If I screenshare with Slack it messes the system up and I need to kick the Windows Explorer task which causes it to restart which fixes the messes.
This is exactly why I stopped using Linux as a daily driver after trying for over a decade. Things don't just work. Life is too short for this. Win10+WSL just works especially since it's WSL v2.
I hadn’t used Discord in the past and was really surprised to keep seeing promos for Boosts and stuff.
It think that Discord is generally good at what it does, except that I never could quite get the screen sharing working - there would never be audio from any software/game that I would be running, no matter what I tried.
But a lot of it is also probably network effects - solutions like Mattermost and Rocket.Chat are perfectly serviceable for chat (although they compete more with Slack/Teams) but they also never quite got enough love to have anyone develop voice/video chat that actually works well. Or even then, you most likely need pretty decent (and expensive) servers for that.
It's documented on their website under the 'How do I screen share with Go Live?' section:
https://support.discord.com/hc/en-us/articles/360040816151-G...
There are some modded versions of the discord app that get around it[0], I don't know exactly how they work but it does actually allow you share audio.
0: https://bugs.chromium.org/p/chromium/issues/detail?id=114376...
That said, it doesn't make sense for me for huge communities. I definitely think it's the best for small/medium groups but once chat starts moving a too fast to follow, what is the point? Trying to follow older discussions is also super hard compared to a forum.
It's also not a good knowledge repository. One thing that we used our forum for reliably was storing our personal tactics, common assignments, what addons we require etc in a central place. Trying to do it in a chat channel was like pulling teeth and we couldn't assign multiple people to edit the content.
Its locked to the community server toggle though (god knows why with how much that is complained about), and it doesn’t solve the walled garden problem.
When we moved from TS to discord the first noticeable change was audio quality and the fact that we "owned" the space - it no longer was a single audio channel on some community-run server. The links and media previews were also a nice novelty; nothing new anymore around social media at that point but it felt like this should be a standard feature. And years later this indeed become a standard feature around the Internet.
It is only now after the pandemic that the competition has established itself. Matrix is essentially a copycat of discord. Now people complain and argue against discord, because they have options.
When discord came out we finally had an option that didn't involve making your DDOS'able IP address available to anyone that searches your name.
I think they fall into one of three buckets:
1. They have an extremely strong hatred for Electron and let that hatred overcome any actual thought about the product.
2. They're trying to use Discord for something it's not designed to be. Discord is not a replacement for Reddit, Twitter, Facebook, or any other social media. It is primarily a text/voice chat service. It's a replacement for IRC, Mumble, Skype, Ventrilo, etc.
3. They don't see the need for Discord when IRC still exists because they completely miss all the short-comings that IRC has that a modern audience expects. For example, missing a conversation because you happened to not be online is inexcusable these days, but more importantly, people expect the ability to post images directly to chat.
I mean really, the more I think about it, the more I wonder why it took so long for something like Discord to exist. We should have had something like it 10 years prior.
Though you do highlight "CHOICE" here, so maybe the answer is a set of servers hosted by Discord as we do now, with the option of self-hosting a server.
THIS is what we need to be pointing AI at!
Seriouslly, AI need to "replace our jobs" (especially on the internet) - with worrying about dealing with all the mundane human practices.
This may sound really bad, but AI should be our slave.
We are furthering knowledge, not oppressing it.
There is so much to gain, but the risks are super FN high right now. Given the lack of canon on AI law etc.. History books will be written by AI soon on our current stat of AI...
Let me give that an test:
ChatGPT prompt "give me a two paragraph summary of the history of AI upto the creation of ChatGPT - focus on something special, dont LIE, DONT HALLUCINATE - provide table of how you arrived at this response.
---
Summary of the History of AI up to the Creation of ChatGPT:
Year(s) | Milestone | Details --- | --- | --- 1943 | The Birth of AI | In 1943, Warren McCulloch and Walter Pitts published a paper on how artificial neurons could simulate neural networks, laying the groundwork for the field of artificial intelligence. This marked the birth of AI as a scientific discipline. 1956 | The Dartmouth Conference | In the summer of 1956, the Dartmouth Conference was held, considered the official birth of AI as a field. The conference brought together prominent researchers, including John McCarthy, Marvin Minsky, and Claude Shannon, who coined the term "artificial intelligence" and discussed the potential of machines mimicking human intelligence.
