It is sad to see many more communities vanish into some “protected” chat system. Many of them not even allowing 3rd-party clients…
I regularly struggle with opening all the Slacks, Discord, Gitter,… for all the communities that I am interested in. 5 years ago all I needed was the IRC client that also required a fraction of memory an CPU power.
At least for chat protocols, anyway. Time was when any home hobbyist could build a reasonable CPU out of transistors on breadboards, but nowdays only multinational mega-corporations have the resources and technology to build CPUs anybody wants to use. Even if you restrict yourself to multi-layer circuit boards and surface-mount components, you’re still nearly fifty years behind the curve.
Chat protocols, HTTP(2,3,QUIC) clients, web browsers. There used to be a healthy ecosystem around these, but now there’s just Google, and the competitor Google is funding to keep antitrust legislators at bay.
Forgetting FPGAs and Risc-V, which are helping in the CPU space: There are massive performance benefits to high end VLSI fabrication of complex chips. If the choice was less stark, I’d also be arguing strongly for ecosystem health and diversity over performance.
And yes, we have lost some of the diversity: PowerPC, Sparc, MIPS, Alpha, PA-RISC – the CPU space was better off when there was competition, instead of just X86 and Arm playing in disjoint sandboxes.
Hey everybody, we’re going to be deprecating our thing in a few months. Not only are we not at the point where we can begin proofing the thing we’re going to replace it with—and then going to spend the intervening time doing exactly that—but we actually have no idea what the replacement is even going to be.
Yet, I understand why IRC is shitty and exclusionary for everyone who has joined the internet after 2010. Especially with the spam problems that IRC networks have to keep up with :(
AFAIU, the idea is to solve for “make our text-based chat widely accessible for human beings” rather than “make our text-based chat great to re-implement on a rainy day”. The other blog post, linked by @notriddle below explains the “why” so much better imho http://exple.tive.org/blarg/2018/11/09/the-evolution-of-open/
I maintain the IRC server software that Mozilla IRC uses. I used to be in contact with the person who managed the Mozilla IRC server but they passed responsibility onto someone else and in the years since the new person has not bothered even once to reach out to us about solving their issues with IRC.
We have plenty of solutions available to deal with the kind of problems they claim to be having. If they want they can make it so people need to be logged into accounts to interact with channels/other users to solve abuse/spam they can do that. If they want a fancy modern client UI then there is several modern interfaces which are very accessible (IRCCloud, The Lounge, Kiwi IRC, etc).
Ultimately it seems to me that they are just making a knee jerk reaction and deciding to jump to some other platform without knowing what they want to move to and without actually looking into seeing if any of their problems are solvable. ¯_(ツ)_/¯
That post has been pretty thoroughly debunked (the author posted the same comment to multiple threads). Mozilla does have the spam mitigations they mention enabled. Mozilla does have a corporate IRCCloud license. This is not some ill-considered knee-jerk reaction borne out of incompetence. It has been considered at great lengths and been a very long time coming
While I’ll personally be sad to see irc.mozilla.org go, I understand why it is necessary and am on-board with the decision.
I just fear that a closed platform such as discord is not a good fit for open source projects in the long run. You are not permitted to use your own clients and are ultimately in a vendor lock in. Of course they can just switch again when their core team became dissatisfied at any time
I’m not an IRC admin, so I don’t have anything to say about the moderation aspect beyond what the article says. But guessing that the fact you can’t delete hateful messages is a big part of it.
But for me, and I say this as someone who loves IRC, the bigger reason is that IRC excludes people who aren’t as technically savvy. Yes I’ve used IRCCloud and no, it isn’t good enough. The only way to get persistent messages is to either pay for a service (not fair to ask contributors to do this, especially those from developing countries), or set up a bouncer/relay (this might be easy if you are an old Unix hand, but not so much for most people). And besides, hosting your own bouncer costs money too.
Even the passage you quoted says “logged in accounts”. Making an account on IRC is hard. I remember the first time I connected to IRC to play a game of Star Fury, and it was enormously intimidating. That’s not a barrier that I want a young contributor who doesn’t have the greatest handle on English to need to surpass.
But guessing that the fact you can’t delete hateful messages is a big part of it.
You can’t not delete them: every message is immediately thrown away by the server. This is one of my main complaints with IRC.
Also, I find it somewhat fascinating that the complexity (or even impossibly) of understanding the software stack that you use is swept under the rug in that whole article.
It talks about controlling your own destiny, but how can you do that if you can’t even implement the basic protocols?
Making an account on IRC is hard. I remember the first time I connected to IRC to play a game of Star Fury, and it was enormously intimidating.
Seems like it would be easy to improve. Anything from an out of band sign up page to a /msg from a bot on joining.
