This blog post by Ploum, who was part of the original XMPP efforts long ago, describes how Google killed one great federated service, which shows why the Fediverse must not give Meta the chance
You must log in or register to comment.
Basically the sequence of events as claimed by the author is that:
- XMPP, niche, small circles
- Google launches Talk that was XMPP compatible
- Millions joined Talk that could coop XMPP in theory
- The coop worked only sparingly and was unidirectional, i.e. Talk to XMPP ✅ but XMPP to Talk ❌
- Talk sucked up existing XMPP users as it was obviously a better option (bandwagon effect + unidirectional “compatibility” with XMPP)
- Talk defederated
This demonstrated exactly the importance of reciprocity. If they play dirty, kick them out asap.
Seems like just another reason why defederation should be completely removed from the protocol. It’s way too easy to abuse and force centralisation.
There are other far less destructive and abusable ways of dealing with spam and content moderation.
I maintain that it’s better to give the users the control, and allow them to decide which instances, communities, and users they want to be exposed to. Bottom up moderation, instead of top down.
For example, instances can provide suggested ‘block’ lists (much like how an ad blocker works) and users can decide whether or not to apply those lists at their own discretion.
By forcing federation, the network stays decentralized. Maintaining community blacklists that can be turned on or off by the individual user protects against heavy handed moderation and censorship, whilst also protecting users from being exposed to undesirable content.
The case with XMPP is that Google Talk introduced addons and intricacies that were unique to them. So they could federate with you in full with additional bells and whistles while you were stuck in an eternal catch-up. They presented a better alternative regardless of the eventual defederation. Even if we have some viral clauses as in GPL in open-source software that ensures protocol compatible software to be compliant, we can only do that to a certain extent plus enforcement is always an issue. Who are going to spend the vast sum of money in court to defend the “federation”?
This aside, enforcing federation alone does not ensure decentralization. These zero-marginal-cost fixed-cost-intensive businesses of the internet has a tendency to centralize as serving one more seat costs no penny plus one more seat diluates the fixed cost altogether.
The Beehaw devs are feeling the need to defederate right now due to that being the viable strategy for scaling their mod efforts. What would the solution be if they could not defederate?
One key difference between link aggregators (kbin/lemmy/reddit/digg) and microblogs (twitter/mastodon) on the one hand, vs social networks (facebook/myspace/diaspora/friendica) and instant messengers (aim/icq/xmpp/signal) on the other, is that the latter is highly dependent on your real-life social network, while the former is not. People using instant messengers and people on facebook want to use them to interact with their friends and family, so they have to use the platforms that those friends and family are on. On the other hand, people are happy to use link aggregators and microblogs as long as there are interesting people and communities to follow, even if they consist entirely of strangers.
Back in the early days of XMPP, when it was still known as “Jabber”, I tried switching to it from AOL Instant Messenger. I told all of my contacts about it, and tried to get them to set up Jabber accounts. I was super excited that instant messaging was finally being standardized the way email was, and we wouldn’t have to deal with AIM vs MSN messenger vs Yahoo messenger vs etc. I think I was also still bitter about being forced to switch from ICQ to AIM because all my friends had switched. I don’t think I got a single person to start using Jabber, though. At one point I even declared that I was going to stop using AIM entirely, and that people would have to switch over so that we could keep talking to each other. Didn’t work, of course. I just ended up not being able to talk to anyone until I finally went back to AIM.
A bunch of my friends use reddit, but we don’t use the site to interact with each other in any meaningful way. This made switching to kbin really easy. Sure, I’ve told a few of them about it, but it doesn’t really matter to me if they switch or not. As far as I’m aware, XMPP never really became it’s own “thing” and experienced the kind of growth that the threadiverse has. Since we’ve passed the point of being self-sustaining, we can keep growing one user at a time, as individuals decide that they’re tired of reddit and make the jump.
Because of this difference in dynamic, we’re in a much better position against Meta than XMPP was against Google. The fact that we can even consider outright blocking Meta is a really good sign for us, regardless of whether we do so or not. Even if we do end up in a situation where 90% or even 99% of users are on Meta’s platform, we can still refuse to allow them to compromise the ActivityPub protocol. Attempts to “embrace, extend, extinguish” will likely just result in non-blockading instances joining the anti-Meta blockade. With the connection to Meta severed, we’ll just go back to enjoying the company of the 1 to 10% that remain, and that portion will likely be much larger than what we have now.