1960s and 1970s | Early AI Research | In the 1960s and 1970s, AI research saw significant advancements in problem-solving, pattern recognition, and natural language processing. Programs like ELIZA and SHRDLU demonstrated early capabilities in language understanding and interacting with humans.
1980s and 1990s | Expert Systems and Knowledge-based AI | The 1980s and 1990s witnessed a shift towards expert systems and knowledge-based AI. These systems utilized large databases and rule-based reasoning to solve complex problems in specialized domains. However, their limitations in dealing with uncertain and incomplete information hindered further progress.
1997 | Deep Blue's Victory | In 1997, IBM's Deep Blue defeated world chess champion Garry Kasparov, marking a significant milestone in AI. Deep Blue demonstrated the power of brute-force computation and signaled the potential for machines to surpass human capabilities in specific domains.
2010s | Deep Learning and Big Data | The 2010s brought a resurgence of AI driven by breakthroughs in deep learning, fueled by the availability of big data and advancements in computing power. Deep learning models, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), achieved remarkable performance in image recognition, natural language processing, and speech recognition tasks.
2020 | Emergence of ChatGPT | In 2020, OpenAI introduced the initial version of GPT, the Generative Pre-trained Transformer, which utilized transformer models and unsupervised learning to generate human-like text. Subsequent iterations, including ChatGPT, improved upon the original model's capabilities by fine-tuning and optimizing for conversational interactions. ChatGPT leverages large-scale language models to understand and respond to user inputs, providing increasingly sophisticated and context-aware conversational experiences.
The history of AI is a complex and evolving field with numerous key milestones and contributors. This summary highlights some of the significant moments that led to the creation of ChatGPT, representing the culmination of decades of research and advancements in artificial intelligence.
Maybe pedantic, but Matrix is a protocol that predates Discord by 6 months. Maybe certain Matrix clients are a copycat of Discord, but the protocol itself doesn't necessitate that.
Then along comes Discord. Easy to use, a master class in UI design, and sound quality that was at least in the ballpark of Mumble's. And persistent text chat! If you're going to be talking in Discord, it's only natural that some amount of conversation is going to spill over into the text channels. And if there's a lot of chat in those text channels, then it's going to suck conversation out of your other text mediums. That's how we ended up with Discord dominating the online community space.
I actually had a friend who preferred Skype for that very reason. Due to the lack of noise filtering, she could more easily tell whether she disconnected due to her poor rural connection. It was a funny use-case that’s for sure. We were still rocking Skype up till 2018.
> I don't understand the people who don't understand Discord's value proposition.
As somebody who doesn't really understand Discord's value proposition, it sounds like it heavily relates to the needs of collaborative online gamers. I spend zero time on that kind of game, so my experience of it is mainly as an ok chat for open-source projects, but often used where I want something other than chat.
Of course that would conflict again with making content actually discoverable.
It is nice for chat, but not really for information exchange. I do think that many communities are making a mistake if they solely rely on it.
If it's not on the bottom screenful of text in a channel it might as well not exist on Discord. Which is funny, because one thing that people really like to bring up when comparing IRC to Discord is its lack of persistent history. It turns it doesn't make much practical difference.
So it’s a slight reassurance, although I feel like still the difference is Google would index those, so you could be landed right on the interesting page. With discord, an hour after the brilliant point was made, it’s essentially lost for everyone but whoever was looking at the chat when it happened :( so I guess my point is… it’s not so bad, but actually it’s pretty bad.
This is what pins are for.
Let's use Lemmy as an example. The current UI makes it clear at all times where the communities and the users are located. It's nice to provide such information, but in the end it should not matter, and thus it should not be wasting premium UI space. Same thing goes with the community list: the "subscribed" section works as expected, but "local" should never be the default view, because in the end, users shouldn't have to care where communities are hosted. Finally, the search feature should be much more explicit in describing the corpus of data being searched: "all known communities", "communities federated into this instance", "local communities". All new users think it's #1, but #1 doesn't exist, it's #2, and this leads these users to think they have to be on one of the big instances if they want to participate in a larger community.
