Centralisation of communications media is an extremely powerful force.
The typical remediation for centralization is regulated public utilities is breaking up a monopoly into a cartel of 3-5 local monopolies.
Also not great.
We are participating here at a very rare occurrence of decentralisation.
For everyone who thinks this is important for our lives and for the world, it is incumbent on all of us to build structures now that hold this ground as the federation grows.
I highly recommend the excellent book "The Master Switch" by Tim Wu.
It details the initial decentralization, followed by centralization and monopoly, of various media from telegrams to film, radio, tv, telephones, cable TV, and the Internet.
It's a fascinating read, and well worth your time.
One of the big mistakes people make, over and over again, is relying on technological determinism.
That is, thinking the architecture of the technology will preserve the topology of the network.
Mastodon is Open Source. It's built with open standards.
This is necessary but not sufficient to keep the network decentralized.
We're going to need social and legal structures, plus cultural norms, that counterbalance Metcalfe's law, which pushes the network towards centralization.
@evan I am not sure. Mail for example converged to a handful of big providers and a long tail of smaller ones.
Is not Mastodon heading is this direction?
@ks yes, unless we put in place social and cultural mechanisms to avoid it
@evan @ks I think you are exactly right about where this ends up (large, regulated players) in the absence of some counteracting forces - but “social and cultural mechanisms” seems incredibly vague. Is there any precedent that shows communities can self-organize out of network effects rather than fall prey to them?
@markallerton @ks not that I know of. :(
@markallerton @ks let me flip it: what non-technical things would you do to keep things decentralized?
I do agree that centralization is coming to the fediverse, but not for any of the reasons that most people on here think. That centralization is coming for the exact same reasons that centralization came to email, and it's a reason that many folks that would like things to stay decentralised keep ignoring.
And that is user safety.
A lot of Mastodon fans keep pretending that Mastodon is inclusive. It's not. But it could be.
As a Black person, simply signing up for a Mastodon account can expose you to vile racist slurs and threats of violence. Most Mastodon users are one popular toot away from discovering that their instance mods are either unwilling or completely unprepared to deal with this.
Because a centralized whitelist was abused in a cynical attack years ago, the fediverse kinda gave up on that idea, and has been very resistant to it ever since.
It's entirely possible for decentralized instances to provide safety, but most don't/won't. I'm super happy with hachyderm.io for example.
But a larger company is going to integrate with the fediverse, and fulfill the most basic user feature request: "As a user of your product, I would like to know that signing up will not expose me to death threats from nazis"
Then more new users are going to flow there.
The best thing to do to "counter" this coming centralization is super easy to do, but from my short time observing here, it won't happen:
1. There should be stricter criteria for an instance being listed on "join Mastodon." Insufficient moderation gets you de-listed. Handle cynical false reports.
2. It should be easier for a new admin to just check a box and opt-in to a whitelisted federation that excludes the worst instances.
@mekkaokereke I agree. I think JM is a great structure for making default policy for the network. Requiring active moderation and use of shared allow/deny lists is a great criterion for entry.
I also dislike a shared allowlist. It shuts down growth at a time when we need to be reaching more people than ever.
There are 12000 Mastodon sites up, according to some estimates. There are 150 sites on the Rapidblock list. I think a blocklist makes more sense here.