(Also, the implicit assumption here about the ability of these people’s ability to use a chat program is a bit… uncharitable, don’t you think? I know I would be insulted if you told me IRC was too hard for me to use.)
None of the reasons listed for shutting IRC down make sense, and all of them sound to me like someone higher up just liked slack and wanted an excuse to switch to it.
Seems like it would be easy to improve. Anything from an out of band sign up page to a /msg from a bot on joining.
Maybe. IRC was “easy to improve” for years, I have yet to see those improvements happen.
And no, I don’t want to improve it, because I have a project to run and not write web interfaces for my chat tool.
None of the reasons listed for shutting IRC down make sense, and all of them sound to me like someone higher up just liked slack and wanted an excuse to switch to it.
Mozilla already uses Slack for internal discussion, but it will probably not be the replacement for community discussion, precisely because of bad moderation.
I don’t agree that the reasons don’t make sense, irc.mozilla.org is a constant target of abuse and just the cost of abuse handling is very high.
I’m kind of glad that the IRC server can’t reach into my client and delete messages from its memory. I can do that myself if I want to. While it tends not to really happen in professional settings, it makes me annoyed and confused when I see messages disappear from the history of e.g. Twitch chats. It’s as if I’m being punished for another user’s transgression. Why don’t I get to know what happened?
The point here is that we can delete harassment targeted at specific users or groups before it reaches them. There’s no use in receiving abuse and just deleting it yourself.
Spreading lies and attacks is pretty common and I see a case for deleting them before they are pushed towards other clients.
Also, no, I don’t believe you have the right to receive a full log of all message that were sent to our network.
All of the things above happen frequently on irc.mozilla.org, so they bring a big load.
I’m not saying I should receive a full log of all messages that were sent to your network. I’m saying I should be allowed to keep a full log of all messages that were relayed to me. I respect that some other people might not want that, but I think a feature that deletes stuff from my personal copy of the history should at least have an opt-out.
I think I’m fine with actually deleting messages before they’re sent to people, or rejecting them at the time of the receipt.
I’ve certainly lamented the ability to retroactively make people’s clients unsee drive-by spam and harassment, but I think there’s plenty of room to get better at dealing with it (both technically and socially) without resorting to rewriting history.
I’ve certainly lamented the ability to retroactively make people’s clients unsee drive-by spam and harassment, but I think there’s plenty of room to get better at dealing with it (both technically and socially) without resorting to rewriting history.
This is not even in spite, I’m sure a lot of protocol designers would like to speak to you. I like your optimism to improve socially, but I have to keep my communities safe, now. Targeted harassment is real and behavioural education won’t make it go away soon.
And the fact is that the Rust project deals with people joining our venues, spreading misinformation (and we’re very conservative flagging as such) and attacking people on a regular basis. IRC is by far the leader. We need a solution for that and clients that heed “delete” updates is currently what we have. You can use a client that ignores them, but yeah, that’s what we have.
How can you delete messages before they end up in other clients? Does Discord support this? How scalable is that?
I know there is auditorium mode on Undernet IRC, where only channel managers receive all messages from unidentified users, and have to acknowledge them manually before they will be posted for everyone else to see
There’s a ton to consider here. I’ll only talk about the case where clients read all messages. (In IRC, that needs a bouncer or similar)
The model of the server: If the client is only connected while the user is active, you can obviously delete messages for all clients who have not read the messages.
If a message has reached the client (in IRC in all cases), you can send a message asking them to ignore and delete a previous message. This is a best effort, but most clients heed that.
In federated systems, there’s obviously the problem that not only clients, but also all federated servers need to cooperate.
As you describe, pre-screening is a good option in networks that expect a lot of abuse. I do that for YouTube comments and I have to delete ~50%.
TL;DR: you need the clients to cooperate.
Deletions always get the reputation of censorship, but indeed, there’s good reasons for having and heeding those in a cooperative fashion.
They are an effective tool for abuse handling. Abuse is often aimed at emotional pressure towards a victim. A standard strategy is deferring a trusted party to filter your messages. Abuse victims let other people pre-read their email or twitter DM. Deletion messages support this workflow on a server-wide scale.
Chat networks are used to spread illegal info/pictures/content. Informing clients of this to take appropriate action is good practice. Even in a federated network, allowing the origin server to inform other federated servers that a message probably shouldn’t be kept is a good thing to have.
Spam is real, keeping peoples clients clean without their interactions is improving their lives.
I trust the chat operators of my networks, which is why I would never run a client that ignores deletions. I’m not alone in this.
In the end, it’s a best effort and is highly dependent on the exact model of your network, but cooperation makes a lot of sense.
That’s pretty depressing. IRC is one of the few protocols that is actually independently implementable and maintainable by almost anyone.