You are spot on. The difference between the products/services/values offered by XMPP and AcitivtyPub based fediverse is a very crucial distinction.
XMPP’s value is derived from its connectivity. It is bandwagon effect at work. A single fax machine makes no sense but what about another one? Or another 100 ones? Now you have a positive network externality.
The bulk of the AcitivtyPub based fediverse works very differently. The value is from the content, be it people shitposting or memes or cats. As people who frequent online forums and communities can tell, the majority of members are mere readers. They are content consumers. Content producers are often the minority. The reason why soneone will stick to a particular platform is because of the content and the expectation that more is coming.
I’m not sure the distinction would make enough of a difference, and focusing only on XMPP might be doing yourself a disservice. There was nothing social about Office, but the OP points out how the same strategy worked there as well. Users, overall, tend to go where the other users are. Some people left Digg for Reddit because they were unhappy with Digg, but the vast majority simply followed because it was where the users (therefore activity) went. Reddit wasn’t even the best of the many options at that time; what was important was the inflow of users. Once that kicks off, others tend to flock like moths to flame.
As you point out, Reddit was not where you interacted socially, yet it became where you congregated because that was where everyone else was and therefore where the easiest access to content and engagement was. If a Meta product becomes the most popular way to consume ActivityPub content, and therefore becomes the primary Source for that content, independent servers will become barren with just a Meta Thanos-snap of disconnecting their API. They only need to implement Meta-only features that ActivityPub can’t interact or compete with, and the largest portion of users will be drawn away from public servers to the “better” experience with more direct activity. (And that’s without mentioning their ability to craft better messaging, build an easier on-boarding experience, and put their significant coffers to work on marketing.)
Sure, there will still be ActivityPub platforms in the aftermath. Openoffice/Libreoffice still exists, XMPP clients and servers still exist, there are still plenty of forums and even BBS systems. But, there is a reason why none of those things are the overwhelmingly “popular” option, and the strategy they will employ to make sure that happens is the focus of the article, not so much XMPP.
Can anyone with expertise explain the structural difference between Matrix and XMPP?
In short: Embrace, start pushing the service, driving users to it. Expand: add non standard extentions, locking users onto your quasi-compatable version. Extingish: break compatibility entirely, preventing users from swiching to the fully open version.
Okay, this is a good article. I was on the fence about Meta, wondering how they’d cause any damage, and this article cleared that up for me.
Same. I was familiar with “Embrace, Extend, Extinguish” but somehow still didn’t get it until Ploum explained it to me slowly.
Great post by Ploum. Really sheds some light on how vicious these things can be and how federation will have to really push for openness and freedom.
The X in Xmpp is for extensible. I find issue that a protocol that is supposed to be extensible was killed by being extended.
Excellent read. Have just re-posted this on 'key, thanks for sharing this.
I agree that Meta is doing something very dangerous to the fediverse… hope they could be stopped in their tracks.An excellent read. My synopsis is that if any big corporations joined the Fediverse they would fracture it, and that no matter what Meta, Reddit, Google, etc. would never want to see a decentralized platform succeed.
Pretty much the Fediverse needs to never let a big company tie into it. Our group needs to work at growing but at a sustainable rate.
@MyMulligan @jherazob I disagree.
I think the key thing is to just make sure that you don’t use non #FOSS clients. #GoogleTalk started as a client for #XMPP, people migrated to it, and then #Google dropped support for #XMPP. If so many people didn’t use #GoogleTalk, the #XMPP network would have remained unimpeded.
At this point it wouldn’t matter, all they need to do is to mess with the protocol and it’d achieve the same thing, Meta and everything in it’s sphere would “work well”, but connecting with true ActivityPub servers would work just glitchy enough to annoy their users and point the fingers towards our side, just like it happened with XMPP
what the heck makes you think that all the #Fediverse users are just gonna leave for #Meta because federation with it is “annoying”? lol
I wasn’t talking about our users, i was talking about theirs, a direct mirror of what the author described with XMPP
Correct, it’s worse, you can very much argue that Google had good faith intentions, you cannot even pretend that Facebook does while keeping a straight face
What’s missing from the article is an actual explanation of how Google “killed” xmpp. Did google force the independent XMPP client developers to not implement cool features or something? Is meta going to buy up and shut down all the independent mastodon instances?
