A discussion about distributed federated systems else-network did bring up some concerning history that we should keep in mind:
#XMPP was, briefly, the dominant messaging system on the Internet. It was fully federated allowing hosting anywhere. Major extensions were being done by major companies (Apple adding voice and video for example). And it all fell apart because the three big providers (Apple, Google, Facebook) ditched it to have more control.
So if a major company does end up adding support for ActivityPub or other fediverse features I would be very hesitant about actually federating with them. As long as they DO connect to the larger network they'll siphon off the user base, and we KNOW that they will turn on it. Corps cannot make promises, cannot be trustworthy, cannot guarantee they won't betray principles. They are structurally incapable of it, no matter what their humans say.
@iarna I heard something maybe tumblr, but also they might be going open-source, which would be cool; I don't think just adding activitypub and calling it a day is necessarily right, definitely needs the open component so users can move & choose
If Twitter disappeared off the face of the planet out of nowhere and Tumblr went open source within the same timeframe that would change the entire landscape of social media overnight
@iarna We've seen this so many times with standards and BigCorp. Embrace, extend, pervert, monopolise, close. It often starts with an apparently well meaning attempt to support open standards and contribute FOSS code. Then Standard 2.0 is built by a committee of vested interests. Before you know it the whole software niche has been buried.
RSS/Atom - Google Reader, anyone? But also location services, friendfeed, blog search.
@iarna You're right on with that.
There are a million ways to subvert Mastadon growing. Twitter could quietly spin off a tiny company, setup their own autoscaling cluster with better marketing, instructions and interface, throw a couple full time admins and a decent moderator staff at it.
Start adding features outside the api... lock em back in
I feel slightly more positive this time around. In the time of XMPP and Google, everybody was looking towards Google and other big tech companies as validators of the technology, hoping they would adopt it. When Google and others dropped XMPP after some experimentation, people got disappointed and stopped believing in it.
This time around, people on fedi seem to feel that nobody needs any companies, and we'll make this a success all of our own. And if any bad companies show up, we'll defederate them on sight.
Whether that will actually happen remains to be seen. I see how reluctant people who run a mail server are to defederate from gmail. (Not that it doesn't defederate from them on the regular, without appeal or explanation.)
But yeah, hoping that This Time Will Be Different, with people fully aware of EEE tactics.
@coffee The most frustrating thing is they didn't actually even drop the tech, despite giving that as an explanation, they just dropped federation.
But yeah, I hope we've all learned some lessons this time around.
Yeah. What happened was that they dropped the standard, and took the tech they already developed in directions that were beneficial to them alone.
This happened even before federation was cut, as Google implemented its own voice/video calling mechanism, ignoring the fact that a voice/video calling standard already existed and was implemented by various open source clients.
So then clients had to implement Google's standard to be interoperable again, wasting developer's time. And that's about when Google started killing off federation.
@coffee Jingle (XEP-0166, amongst others -- the proposed extension that Google implemented voice and video with) was started in 2005. I'm not aware of any efforts to standardize voice and video before that. Some folks had implemented that sort of feature over XMPP previously but I don't recall any attempt to standardize it outside of Googles efforts?
Am I misremembering something here?
I think the prior art was XEP-0095, which was a Draft Standard from 2003 until 2017, when it was deprecated in favor of XEP-0166/Jingle.
Jingle only became a Draft Standard in 2009, and until 2017 there were (at least officially) 2 competing standards.
The other side of this is regulation, of course. The EU, the EFF, and many others are hard at work on a raft of laws that will hopefully prevent "embrace and disown" community theft. Mastodon being successful against a backdrop of Musk means it's so much easier to make these arguments to the relevant regulatory bodies.
@iarna I would be really surprised if any corporation would be eager to tackle moderation, a problem that doesn’t exist with XMPP.
@iarna Right, but no one attempted to moderate it. Corporations are unlikely to offer social media without moderation, and are also very reluctant to offer moderation due to the politically charged nature of it.
@iarna Google's next product to create, get people using, and then pull the rug out from under people?
Yeah, been there. Too many times.
@iarna my experience with XMPP was in a large part corporate IM servers in LAN, so that was isolated, not federated.
And appart of that, indeed the jabber - Google talk connectivity was brief and not used a lot (people in GTalk communicated mostly with other people in GTalk).
So maybe it's different with the Fediverse where an isolated instance is nearly useless.
@iarna as i remember XMPP die because of bad UX. Protocol had many extensions supported by _some_ clients and _some_ servers. So you cant say «XMPP support feature X», only «Clients A and B on servers C and D support feature X»
@iarna what we used to call E3 - Embrace, Extend, Extinquish approach. Microsoft used to be an expert player of that game. Then it became a triumvirate.
@townsend If there were EFF vetted legislation to accomplish that, I'd certainly feel better about it. I'm not holding my breath on that being reality in the near term however.
@iarna and how do you plan doing that? AP is by definition a pretty open protocol, I can actually listen to your public stream without registration or real efforts. You can only stop me talking to you, no more.
@iarna I agree; and it's not the opposite either.
What I'm saying is that there is no [feasible] way to stop them.
@grin I'm mean, sure? Like, there are protocol extensions one could imagine to make it easier to fully disconnect from servers you don't like who are otherwise well behaved, but no open network will ever be able to stop poorly behaved services from implementing that kind of attack through purely technical means.