I think we’re seeing the rise of the nominally open but practically proprietary internet, thanks to the death of accessible protocols.
Fully agree.
It is sad to see many more communities vanish into some “protected” chat system. Many of them not even allowing 3rd-party clients…
I regularly struggle with opening all the Slacks, Discord, Gitter,… for all the communities that I am interested in. 5 years ago all I needed was the IRC client that also required a fraction of memory an CPU power.
At least for chat protocols, anyway. Time was when any home hobbyist could build a reasonable CPU out of transistors on breadboards, but nowdays only multinational mega-corporations have the resources and technology to build CPUs anybody wants to use. Even if you restrict yourself to multi-layer circuit boards and surface-mount components, you’re still nearly fifty years behind the curve.
Chat protocols, HTTP(2,3,QUIC) clients, web browsers. There used to be a healthy ecosystem around these, but now there’s just Google, and the competitor Google is funding to keep antitrust legislators at bay.
Forgetting FPGAs and Risc-V, which are helping in the CPU space: There are massive performance benefits to high end VLSI fabrication of complex chips. If the choice was less stark, I’d also be arguing strongly for ecosystem health and diversity over performance.
And yes, we have lost some of the diversity: PowerPC, Sparc, MIPS, Alpha, PA-RISC – the CPU space was better off when there was competition, instead of just X86 and Arm playing in disjoint sandboxes.
Earlier this year, also KDE stated they were “looking for a better way of chatting and live-sharing information for several years now.” – https://dot.kde.org/2019/02/20/kde-adding-matrix-its-im-framework
Classic Mozilla-style decisionmaking:
Hey everybody, we’re going to be deprecating our thing in a few months. Not only are we not at the point where we can begin proofing the thing we’re going to replace it with—and then going to spend the intervening time doing exactly that—but we actually have no idea what the replacement is even going to be.
(FWIW: I hate IRC.)
inb4 slack
Worse - Discord.
Please note that the Rust project was not involved in any decisions and is also not part of any evaluations. (We also didn’t want to)
We were just pre-informed and still have 2 active community channels and a working group there.
Half the company already uses slack. The other half still uses irc. It’s maddening. Any given solution will be better than the current situation.
I love IRC. I write IRC bots all the time.
Yet, I understand why IRC is shitty and exclusionary for everyone who has joined the internet after 2010. Especially with the spam problems that IRC networks have to keep up with :( AFAIU, the idea is to solve for “make our text-based chat widely accessible for human beings” rather than “make our text-based chat great to re-implement on a rainy day”. The other blog post, linked by @notriddle below explains the “why” so much better imho http://exple.tive.org/blarg/2018/11/09/the-evolution-of-open/
I’m not immediately going to link this other blog post by the same poster at the top level, since they’re so closely related anyway, but http://exple.tive.org/blarg/2018/11/09/the-evolution-of-open/ seems like good reading, too.
That article does make one or more great points and I think you should share it!
Interesting post on the blue site
Quoted here:
That post has been pretty thoroughly debunked (the author posted the same comment to multiple threads). Mozilla does have the spam mitigations they mention enabled. Mozilla does have a corporate IRCCloud license. This is not some ill-considered knee-jerk reaction borne out of incompetence. It has been considered at great lengths and been a very long time coming
While I’ll personally be sad to see irc.mozilla.org go, I understand why it is necessary and am on-board with the decision.
Perhaps it would be useful to explain why they’re insufficient, rather than just saying that you have them.
I just fear that a closed platform such as discord is not a good fit for open source projects in the long run. You are not permitted to use your own clients and are ultimately in a vendor lock in. Of course they can just switch again when their core team became dissatisfied at any time
I’m not an IRC admin, so I don’t have anything to say about the moderation aspect beyond what the article says. But guessing that the fact you can’t delete hateful messages is a big part of it.
But for me, and I say this as someone who loves IRC, the bigger reason is that IRC excludes people who aren’t as technically savvy. Yes I’ve used IRCCloud and no, it isn’t good enough. The only way to get persistent messages is to either pay for a service (not fair to ask contributors to do this, especially those from developing countries), or set up a bouncer/relay (this might be easy if you are an old Unix hand, but not so much for most people). And besides, hosting your own bouncer costs money too.
Even the passage you quoted says “logged in accounts”. Making an account on IRC is hard. I remember the first time I connected to IRC to play a game of
Star Fury, and it was enormously intimidating. That’s not a barrier that I want a young contributor who doesn’t have the greatest handle on English to need to surpass.It’s been linked elsewhere in this thread, but the author of this article’s other post is a pretty great read: http://exple.tive.org/blarg/2018/11/09/the-evolution-of-open/
This is bittersweet for me, irc.mozilla.org has been a huge part of my life for almost ten years now. But it’s the right call.