If the problem is that Facebook might develop a superior UX, maybe the fediverse should work on a better UX instead of screaming about some scary boogeyman and how the users are too dumb to know any better?
google added things that only they knew about, which broke the other XMPP apps, driving people to install google talk instead. Sure, eventually people figured their shenanigans out and their apps started working again but the damage was done. Repeat that a few times so that most people were using google talk, then flip the switch and everything that’s not google talk was basically a ghost town. People don’t stick around in ghost towns
Is there some way to work a limitation into a licence? Something around only being able to present federated content with included algorithms. That would instantly make it unattractive to all the big players who profit off their specific ad driven algorithmic feeds.
Perhaps the various concensus theories and mechanisms that came out of crypto could somehow give inspiration on ideas to protect this service from the shitty financial actors that come in and ruin all of the good services.
I’m not saying actually using crypto, just maybe some of their concensus mechanisms/ideas for preventing bad actors could be put in place.
I’m gonna throw this out there:
If Meta is going to join the fediverse (or implement something with activitypub) there is absolutely nothing we can do to stop them.
It’s an open protocol. They can use it.
The only thing we can do is force them to follow the AGPL and/or fork the code if they get crazy with change requests.
deleted by creator
Bing AI summary:
The blog post “How to Kill a Decentralised Network (such as the Fediverse)” by Ploum discusses how the GAFAM empire controls the internet in 2023, except for a few small villages that resist the oppression and form the "Fediverse"¹. The Fediverse gains fame and attention through debates around Twitter and Reddit¹. The post also discusses how capitalists are against competition and how Facebook has been careful to kill every competition by buying companies that could become competitors¹. However, the Fediverse cannot be bought because it is an informal group of servers discussing through a protocol (ActivityPub) and running different software¹. The post also discusses how Google made XMPP irrelevant by joining the XMPP federation¹.
Source: Conversation with Bing, 6/23/2023
(1) How to Kill a Decentralised Network (such as the Fediverse). https://ploum.net/2023-06-23-how-to-kill-decentralised-networks.html.
(2) How To Kill Poa Annua (Annual Bluegrass): Your Step-By-Step Guide. https://www.domyown.com/how-to-kill-poa-annua-grass-a-572.html.
(3) How to Kill Clover in Your Lawn | Scotts. https://scotts.com/en-us/how-to/how-to-kill-clover-in-your-lawn.html.Those sources… are… interesting…
AI will replace all of us at our jobs.
With Bermuda grass.
I hope people, especially instance owners and devs, listen to this warning.
Unfortunately, the fact that lemmy isnt in the pact to not federate with meta means history will repeat itself
Even then, i doubt theres enough strong willed people to actually resist big payouts in exchange for access to the federation. Its truly pathetic how people are easily convinced
There’s nothing “pathetic” about it. You’d have to be stupid not to take a good deal. That’s like not scoring a goal while the goalie isn’t paying attention, because you think it’d be unfair.
Solving this problem isn’t the responsibility of individuals and their “willpower”, it’s the responsibility of governments. It’s their job to regulate markets and ensure fair competition. Getting mad at some sysadmin who accepts money from Meta might make you feel better inside, but it’s not going to change anything. Do your research, write to your representatives, threaten to vote them out, and then actually show up on voting days.
Aye great read and very illuminating. We gotta protect the fediverse from corporate insidious destruction. This quote stood out to me:
And because there were far more Google talk users than “true XMPP” users, there was little room for “not caring about Google talk users”. Newcomers discovering XMPP and not being Google talk users themselves had very frustrating experience because most of their contact were Google Talk users. They thought they could communicate easily with them but it was basically a degraded version of what they had while using Google talk itself. A typical XMPP roster was mainly composed of Google Talk users with a few geeks.
In 2013, Google realised that most XMPP interactions were between Google Talk users anyway. They didn’t care about respecting a protocol they were not 100% in control. So they pulled the plug and announced they would not be federated anymore. And started a long quest to create a messenger, starting with Hangout (which was followed by Allo, Duo. I lost count after that).
But XMPP users were presumably still around and outlasted Google and their apps. We’ll be the same even if Facebook churns the protocol, because the whole point of being on Mastodon or KBin is to not be on Facebook.
you missed the point where the open source devs were in a constant race to adapt to all the google-“innovations” and actually troubleshoot on them which ends up demotivating
did Google force them to do that, or did the open source devs just make a mistake?
So how do you know who to trust?
