Notices by Strypey (strypey@mastodon.nzoss.nz)

@AmarOk being able to see the person speaking is still a big improvement on seeing nothing at all.

@AmarOk I'm using the metered data use case to demonstrate the difference in resource use. Which you now acknowledge is serious enough to consider not using video at all. In which case sending only one video at a time seems like a useful middle ground, for when some or all participants don't quite have a powerful enough device or stable (or cheap) enough net connection to make a full video conference viable.

@AmarOk
> if you are 2, 4, 16, 32, you will just receive 1 video and 1 audio stream

I understand that, but the size of that stream is going to vary quite a bit, isn't it? I can't see how a stream that contains 32 moving images isn't going to use a lot more bandwidth than a stream that only contains 1 moving image.

@AmarOk now imagine 5 users who only own mobile devices and only have access to metered mobile data. They take turns being the hosting peer, one per conference, to share the cost. If the hosting peer only has to receive and distribute 1 audio/ video stream at a time, instead of 5, that will be much cheaper for them.

@AmarOk
> this will change nothing

Let's say I'm using metered mobile data. Each packet of audio/video I send or receive costs me money. PTT that doesn't send audio OR video when I'm not talking saves me a lot of money while still allowing me a video conference experience. Also, surely receiving 2 or more video streams from the hosting peer (plus background noise) uses more of my data than a single voice/ video stream?

@wolf480pl
> if you found a way that gives people who pay better latency,

Yes, the whole scheme depends on @beadsland 's proposition that P2P-assisted streaming can work more efficiently than a broadcast server.

> you could make a centralized non-blockchain ledger.

Yup or use something like Holochains, which are neither blockchains nor centralized.

@lunch @alcinnz

@wolf480pl
I'm only continuing with this as a thought experiment, but ...

> at least one viewer must pay to watch

You could engineer the system to return everyone's tokens if they all pass on chunks like good netizens. But once you've gone to the trouble of introducing a payment system, you might as well use it to facilitate audiences paying producers too.

> And IMO it's likely that most viewers won't be willing to pay.

Netfix and Twitch suggest otherwise.

@lunch @beadsland @alcinnz

@Alonealastalovedalongthe if you're interested in voice chat apps, you could test Matrix by installing RiotX (or any Matrix client) and joining the chat room for our testing group:
#voicechat:matrix.org

You too @jamesmullarkey 😃

@hirojin yeah, you shoot that messenger 😂

@AmarOk if PTT for both mic and camera was an option in Jami conferences, I would definitely use it.

@AmarOk
> But you can just stop your video when you want.

True, but the ability to mute mics doesn't have the benefits of push-to-talk, where apps don't waste resources encoding and sending background noise, and muxing multiple streams of background noise in with the stream of the person talking. PTT also helps groups identify who wants to speak next.

> The video is the big consumer of bandwidth.

So sending video only while speaking would save even more resources than PTT audio, yes?

@AmarOk not encoding and sending an audio and video stream, except when speaking, might make performance better for the peers who aren't hosting the conference too. Especially if they have old devices or poor net connections.

@jamesmullarkey
> Read more from others about this

... is missing:
* https://drewdevault.com/2018/08/08/Signal.html

* http://web.archive.org/web/20190427000021/https://sandervenema.ch/2016/11/why-i-wont-recommend-signal-anymore/

* http://web.archive.org/web/20190509025806/https://www.coactivate.org/projects/disintermedia/signal-fault

@jamesmullarkey
> You should use Signal and encourage everyone you know to use it instead of WhatsApp. Simple as.

Yeah, nah.

@jamesmullarkey
> WhatsApp touts the ability to share your location with people as not only a feature but sensible and worthwhile.

Signal does this too, at least on Android, with dependencies on GoggleMaps that make it difficult to use with free code navigation apps:
https://github.com/signalapp/Signal-Android/issues/6351

@jamesmullarkey > Signal will let you video chat with only one person – you can talk to several people at once with WhatsApp.

... or with the Jitsi Meet integration in Riot. Or with Jami, a P2P chat app in which the voice/ video conferencing currently works better than text chat (no group text chat yet).

@jamesmullarkey
> Edward Snowden is a major advocate and he should know as his life has depended on it.

He has said he uses it, but I don't think his life ever depended on it. Signal has only existed since 2014. Snowden arrange the NSA leaks in 2013 using Tails GNU/Linux, PGP, and Tor, not Signal:
https://www.wired.com/2014/04/tails/

@jamesmullarkey
> Activists and journalists use Signal

... maybe, but Moxie has specifically said this is a bad idea, since Signal is intended to protect Jane Average from mass surveillance, not people likely to be the subject of targeted surveillance.

@jamesmullarkey
> Signal only ever asks you for your phone number

... which in most cases would allows any compromised account to be traced back to a human user. Signal claim they need your cell number to prevent spam, which is garbage, since both Wire and Riot provide encrypted chat without asking for one, and don't have noticable spam problems.

#Signal #Wire #Riot