Hacker News new | ask | show | jobs
by moojd 1095 days ago
This is really nice. Just needs one click joining for voice channels. This is an under-appreciated killer feature of discord that slack missed when implementing huddles. The ability to instantly hop in and out of voice channels is what keeps discord from feeling like any other teleconferencing app. Any friction to this makes it so I am less likely to just casually jump into a call.
7 comments

Please no.

As someone who doesn't use voice channels, I can't even count how many times I've accidentally joined them because I mis-clicked. And then I have to find the "end call" button because, of course, it's small and located away from the channel itself. It is so unbelievably annoying.

The best feature Discord added recently was the ability to hide channels, so I can finally, once and for all, forget about voice channels.

I've been using Discord since 2017 and in that time I think I've literally never accidentally joined a voice channel? The app even prompts you if you're sure about joining under some circumstances.

GP is right, frictionless voice channels is a killer feature. Maybe a toggleable prompt or some other feature to prevent accidentally joining can be an opt-in.

> I've been using Discord since 2017 and in that time I think I've literally never accidentally joined a voice channel?

I've also been on Discord since 2017, and I've done it many times. Especially on servers which put voice channels right next to text channels. And I'm not the only one either, judging by another comment[0].

[0]: https://news.ycombinator.com/item?id=36439417

> The app even prompts you if you're sure about joining under some circumstances.

The app is even worse! It brings a full-screen pop-up and if you haven't granted Discord microphone permissions, it will bug you, every, single, time.

Maybe ask the server admins to make a voice channel category and then not only are text/voice separated/grouped but you can collapse the voice category when not using a voice channel. I really just don't see this problem, it's not even that I haven't done it but also I don't see other people doing it. To be fair though, like half the servers I'm active in organize their channels in this separated way.

Anyway, I think opt-in for some kind of friction makes sense.

Part of the issue here is that historically, audio channels didn't have any association with text channels, and so you'd find setups where people would pair them up and interleave them.

Nowadays each audio channel has an associated text channel, making it a lot easier.

Yeah, it's obvious - just have a user setting: "Join voice channels without confirmation". Requiring confirmation should be the default, to prevent users from accidentally joining.
At least on all the commercial platforms you can now say that specific apps don’t have access to the microphone so this is a fixed problem.
Any design which removes friction in an application gets my blessing. Having to long press and confirm or approve some benign action is super frustrating. Is accidentally joining a call really that bad? "Whoops" *Drop call*
Exposing your microphone to countless strangers could in theory have worse outcomes than “woops” based off numerous factors, such as what you’re doing/talking about/listening to and how long it takes you to notice that you are broadcasting the audio of the room that you’re in.
I use push to talk because I'm considerate.

Keep your mic muted if this is a concern.

Discord makes it clear when you're in a call.

I simply do not believe in adding friction because of bad users.

Push-to-talk is literally adding friction because of “bad users”.

If everybody were “considerate” then a simple toggle wouldn’t cause any issues.

Push-to-talk is a feature, not friction, as far as I'm concerned. I refuse to use apps that only have auto-record.
Also once the session is up, it's too late for your purposes. PTT or "just don't talk" is not a solution at all.
Necro, sry.

Revolt has your mic muted by default.

It's not a huge deal in same sense as exposing parts of your initials, half masked e-mail addresses, or the last 4 digits of credit card numbers are not a huge deal, each of which could allow your adversary recover full details in some cases should they exist.
You just described the appropriate solution to this problem: better visual separation. You don’t need to add friction for people that like voice chats, just make it clear what you’re clicking on so you don’t have an oopsie
When someone clicks a voice channel, prompt the user to confirm that they want to join along with a “don’t ask again” option. Have this option be reconfigurable in user settings if you decide that you do want the prompt later.
The wonderful thing about free software is that you're free to change that behavior to something else that you like better.
Please yes. This is the killer feature. To solve your issue all you need is an otional modal to confirm joining. The idea is that you are joining a room not having a phone call. This enables things like joining by yourself while other participants join.
Yep, I wish there was a feature to hide them completely by default.
I believe they recently added that modal, in a recent installation it prompted me for confirmation before I clicked the "dont ask again" option
This was normal back in the days of IRC and Mumble. I should try introducing mumble into our company, come to think of it. I remember not understanding the client-side TLS certificate thing, so like, it had strong encryption way ahead of its time and the UI was basically as smooth as it could be because we all made it onto the server, even if we didn't understand what it was and the server just accepted any cert.