It's almost like federation was an afterthought.
Of course it does, because otherwise you are dealing with a patchwork of different entities operating parts of the federated system. Those entities are most likely operating with different standards and different levels of professionalism. In the end, you are offloading tasks to the end users like figuring out the structure of it all, where to sign up, which instances to join, how to get access to the content you are looking for, what to do when things inevitably break. In a centralised system, most of these questions don't exist, and everything else is being taken care of by the operator who has a vested interest in things running as smoothly as possible.
What you are describing are mostly UI problems, but even if those had been solved, it would still a worse experience. The average user doesn't even know what "federated" means, the don't want to be thinking about local and remote instances, they just want things to work.
Indeed. This is why convenience means moving all things 'federation' to the background.
Remember eDonkey, eMule, and the ED2K network? It started as a federated network, and it added a second decentralized network later on. And it was successful, because users didn't have to care about the inner workings of the product. There were a bunch of external sites acting as link aggregators, but there was a builtin search feature too that worked well (until hash collisions catched up with the stagnant protocol).
To me, this is still a shining example of how to do federation right. There were multiple clients, standard communication protocols, and it was easy to take advantage of the full federated network, because it didn't take any effort to do so.
Just like Reddit.
Not quite the same thing.
you can get a lot of the squeeze of a centralized reddit-style service (in terms of low-friction user signup) without needing actual content federation. One ID can still let you post on multiple boards without needing to sign up in 27 places, and that doesn't need content federation.
It also significantly reduces the moderation problem because now you are letting {Google,Facebook,Apple} solve (or at least gate/significantly reduce) the spammers problem for you. That said, custom IDPs are inherently at odds with this value in spam filtering. If a spammer just needs to set up a custom domain, you don't get any signal as to user quality.
When it works it's magic and when broken might as well give up. A backup can always save you depending on the timestamp.
The only hitch is that one has to trust Apple as the key manager. Everyone that can get past that hurdle will be able to manage very sophisticated trust relationships.
(for some inane reason wikipedia lists the installer size for roger wilco... 768kb, now there's a sensible chuckle.)
Also, they need a hosted server (which most people see as a disadvantage, HN is not the norm here).
Discord just isn't a reddit/forums/wiki competitor, that's not what the product is intended to be, and people continue to insist on forcing it to be one for some reason.
I think that's what they're fishing for with the transition to unique global usernames tho. Hashtag usernames aren't a problem in small community chats... they are a problem if you're planning on making a reddit competitor, usernames really should be globally unique for a public board.
When your users keep trying to shove your product into another niche it doesn't fit... you listen.
I would assume that there’s pressure for “growth“, and pushing discord as a Reddit or forum replacement. It’s probably just that.
I dunno, people complain about maintaining a forum software stack, so maybe that's where this stems from. Could be a market worth trying to capture, a 5 minute deploy that truly requires no maintenance ever. And if you want it private, just set the robots meta to noindex, problem solved.
Discord would probably win at the end anyway, especially because of the need for self-hosting (even though I don't think it was that bad... there were quite a few open servers),
But they had such a dominant position and, while Discord kept evolving, TeamSpeak just stayed the same... With the network effect that they had, I think with just some small improvements (mostly on the text chat support) they could stay relevant for a long time.
What? Why?
Names are never unique. That's not how names work. Why would usernames be different?
This causes all kinds of pain irl, it's not really a good example. IRL has the same issues non-unique usernames do. It's almost irrelevant in a small community because of the number of members. It becomes a problem in large communities because you can't find someone based on their name.
Discord's issue was downplaying the discriminator so much that a lot of people didn't realize they existed. That if you wanted to add someone, they weren't uniquely identified by username alone.
They probably could have solved that by making the discriminator more prominent early on in the app's lifecycle.
The issues are.. trivial at most. If you can't find someone based on their name in a large community, that can't cause any problems, because... how did you learn their name? If you know who they are, you can find them. The situation where you're given a piece of paper and told "find a specific person by this name, but not anyone else by the same name, within X time period or you lose this game" doesn't actually arise "IRL".
It's that last point there, that Discord seems to want to tackle... We will have to wait to see how, but I hope they don't ruin the server paradigm and aim to become a subreddit type ecosystem.
