tl; dr it was a standard and so could not improve fast enough with the improvements being discovered around distributed systems and mobile needs. I get the author's point, and we're seeing a lot of standards being outstripped by entrepreneurs these days, but I hope things slow down and standards become useful again. They are really important to maintain the open source world.
If you go to a nice restaurant that makes more than enough money to support itself and you enjoy the service and the food and then someone builds 100 McDonald's all over the state this doesn't instantly transform the restaurant into a failure.
If there are no new users due to people not being able to figure out how to use it, I am OK with that. I pretty much figured out how to use IRC (started with mIRC) at age 10 or so. I am not sure it has to do with lack of awareness either, many widespread projects have their own channel on Freenode which they put on their website.
Why do you think it barely has any new users, and how do we measure it?
It totally did not fail. It is the most useful realtime chat for many topics I am interested in:
#python for the python language
#php for the php language
#javascript for the javascript language
#django for the django framework
#laravel for the laravel framework
#symfony for the symfony framework
#startups for startup stuff
#css and #webdev for web development
#mysql for mysql and mariadb
#math for math related topics
And many more.
Show me any other chat system where you can go online, ask a question and get answers from professionals in the field instantly. I would immediately use it. But there is none.
Isn't the speed and usefulness of Q&A a function of the community itself and not IRC though? For example, the Matrix devs hang out on many matrix channels and are fast to reply to any questions. Experts over at the hackintosh discord are also very fast to reply and extremely helpful to help troubleshoot issues.
The point with IRC is that for some time it was the place where everybody was ... at least till the EFnet/ircnet split and introduction of freenode and all the other networks.
Makes you question what failure is. IRC played an extremely important role in spreading computer science knowledge, bootstrapping what we take today as a granted IT world. It helped build communities for the introverts, with various clients introducing basic scripting and programming technique to the curious minded. For me IRC played an irreplaceable role in my life as a developer and I would never be where I am if it wasn't for the networks built on that protocol. I think that IRC never failed, it was just outgrown.
Also it had an enormous impact on early online gaming. IRC was the default way many clans assembled, communicated, recruited, and coordinated match play and scrimmages between each other. Channels like findscrim allowed CS teams to match with other parties of equivalent skill looking for scrimmages in real time all over the world, 24 hours per day.
Indeed, I remember when we wrote an IRC bot for organizing scrims, and it would also keep a history of the results and so forth. Fun times. I miss it. :)
Is IRC really a failed endeavor? I would consider something to be a failure if it never took off the ground, instead we have a protocol that is over 30 years old being finally supplanted with newer technologies. There is no inherent reason why IRC failed, except that it was designed with the sensibilities of 1989 in mind, but the requirements of the modern day are not what they used to be. I enjoy the simplicity of IRC as much as the next guy, but when you have things such as a lack of chat history availability being something you have to work around, I understand why it has been supplanted by newer products, as has a vast majority of things from those days.
I didn't even bother reading the article after reading the headline. As you pointed out, IRC was far from a failure. Even with the coming and going of all sorts of realtime chat networks in the 1990s and 2000s, IRC still maintained vibrant stable communities of users. Freenode was a great more modern example. It's succeeded through a whole variety of what one could more reasonably consider 'fad' technologies.
Agreed. It's so weird to look at a project/protocol and think that anything less than eternal market dominance is failure.
IRC heavily shaped the modern web, and had market dominance for decades, and built so much user loyalty that die-hard fans are still refusing to migrate off today[0]. IRC isn't the future of chat. But it is completely normal, and often even desirable that successful things should someday die when new paradigms become more useful.
The only way to call IRC a failure is to have such a narrow definition of success that almost nothing in the world could ever be considered successful. Discord is 5 years old. I guarantee you Discord is not going to last 30 years before everyone migrates to a new platform. It probably won't even last 10 years.
Matrix is (in my mind) the future of open chat. If Matrix ends up lasting 30 years before we come up with something better -- and if it gets even half of the market penetration that IRC got -- then I'll be ecstatic. That would be success beyond my wildest dreams.
To be sure, IRC is alive. I just know I've stopped using it a lot less. I make the point that I (and perhaps many of us) used to use tools everyday that were standards based, now they're not, and that's a shame. Especially true with chat clients. It's like it's 1999 again, when you needed to be on MSN IM, AOL, and Yahoo chat to reach all your friends. Then pidgin came along. What a glorious standards based time that was. It was also the time I got into IRC. Then slack came. All of a sudden the people I needed to reach were more likely on slack then IRC. I also got on mobile around that time much later than everyone else. That's when I really started using XMPP and IRC less. That's all well and good, but now where are we? GroupMe, Slack and Teams open on my laptop 24/7. Multiple chat clients. Hello, 1999. It's been a while.
My point isn't "yeah it really is dead" but "man we need to get back to those pidgin days somehow"
100% true. I use IRC for work and I have that feeling every single day.