It's a good reminder indeed, I hadn't considered the old gaming toolkit we used as teenagers and I work with 100% technical people that can figure out an apt install mumble

Mumble's latency is unbeatable imo, it's basically their main focus and shows.

The sticking point for me is the lack of persistent messages, something the devs strangely think is a privacy plus. Issue open since 2016: https://github.com/mumble-voip/mumble/issues/2560

If you drop out for a minute you won't have access to anything that was posted in chat, which makes it useless for anything other than voice only comms, that might suit some business purposes but I've always needed to post links or screenshots in chat during meetings.

I think in the case of Mumble, the chat feature is intended to be used to let people fix sound issues, nothing more.

This sort of is yet another data point related to a conundrum: on one hand, applications that integrate multiple functionalities are convenient but there's always one or more functionality that doesn't fit your needs; on the other hand, applications that focus on one functionality do it well, but you have to hand pick one-by-one to fit all your needs, perhaps add some glue scripts/plugins to make them work nicely together, and this is inconvenient (in particular if you are not alone and not in an organization that "dictates" those choices) and a slow process.

Basically it is Office365/LibreOffice/Emacs or Browsers (with plugins to add chat, email clients, etc.), issue trackers (sometimes feature Wikis and more), Git online front ends (sometimes offer Wikis and issue trackers), versus e.g. raw Git, Vim/Notepad/Nano, bare-bones browsers (Surf, Midori browsers), ...

I favor the dedicate apps for things I do a lot - usually I eventually know them pretty well and I have the know-how to make them interoperate. For more casual stuff, I'm like the random user - the convenience just wins.

But for communication applications, as long as people jump the bandwagon of the latest private service provided "for free", you have to go with the flow or only be the nerd that talks with nerds on obscure networks.

> The sticking point for me is the lack of persistent messages, something the devs strangely think is a privacy plus

This is my complaint about Discord and such as well, though. People seem to love platforms where what you typed last week is never looked at again and isn't searchable/accessible from a regular web search. HN is also moving more and more in that direction: threads always dropped off very fast (matter of up to 3 days if you have a top-of-the-year popular thread) but then, at some point, editing was restricted to only be 2 hours (countless times I run into not being able to add a correction or addition, so posts are now less good/useful in posterity), and last week I noticed I couldn't reply to someone anymore after 18 days (they had replied to me, I had finally gotten around to checking their suggestion, and wanted to reply back ... alas).

Be it bad ears or a pretty rudimentary testing method, but the experience we had at a lan was that we could hear each other through mumble before hearing a shout from across a room.
I know that feeling. For work, I have to do basically everything in a VM, but even on the host the input latency (as measured by 1000 fps camera filming the keyboard with the screen in the background, while I strike a key) is not great or anything. Being used to this latency causes me to feel like the disk encryption boot screen shows the asterisk before I truly typed the letter of the password. (Relevant: https://danluu.com/keyboard-latency/)

I am on my way to bed but, damn, you're getting me curious about the delay that Mumble has versus speed of sound. Knowing it uses Speex ("This is an example of Speex ..." x100) which has a minimum delay of 30ms (https://en.wikipedia.org/wiki/Speex), and that you have <1ms ping on gigabit LAN so ignoring that by comparison (even if you>mumble>someoneelse is 2 hops), sound travels...

    $ qalc
    > 340m/s * 30ms
    (340 * (meter / second)) * (30 * millisecond) = 10.54 m
That would be the absolute minimum possible (unless modern Mumble switched to Opus and you used that version) for it to be true. But, considering the first paragraph, I am also very ready to accept it just sounds like Mumble is quicker just because it's so contrary to the norm.

If you say it was ~11+ meters (36 ft), I may be curious enough to create a test setup :p. But I'm assuming you mean the person sitting next to you, not someone who put Mumble on huge speakers across a hall.

It would have been five or six meters with someone speaking loudly into the mic and someone else with one headphone cup on. So you'd need another person to test.

