's their turn at the mic, delay in real-time interaction, the mental switch to "remote mode" for local participants, etc., all cause the meeting to slow down... a lot. It's like multiple processes running on one CPU - context switching is painful. We can try to pile up the remote participants to go all at once, so that there're fewer context switches. That's what folks do in big conferences: the remote participants are queued up until the local ones have finished, and then the remote ones go all at once. Unfortunately that turns it into a Q&A type thing at the end, and not a discussion, but with that big an active audience that's probably all it could be anyway. > But, by that point, many of the disadvantage of audio input > relative to someone reading from Jabber have disappeared and the > other potential problems with audio input -- noise, level > setting, people who are hard to understand even if they are in > the room, and so on-- start to dominate. Yes, audio quality and volume control and a bunch of related things are very important for this to work. IANAE on that - there are professionals who do that stuff for a living. > Would I prefer audio > input to typing into Jabber under the right conditions? Sure, > in part because, while I type faster than average it still isn't > fast enough to compensate for the various delays. But it really > isn't a panacea for any of the significant problems. OK, so what are the significant problems? What have the WGs you've been participating in not been doing, that makes you feel like you don't get to participate remotely? -hadriel