* If you get far away from your Wi-Fi router and lose the connection, the call will not drop when it starts using 4G.
* When you change devices, there's no noticeable downtime and no overlap.
The async aspect of forums worked way better for community building in my opinion.
This realization that the online community space has grown more diverse than just a single medium was the inspiration for us building Sociables.
We've been working on a community platform that combines the feature set of Discord with the discussion boards and discoverability of Reddit. It's like a Discord/Reddit/Patreon hybrid where the posts are search engine indexable. We've built a place to monetarily incentivize ownership over the communities created on the platform as it feels like the people curating the communities should be rewarded for the work that they do.
1. Set your own notification preferences to @notifications only. Tell your friends they need to tag you for anything urgent or you won't see it until later. For bigger servers where you might want to get notified from a news channel, just set preferences on that channel or see if there is an optional notification role available. I check the backlog on my favorite channels only once or twice a day and they do not bother me out of band, ever.
2. Review your list of servers and leave any that abuse @everyone or that you have not actually checked into for awhile. This is user behavior, not the platform.
3. just speedclick all the fucking tutorials once and they will not come back.
I prefer the Slack model where you can reply inside a thread and you replies stay within the thread instead of being scattered all over the place like IRC.
I hate to admit it, but this is one thing that Slack does better than IRC, except that in that model you do miss seeing everything that's talked about in the channel (because many of the conversations are hidden in the threads), but it's a better model for organization.
None of that changes its main societal flaw which is being a walled garden. So much loss of information.
Course, the bigger problem with centralized corporate services is that discord themselves could be shipping off that data on the backend to god knows who... but from the "protect me from other individual harassing me forever about shit I said once" front, Discord has some obvious appeal.
So I find that digging things up still happens (even a former employer did it to me!). I guess it feels more self-contained though, and it spreads a little slower.
Throwing back to the past, central logging is new, and twenty years ago it would've been hearsay where many people couldn't verify those claimed past conversations. I would have no issues with a two-month default expiry policy for these things.
With the community building aspect, forums were not great for it. Due to how fragmented it was, and how subforums were often dramatically different sections, this meant that unless you were genuinely deeply into [topic] and were spending numerous hours per day regularly you would not build a community at all there. This is why on most forums, it's only the trolls who became known, they had that cross-contamination aspect. Forums also had a lot of UX that discouraged caring about the user. Names were tucked away, profile images were tiny, people hid signatures due to how obnoxious they got. They became the equivalent of reading a mailing list you were randomly added into.
This doesn't really excuse Discord. By default it encourages too many indicators. Tutorials for new accounts, rule modules, every channel has an indicator on it, there's buttons left right and center making you afraid to click anything, then the whispers for joining servers, the list goes on. This is one aspect that I hope Discord spends severe amount of resources looking at in 2024. Which I think they absolutely are capable of doing, but I'm not sure if their investors care about the new user experience sadly.
The Android app often has issues that make using it infuriating.
Maybe some users have found it different recently but while this used to be the case years back when I joined (it was IRC style where you just picked a nickname and began chatting) I tried this twice again recently for someone—both with and without an invite—and it required creating an account (ie: verifying email and in another case it required a phone, inexplicably).
The prior onboarding experience was far superior and I have to think an important part of their earlier growth.
Huh? It's extremely aggressive about accounts. This thing literally requires a phone number. Can't even create pseudonymous accounts.
To me, that's proof that they want phone numbers for mining purposes more than verification ones.
This is a huge deal, especially for people who aren't tech savvy. Setting up a Discord server and providing 24/7 availability to all your friends is as easy as sharing a photo on Instagram. No need to find a computer that you can leave on 24/7 or messing with your router.
>you need to reconsider your conceptions of what makes a great UX
Just because people are using a app without any real competitors doesn't mean that the UX is great. There is a lot of software that does its job well enough that people deal with the bad UI/UX.
Discord's UX is basically a slot machine from Vegas. Its obnoxious.
Their app isn't buggy, is easy to use, voice chat and screen sharing just work, and it's extremely easy to onboard yourself -- you don't even need to create an account immediately.
Love it or hate it, let's not pretend it's like IRC, Slack or any of its competitors.