The End of Freenode – the End of IRC?

Internet Relay Chat (IRC) is an instant messaging chat protocol whose origins date back into the 80ies and is one of the few “old” communication protocols with European (more exactly: Finnish) rather than American roots. In 1993 the server/client protocol was standardised by the IETF as RFC 1459. IRC since then survived all and any chat services introduced later, both proprietary ones like ICQ and free ones like XMPP. Until now specifically the Free and Open Source Software (FOSS) community was holding to IRC, even though it recently underwent attacks by optically improved proprietary services like Slack. It however is not unlikely that these proprietary services vanish as quickly as they appeared. It would not be the first time that IRC stands out as the rock in the wild sea. The comic drafter Monroe has summarised this in his unique mannor as can be seen in Fig. 1.

IRC has been resistant against not only proprietary, but also against free protocols. Jabber/XMPP never gained similar traction because it is enormously split into several extensions (XEPs) and the still new Matrix has been critised as a quasi-monopoly of the primary matrix.org server, since there are only few people willing to host the resource hungry Matrix server software. In addition, the lack of a formal protocol specification poses significant difficulties for the development of alternative clients other than the official Element (formerly Riot). Both IRC and XMPP on the other hand feature a multitude of different clients which serve any possible preference. In total, compared to IRC I expect Matrix to have a rather short life span due to its complexity.

It is possible to find multiple reasons for IRC’s survival. One of them is its technical simplicity. The communication protocol used between server and client is simple enough to be implemented by nearly every programmer as a training exercise within a few days. From and end-user perspective, this might be an irrelevant advantage, but from a technical view it is the very base for an enduring interest in the protocol’s implementation. Platforms whose communication protocol is technically difficult to implement will necessaryly have a shortler life span compared to such with simple protocols. The situation is similar with regard to Gopher, which long has been removed from typical users’ experience, but which is still enjoyed by technical minimalists. With Gemini there exist efforts to moderately modernise the protocol, which have a parallel in the IRC world with IRCv3.

The technical base however cannot be the only reason for IRC’s persistance, since even technically well designed protocols are not guaranteed to be adopted; in fact, it is entirely valid to critise IRC to not be up to date anymore in several aspects. The answer lies within IRC’s specific use, that is, within the IRC networks. IRC is one the very few media of social communication that profits from the Network Effect, an effect which is normally used to draw users away from free communication media into closed silos. Specifically Freenode was longstandingly known to host a chat channel for nearly every FOSS project, often enough even the project’s official real time communication medium. Even after a sadly growing number of FOSS projects decided to withdraw their official channels from IRC in favour of silos such as Slack, there are usually enough enthusiasts to be found still to run an inofficial chat channel in IRC, something that until recently normally meant: on Freenode. As a result, it was often sufficient to try on Freenode the channal names #project and ##project in order to find a community related to the project in question. Freenode thus grew into a kind of digital café: one went there with the intention to talk about a specific project, because one knew there will be persons around who are related to it. More often than not – including in the author’s case – there grew friendships from these informal meetings which extended into real life. Outside the FOSS community, it might be similar for other IRC networks. Other platforms were not able to establish themselves in a similar way; often enough their development was up front focused on other use cases like communication under friends and relatives, or in extreme cases only one-to-one communication. IRC on the other hand was from the beginning designed for meeting unknown, but likeminded persons. This approach is even mirrored in the terminology used: one does not participate in “chat groups” or “group chats”, but instead in “chat channels” or even more plastically in “chat rooms”.

First of all there is the question whether it is possible for Libera.Chat to establish itself as a competitor to Freenode at all. A plausible fear is that Libera.Chat is going to suffer from the same fate as LibreOffice: the volunteers disassociate themselves with the holder of the rights to the name and attempt for years to regain the old name’s (OpenOffice’s) fame. After twelve years the name LibreOffice is still mostly unknown outside Linux circles. If one takes this as measurement for Libera.Chat, then the network is on a year-long path of unknown usefulness, while Freenode still profits simply from its name’s popularity alone and continues to attract users even though its semantics changed. As of now, there are hints that this scenario will not come true. For this conclusion it is not required to apply far-fetched thought patterns as they surface in statements like “This channel was originally created … on freenode, and stayed with the network on 2021-05-19 after that domain was lost”¹¹ This cryptic statement is better explained by this Sysop statement: „The network is the people, not a domain name“. , but it is sufficient to take a look at the statistics and the exodus of prominent projects.