It was also in 2009 so it could even have been Celt rather than speed or opus

I used to use Skype about 10 years ago for playing, then raidcall, then I eventually realized I could host a teamspeak server for the server I was hosting, so we got a non-profit license and hosted our TS server.

I played Eve Online and had to use Mumble. Damn, I wish more people used Mumble, it's such a nice piece of software and it's open source! If only we could have text chat mixed with the audio backend from Mumble, all this together in one app, I believe it would be amazing.

It's crazy how collaboration solutions keep missing this. I passionately hate "calls". You don't catch people with a lasso to talk to them in real life.
Not sure I understand. They're called "meetings" IRL right? Just that they're now online, a lot of the time. We use the word call and meeting interchangeably, I'm not really seeing the difference.

(The only difference between before covid and after is that we are more people and have more calls that take longer. Text chatting is a lot more structured and efficient and inclusive because nobody has to wait for a turn while the topic evolves past five persons before it gets to you, but whenever a topic gets more than a handful of messages, invariably $boss will propose to stop wasting time and "plan a meeting" .... but I also have that experience with IRL meetings when there are more than, say, five people involved)

Seeing who is in the channel is the big difference. It's one thing to just phone someone up, potentially disturbing them from whatever they were doing and suddenly demanding their attention through a ring. It's another thing to see they're in an open communication space and hop in to join impromptu. There's significantly less social friction involved.
The problem is that in the real world you can have more than one conversation in a space. With a group call/channel you get exactly one conversation or a bunch of people yelling over each other.

You could probably make that work in VC but the UX is going to be awkward and un-intuitive no matter how you slice it.

So the best option (online/available status and 1:1 calls which is what we have now) is probably the best alternative.

Discords usually have multiple audio channels, solve the "exactly one" issue. If you really want to have two conversations, one group drops down to another one.
Yes however that's no different than just being in multiple calls but just swapping between them (ala skype for business or teams).

It's still exactly one conversation you can be part of at a time. This compares to in person conversation where you can be part of a greater conversation while also carrying on a smaller, secondary conversation with the people immediately adjacent to you.

I've seen some conference software try to address this with a "tables" system where you sit at a table of a few people but are in a greater discussion room. Your table conversation is only heard at the table but the room conversation is heard at all the tables. then the table channel is "open mic" and the room "push to talk" (or both being bound to different "push to talk" buttons).

That kind of addresses the issue but it's awkward and doesn't translate well to anything other than conferences. Maybe a 2d/3d game style environment with range chat would work the best for something like this but I've yet to see an interface for this that isn't miserable outside of actual games. It'd solve the "walk up and talk" issue but it'd add 10 other issues all of it's own.

It’s about the ceremony involved to initiate an informal meeting.

IRL you might just grab an empty office or room and hit the whiteboard for ten minutes.

In other tools you need to invite specific people to a chat, and they have to find and respond to that specific event.

In discord you say “let’s meet in the engineering channel” and anyone who also wants to join can do so.

It’s a seemingly simple difference but I found discord so great for remote work relative to slack. It’s just much more casual, like real life.

I think he's referring to just walking up to someone and talking to them in the office.
Lol! This is such a great analogy I might lasso some folks just to tell them about it.
It's like talking over cubicle walls vs scheduling conference rooms.
the very first time i went looking for a discord plugin, i was looking for a way to block that after having entered too many chat channels accidentally. it might be a killer OPTION.
You can do this in slack by starting a huddle in a common channel. Its not as good but helps somewhat.
I think they mean that you cannot hop between different huddles in a single click.
Yes, I believe you are correct, discord makes it easier to see what you are potentially getting into much easier.
Why is slack preferred over teams, discord or revolt?
I'm not sure, it seems like a corporate thing? I prefer discord personally, unfamiliar with revolt, will say that teams in an academic setting was immensely frustrating.

But I use slack for work and some org's community space is also on slack so I make the best of it that I can.

But you can do that
Nice! I tried but I must be missing something. Is it a setting? Only way I can see to join voice is to click on the channel and then click the voice button?
You can create an invite link to a voice channel instead. There's even a wiki how to this https://www.wikihow.com/Invite-People-to-a-Discord-Channel-o...