This is one of the things that bugs me the most. It makes it really difficult to have a shared conversation, such as going around the room answering a question.
It's because none of the video chat apps are competing on features, only on user lockdown. Their KPIs is how many users they have, and that only needs enough features/price to convince the corporate buyer which rarely accepts feedback from larger set of employees.
As such, outside of certain older very expensive corporate videoconferencing setups that remember targeting inter-corporate meetings and the like, the actual feature set ossified. You get some things that are easy to demo in 1:1 call, you get locked down room systems that leave me weeping for Webex ffs, and you get omnipresent WebRTC in-browser setup that often murders your CPU unless you get lucky.
I find it really bad that the best video conference experience I ever had involved Cisco Webex hardware and Bluejeans, all interconnected with so-maligned SIP. And before that, the glorious days of wider ITU-T based video conferencing, like NetMeeting with its whiteboards and active screensharing.
Imagine if video conferencing solutions competed on top of common protocols, and you'd separately invest in infrastructure and clients? Instead of the rare enterprise cases where Lync/Teams/Webex/other SIP is set on dedicated lines in large corporation, make similar setups common everywhere (WebRTC/SIP bridges/transcoders on the edge, with dedicated bandwidth bough by company, etc.?) with vendors competing on something more than "you have to use our stack because we don't interoperate with anyone else".
Sincerely, someone who has had enough keeping 3-5 video conferencing apps around.
Yep, it's pretty abysmal for class settings.
I know it's partly down to configuration/instructors but being shoved into a room by an instructor with 40 other people forced to stare into a webcam while watching everyone else doing the same is not an ideal environment.
There needs to be a lot more attention on how and when to select additional participants in the general case where there's only one or two priority speakers and the rest are basically just sitting around listening.
Now of course me personally I'd just rather not use a webcam at all (Or only when specifically called on), but often that's not a choice unfortunately.
But I 100% agree with you that this is a design problem to be solved first with better interfaces and workflows, not AI. Innovation is the sum of behavior change and AI is not going to be doing our knowledge work for us anytime soon. Given we're doin the work, we deserve better tools to do it ourselves.
I'm currently building grain.co that layers a new kind of video conferencing on top of Zoom, focused on brokering the recorded information gathered during sync calls into the async tools where knowledge lives and actual work is done after the convo ends.
Our approach is to make it easy to annotate in real-time, clip out those parts you want to save/share, and then push them into async tools where anyone can view them like it were text in a document or a message in Slack.
EDIT: not a criticism, just made me think of a tangentially related thing:
This is so emblematic of how AI is a moving target. Once something becomes an everyday tool it's no longer AI?
Not so long ago, being able to transcribe speech from multiple people in natural conversation would be the hallmark of AI research.
Indeed!
I just had longer post about it https://news.ycombinator.com/item?id=25689370 because people try to solve problems that are already solvable by P2T
I've been using [1] for system-wide push-to-talk on OSX and it has worked well for the most part.
So you do need somebody technical to make the space work how you need, and if your users are all non-technical you probably only want one type of rule everywhere or it'll confuse them.
If you do that, it works a treat.
1. You don’t see who is speaking when there are many people present— you have to scroll through participants to find the person talking. Lots of new employees, I didn’t know who was talking about half the time
2. Connection issues. I couldn’t hear the speaker but others could, for extended periods of time. Missed entire presentations due to this unfortunately, as I couldn’t hear or have any means of connecting to the speaker.
Don’t get me wrong it’s fun software, but it’s not ready to be your rock solid all-hands replacement quite yet. Had the pleasure of experiencing Welcome at one of the YC networking events, and that worked really well, curious to hear if they’re interested in filling this gap
Was the speaker at a podium? Were they spotlighted? Were you close to them?
There are several reasons why you might not have been able to connect that are part of the design. We found initially people were confused by the connection rules, but once they got the hang of it, it all "just worked".
I agree completely that it takes time to get used to, and perhaps you did have genuine connection issues, but we ran with 150 to 200 people for 48 hours without any significant[0] identified connection issues, which is why I wonder if it was mis-identification of a "symptom".
I'll check out "Welcome".
[0] There were occasions when people needed to reload the tab, but doing so always fixed the problem.
FWIW the person I could no longer hear was located across the globe, so maybe that’s related.
I apologize for not having a more thorough write up of what happened, but it was certainly a real issue.
It does sound like it was a real issue ... have you provided feedback to Gather.Town? Perhaps it is related to the speaker being across the globe, and they could find that useful, even if anecdotal. It's still a data point, pointing at a potential underlying problem.
Regardless, thanks for the reply. (and NRN)
As for the other issue, active speaker detection is on our near-term roadmap. It's become more pressing for us too, as our team has grown and no longer fits in one page :P
Thanks for the feedback!
I’ll keep the feedback method in mind for future usages!
Also, I get it, having started to work with the WebRTC stack myself. While the service is free to use I think I can stomach a few issues while it’s actively being worked on :)
I swear, the best video chat experience I had was from The Go Game. At no point were you dumped into a massive shared room. Instead, people can join others at-will into small "rooms" of up to six people, which is much more manageable. Then you can implement whatever games you want and let people decide what they want to do. No need for next-gen hardware, no fancy "VR spaces", just a different interface paradigm that actually scales.