Even more, it appears as if Freenode’s new IRC operators are striving to thoroghly destruct Freenode. One has called the aforementioned netsplit an attempt of “suicide”. This alone may not be convicing (Freenode’s staff views their actions generally as a path to take IRC further), but the moderational qualities shown by Freenode’s new operators are an entirely sufficient guarantee for emptying the network from users. The cases described above – the freenodecom bot and the manual forced handover of #fsf – show that the new Freenode staff lacks leader qualities. In both cases not only was the technically hardest sanction (K-line resp. forced channel takeover) applied within a short timespan, but this moderational action was conducted without any following of an escalation process, namely without any previous warning. This is not a professional way to deal with breach of network policies, especially given that these policies were changed just shortly before. Thus, one has rightly critisised that Freenode’s new operators are abusing their power. As if that was not bad enough, these abuses are executed in a way to openly show the technical power available to the IRC operators and display how much fun it is for them to exercise this power against those who cannot withstand. The aforementioned #fsf Chatlog can serve as reference to this behavior (nicks with staff mark and the root nick are Freenode operators):

2021-06-13 06:04:23 → TheRedHorseman (~bagira@freenode/staff/phanes) has joined #fsf
[…]
06:43:07 → Foxy (znc@freenode/staff/foxy) has joined #fsf
06:43:46 → job (job@gateway/shell/tilde.team/x-dhtjyelnzoocvubg) has joined #fsf
06:44:04 → root (root@internet.relay.chat) has joined #fsf
06:44:19 → azend (~quassel@unaffiliated/azend) has joined #fsf
06:44:20 job            oh no
06:44:26 root           sudo su
06:44:35 TheRedHorseman o/
06:44:37 ← fling (~fling@fsf/member/fling) has quit (Ping timeout: 245 seconds)
06:44:50 Foxy           \o
06:45:02 * TheRedHorseman trots around in a circle, eyes gleaming red in the dark
06:45:16 → altwulf (~altwulf@64.227.64.196) has joined #fsf
06:45:46 ← wolftune (~aaron@75.164.131.0) has quit (Quit: Konversation terminated!)
06:45:53 → wolftune (~aaron@75.164.131.0) has joined #fsf
06:45:57 -- Mode #fsf [-o ggoes] by OperServ
06:46:50 ggoes          never a dull moment
06:46:56 bandali        ^

What happens subsequently is unworth of publically citing, but this is already enough to see how one does not approach a conflict. This is a way to demonstrate to others that one finds them inequal. It fosters a climate of fear which nobody is going to enter into voluntaryly. Therefore, it is not far-fatched to attest Freenode to be on the way to turn into a kind of digital Khazad-dûm. Different from how Oracle dealt with OpenOffice, the existing reputation of the name Freenode is systematically damaged. That actually will benefit Libera.Chat alone for the reason it is not Freenode with its negative associations.

Until May 19^th, 2021, Freenode was the largest IRC network still in existance, which provokes the question if its destruction is equivalent with the end of IRC as a whole. Some projects have decided to not migrate from Freenode to Libera.Chat or OFTC, but instead to entirely different media; NixOS for instance is migrating to Matrix. The FSF and the GNU projects on the other hand specifically decided against this option, because Matrix did not satisfy their requirements of Free Software use²² Interestingly they also decided against XMPP, but with another argument: it was found to not provide sufficently more features compared to IRC to justify the switch of the communication medium. . Given the difficult development situation with Matrix (see above II.1) this probably is a sensible decision. If however a sufficient number of projects take the Freenode breakdown as an incentive to turn down IRC entirely, then May 19^th, 2021, will be recognised as the beginning of the end of IRC in the Internet’s history.

Something one should also be worried about is whether the decisions by Freenode’s new IRC operators damage the reputation and reception of IRC as a whole. The hacker community is traditionally seen as difficult and one requires a rather thick skin in order to deal with it. Torvald’s public outcry against NVIDIA exemplarifies this. The name Freenode was for about twenty years synonymous with the place one enters into if one wanted a faster and more informal discussion about a FOSS project than it was possible on the project’s fora and mailing lists. If this place now is lead by a narcisstic IRC operators team, this could damage the reputation of both the medium IRC and the hacker community as a whole. This is something that cannot really be excluded at this point. It can only be hoped that either Freenode’s new IRC operators recognise their responsibility or they complete their destructive work consequently enough to create a gap that Libera.Chat is able to fill before the larger building collapses.

Currently it looks as if the latter scenario becomes real. Although the exact reasons leading up to the events of May 19^th remain in the dark, one should recognise the Libera operators’ talent. Just on the very day that Lee uses his newly gained power on Freenode, nearly the entire staff team not only resigns, but immediately can point to a nearly fully functional alternative network. This clever double action has generated a noticable news media adoption, which has moved the aging medium of IRC back into the light of the public once more. This has had immediate consequences: those who followed the chats happening in the first days of Libera.Chat in the meta channel #libera could observe many people who were in first contact with IRC at all or at least since a long time and who were astonished by the medium’s vividness. Whether this effect is of lasting duration is something yet to be seen. In any case, the IRCv3 project should try to leverage this unexpected new attention which is given to IRC.

IRC has seen many other protocols and services come and go and it is sensible to predict that the events surrounding Freenode are insufficient to take the IRC protocol as a whole out of order.