I would say this 1999 vibe is caused by smartphone and the "issues" that XMPP and IRC have with multiple clients for the same account. But it's a pity that, instead of improving those protocols, now we have to deal with multiple clients for multiple services.
Why people understand that you can use email with whatsoever client you want, but no with chats?
When Slack still had the IRC bridge, I participated in my company's team slack using Irssi. There were a few things that didn't work great, but it was so snappy compared to the official client and I could run it in a tmux alongside everything else I was doing (I was in a devOPS role; light on dev, heavy on ops). Plus it didn't fill my screen with animated gifs all the time... that was nice.
There's still a community I hang out with on IRC, and I started thinking about finding an IRC client for my new Macbook Pro (have been using X-Chat on Linux for months). Just the other day on HN I learned about Erc, which is an IRC client that's built right into Emacs. Tried it out last night and it's working awesome so far! Now if only the 8 Slack teams I'm a part of would switch over...
today, I get a notice in email from Slack when Y-ImportantPerson speaks directly to me, which means a record is entered in a graph database somewhere that says "mistrial9 spoke to Y-ImportantPerson" --meta-data-- .. I am impressed by the addition of constant logging of all communications activity on Slack, as a solution to hacking activity ??
Is this the same social promise that brought us directly into the F*Book era? "people don't care" is not a real answer...
* An actual example session as experienced on gui clients is simply filled with people talking on various channels that the user has selected on prior sessions.
* Clients like hexchat are plain but functional. I'm sure if a designer was hired they could figure out a way to make it prettier and harder to use.
* I don't generally worry about mobile support because I prefer in depth discussion with a real keyboard on a real computer.
* It "failed" because most of the dumb people who don't have anything useful or informative to say prefer to say their nothings via the mobiles and you can't sell people IRC hosting at 10 usd per user per month.
* The only thing it really needs is trivial persistence via a cheap bouncer integrated into clients with fewer steps. As in add credit card number click go.
> I'm sure if a designer was hired they could figure out a way to make it prettier and harder to use.
I hate the UI/UX today. It is unbelievable how much information fits into my 80x24 terminal running irssi, whereas for example Discord takes up an entire workspace and you cannot customize it to your liking, there is just too much unnecessary spacing, and you cannot make the list of friends disappear and so forth without using an extension. Discord started adding unnecessary crap which made me resort to using uBlock to hide them. I cannot do this with Electron apps, so I am using whatever I can (and have to) from the browser, but I wish I did not have to use them in the first place.
In any case: IRC will never die for me. Xterm + irssi will never die for me. :) I will continue growing old with my buddies on Freenode! :D
IMHO it didn't fail. The internet "failed", becoming something much more centralized and oligopoly-like.
In my area, IRC servers were maintained by local ISP administrators plus some help of local volunteers. As local, dial-up ISPs consolidated and ended up absorbed or put out of business by telecoms, IRC networks become orphans.
It didn't fail. Not a perfect analogy, but saying "why IRC failed" is like saying "why horse and buggy failed". IRC was quite successful. Still is really, just is used a LOT less now.
IRC failed because your parents don't know how to use it.*
*Or, given that this is 2020, your parents probably do know how to use it, but your children can't be bothered to figure it out because the walled garden alternatives already have all of their friends online.
> Clients joining and parting channels and servers didn't
have much state
State and state management is the main design problem at the heart of IRC. There's actually a lot of state considering that this is state that has to be replicated across all servers. (All servers have to know about all users currently connected)
Servers are connected in a spanning tree. All servers need to have the same state. State changes are initiated locally at each server and propagated through the spanning tree (which simplifies state propagation a lot). State changes take time to propagate, which means that servers will not have the same view of the world at any given time. Network partitions means the servers on each side of the partition have to forget all the state that belongs to the other side of the partition. When networks join conflicting state has to be resolved and there aren't really any good ways to do this.
Perhaps we have different views on "state". Perhaps you are talking more about the user experience and not the design challenges inherent in the low level design?
One thing that makes me very sad is that XMPP was such a pig's breakfast. It would have been really nice to have a messaging system backed by an IETF standard that wasn't infested with whatever technology was popular at the time of its creation.
One of the common complaints I have heard about IRC is that there's no history if you're not connected (contrast to, say, Slack). In most cases I consider that to be a feature; one of my biggest complaints about the way many teams use Slack is that it becomes both a real-time chat system and the canonical state of truth about decisions and how they were made.
It's a little more work, but I would personally prefer if chat was used to discuss things, and then someone took the time to actually summarize the discussion and conclusion into a document somewhere. Maybe it's Markdown that goes into Git, maybe it goes onto a wiki, maybe it's just a long-form email that people can review at their leisure. But there's got to be some kind of better record than a disjointed Slack chat.
reply