You can’t not delete them: every message is immediately thrown away by the server. This is one of my main complaints with IRC.
Also, I find it somewhat fascinating that the complexity (or even impossibly) of understanding the software stack that you use is swept under the rug in that whole article.
It talks about controlling your own destiny, but how can you do that if you can’t even implement the basic protocols?
Seems like it would be easy to improve. Anything from an out of band sign up page to a /msg from a bot on joining.
(Also, the implicit assumption here about the ability of these people’s ability to use a chat program is a bit… uncharitable, don’t you think? I know I would be insulted if you told me IRC was too hard for me to use.)
None of the reasons listed for shutting IRC down make sense, and all of them sound to me like someone higher up just liked slack and wanted an excuse to switch to it.
In spite of weaker moderation tools.
Maybe. IRC was “easy to improve” for years, I have yet to see those improvements happen.
And no, I don’t want to improve it, because I have a project to run and not write web interfaces for my chat tool.
Mozilla already uses Slack for internal discussion, but it will probably not be the replacement for community discussion, precisely because of bad moderation.
I don’t agree that the reasons don’t make sense, irc.mozilla.org is a constant target of abuse and just the cost of abuse handling is very high.
I’m kind of glad that the IRC server can’t reach into my client and delete messages from its memory. I can do that myself if I want to. While it tends not to really happen in professional settings, it makes me annoyed and confused when I see messages disappear from the history of e.g. Twitch chats. It’s as if I’m being punished for another user’s transgression. Why don’t I get to know what happened?
The point here is that we can delete harassment targeted at specific users or groups before it reaches them. There’s no use in receiving abuse and just deleting it yourself.
Spreading lies and attacks is pretty common and I see a case for deleting them before they are pushed towards other clients.
Also, no, I don’t believe you have the right to receive a full log of all message that were sent to our network.
All of the things above happen frequently on irc.mozilla.org, so they bring a big load.
I’m not saying I should receive a full log of all messages that were sent to your network. I’m saying I should be allowed to keep a full log of all messages that were relayed to me. I respect that some other people might not want that, but I think a feature that deletes stuff from my personal copy of the history should at least have an opt-out.
I think I’m fine with actually deleting messages before they’re sent to people, or rejecting them at the time of the receipt.
I’ve certainly lamented the ability to retroactively make people’s clients unsee drive-by spam and harassment, but I think there’s plenty of room to get better at dealing with it (both technically and socially) without resorting to rewriting history.
This is not even in spite, I’m sure a lot of protocol designers would like to speak to you. I like your optimism to improve socially, but I have to keep my communities safe, now. Targeted harassment is real and behavioural education won’t make it go away soon.
And the fact is that the Rust project deals with people joining our venues, spreading misinformation (and we’re very conservative flagging as such) and attacking people on a regular basis. IRC is by far the leader. We need a solution for that and clients that heed “delete” updates is currently what we have. You can use a client that ignores them, but yeah, that’s what we have.
How can you delete messages before they end up in other clients? Does Discord support this? How scalable is that?
I know there is auditorium mode on Undernet IRC, where only channel managers receive all messages from unidentified users, and have to acknowledge them manually before they will be posted for everyone else to see
There’s a ton to consider here. I’ll only talk about the case where clients read all messages. (In IRC, that needs a bouncer or similar)
The model of the server: If the client is only connected while the user is active, you can obviously delete messages for all clients who have not read the messages.
If a message has reached the client (in IRC in all cases), you can send a message asking them to ignore and delete a previous message. This is a best effort, but most clients heed that.
In federated systems, there’s obviously the problem that not only clients, but also all federated servers need to cooperate.
As you describe, pre-screening is a good option in networks that expect a lot of abuse. I do that for YouTube comments and I have to delete ~50%.
TL;DR: you need the clients to cooperate.
Deletions always get the reputation of censorship, but indeed, there’s good reasons for having and heeding those in a cooperative fashion.
They are an effective tool for abuse handling. Abuse is often aimed at emotional pressure towards a victim. A standard strategy is deferring a trusted party to filter your messages. Abuse victims let other people pre-read their email or twitter DM. Deletion messages support this workflow on a server-wide scale.
Chat networks are used to spread illegal info/pictures/content. Informing clients of this to take appropriate action is good practice. Even in a federated network, allowing the origin server to inform other federated servers that a message probably shouldn’t be kept is a good thing to have.
Spam is real, keeping peoples clients clean without their interactions is improving their lives.
I trust the chat operators of my networks, which is why I would never run a client that ignores deletions. I’m not alone in this.
In the end, it’s a best effort and is highly dependent on the exact model of your network, but cooperation makes a lot of sense.