@iarna apart from that there is the ethical aspect, whether we may call "open network" something blocking based on some agenda.
(And it is not "you don't like", it is "everybody must not like", kind of a big difference.)
The history is true, but right now the legal scenario is way more developed.
Now that the #DMA exists, I am very confident that degenerations such as the one you mentioned would be prevented.
@iarna I misread this as "cops cannot make promises, cannot be trustworthy..." and I was like "I'm not sure why the topic changed, but right on."
@iarna I'm seeing a bit of a push for non-tech orgs to run their own instances, which sound intriguing: NGOs, public broadcasters, universities, and the like would be (IMO) ideal hosts that wouldn't try to close off their systems in favour of proprietary protocols.
@iarna a similar thing happened with email: running your own server these days is quite the challenge because of the big corporations. Even if you follow all the best practices you can fall into a blocklist and have your email classified as spam without appeal.
I've been hosting my own email for 22+ years, and every email I send to Gmail will go straight to the spam folder, unless the person has emailed me first.
@iarna It is is worth noting that the laws are different today, than they were a decade ago. We've got DMA/DSA and chances are they will enable greater competition and prevent building walled gardens. I am cautiously optimistic.
@iarna I don't think that's entirely accurate. XMPP was a good standard, but far from perfect. No calls, no file transfers. Simple clients were not functional, and functional clients were too complicated. Without GCM support, the phone battery drained in half a day (it was implemented later, but it was too late).
Besides Google, Facebook and Apple (did Apple have an XMPP server/client??) there was LiveJournal, VK, Yandex (held on the longest), but all of them first pupated (disabling the federation or changing the protocol beyond recognition), then closed because no one used them.
What I mean to say is that there really hasn't been any kind of user migration from "free" XMPP to the closed network of corporations. XMPP was attached as a bonus to some service, whether it was a blog on LiveJournal, a mail on Yandex or an account in the social network VK. XMPP popularity grew rapidly thanks to new users, but just as quickly fell to its initial level, and came to naught with the arrival of mobile-first messengers, where the three pillars - messages, (video) calls and files were implemented steadily and unambiguously for all users.
@johan When launched, iChat was just XMPP, both server side and in peer messaging. (Yeah LiveJournal's support was, IIRC, not integrated into the site itself? Just a server you could use if you had an account there.)
@johan To clarify my thinking: I didn't see migration of "open" XMPP users to closed as a thing, Rather I saw it as interfering with recruiting for the open services. That is to say, if the only way to reach someone was to add a new chat client, folks were accustomed to that. The user experience of "just type in chat box on a site you were already using" is obviously superior.
@johan File transfer was available from early on, and I don't recall finding support difficult, but it -was- a long time ago, so I'm not super confident there. (Possibly not in the Goog/Facebook web clients?) Video/audio support was definitely more limited. The was largely pre-smartphone, though the video/audio spec getting implementations (in Google Talk initially) was borderline.
@johan But that was also shortly before those became standard features everywhere. I felt like XMPP use collapsed well before the dominance of modern mobile messaging.
@iarna ...I'm trying to remember how we communicated back then 🤔
Anyway, I used Jabber on my desktop even when there were no "live" people left. Because two popular (in narrow geek circles) microblogging services - Juick and Point worked fine (and still do) through a jabber bot. It is very convenient, you can keep several discussion threads at once, not bothering with a slow web interface 🙂
...And then we came to what we came to: everyone has WhattsApp, most people have Telegram, no one is capable of anything else.
@iarna File transfer was available from the beginning, and I don't remember support being difficult, but that was a long time ago, so I'm not very sure about that. (Maybe not in the Goog/Facebook web clients?) Video/audio support was definitely more limited. This was mostly before the advent of smartphones, although the video/audio specification that got implemented (in Google Talk originally) was borderline.
Well, for some, after specifying a server, login, password (ugh, hard!) adding a proxy server to transfer files was a real challenge. 🤔 Symbian smartphones were already quite there (yes, without a front camera), but internet prices didn't allow for voice traffic...
In any case, audio over Jabber is SIP bolted on sideways and has nothing to do with XMPP.
@iarna At that time, most clients were multiprotocol.
And Yandex had Jabber integrated into the webmail interface, so if your contact also had mail on Yandex - you could not exchange emails, but communicate in real time.
@iarna That's basically what I wrote: Jabber was an add-on to the main service. I don't remember exactly whether federation worked in LJ, but in any case it was possible to use via J2J transport.
@johan Uh, yeah, I can't recall either. I had an account on my own server so while I logged into it from a multi-client, I wasn't adding friends on it.
@iarna Jabber was great and the first really open federated messaging platform, I agree. But dominant? ICQ and AIM and other closed platforms had massive lasting user bases that XMPP never saw.
Other platforms like experimented with federating to XMPP but they had no incentive to commit to a platform that couldn't deliver their advertising.
@PCOWandre There were a few years where every Facebook, Google and Apple user could all chat with each any XMPP user. So yes, dominate. This meant that you could use an independent client and still remain connected to everyone you knew. ICQ was long long gone by then, and AIM/MSN/Skype were waning.
The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!