Just for the record: what Google/FB did to #XMPP was not 'embrace, extend, extinguish'. I bring this up because people are focusing on FB developing a divergent protocol/extensions. Maybe that's a valid concern, but it's unrelated to XMPP.
While Google Talk was actively developed, Google folk actively participated in the community. In fact they contributed important extensions such as Jingle, the protocol we still use today for audio/video calls in XMPP, and other bits and pieces.
The big problem was simply that these were the largest nodes on the network. They had so many users, they had nothing to lose by putting up the walls and shutting out what was (to them) a small minority.
The answer to this problem is nothing to do with the protocol, but just ensuring that the network is diverse and distributed, not centralized in one or two seats of power. Now how to actually do that, is a harder question. But it's the one we should be discussing.
@mattj legislation is one possible route, especially in the EU. But I'd prefer to see a self-organising network of cooperatively owned and democratically governed nodes.
For example, when I started work on Snikket, a project to make XMPP easier, I deliberately chose to focus on helping people establish small servers, based on existing social relationships. I specifically did not want to become just another large public service, partly due to what happened with Google Talk.
A large network of small nodes is more robust against disruption. It's what any healthy decentralized network should be aiming for. Applies to XMPP, Fediverse and the internet in general.
And to circle back to the start of the thread: if we can avoid reaching this super large node situation, the ease of an 'embrace, extend, extinguish' attack on the network protocol by such nodes is diminished significantly.
@mattj I'm confident that the Fedi has sufficient critical mass in terms of users, instances, community spirit and principled contributors/developers. Whilst facebook's possible foray into the Fediverse must be shunned for what they stand for, and will always strive towards, I am not concerned.
@mattj Whilst I really do agree with this argument, I've been a member on a number of small networks for federated services, and more than a few have closed down with little or no warning. Kinda makes it difficult sometimes to keep the faith.
@ChrisWere That's why I focus on existing social groups for Snikket. My family Snikket server is of value to all of us. If I was running the service for strangers, I'd be less inclined to keep it running when times got hard. But yeah, I hear you.
Note though, large services are far from immune to shutdowns. Happens to commercial and non-commercial services all the time.
This is what we need in matrix and the fediverse too. Accounts - and particularly people's data - being stuck on one instance is a weakness of both networks.
It would be great if all posts to the verse signed with a key that would then allow me to reliably identify myself as the author. And rebuild my posting history after a sudden instance collapse, by pulling copies from any instance where people were following.
For those who are just here to chitchat and share ShowerThoughts, their history doesn't matter. Some even auto-delete it after a couple of weeks. But for those of us who work here, it's a useful record.
@mattj I don’t really fear Embrace, Extend, Extinguish. The far more successful attack method is Divide and Conquer. Or in the case of Google Talk it’s more like Divide and Ignore. That’s what could happen to the #fediverse too.
@mattj I very much agree with you, but with my dismal scientist hat on, I have to say that cost efficiencies push towards centralisation. You can run one 10,000-person server more cheaply in both people and tech than 10 separate 1,000-person servers that talk to each other. So any federated system that wants to continue to be one needs to take this into account, and build in incentives to stay with small servers.
Add comment