Aside from the usability learning curve he talks about, the other very pointed criticism is scalability problems that instance owners face. From what I understand, Lemmy and Mastodon are both similar in that they use the ActivityPub protocol. Could Lemmy get too big to scale and still be decentralized?
Source: “I used Mastodon for almost a year, just trust me bro.”
He seems to know what he’s talking about and hosted his own instance for a long time, why specifically do you distrust the author?
For his argument, length of use doesn’t matter. He’s more concerned with the new user experience, which can be determined on the order of days.
His criticisms are valid. Getting started on the Fediverse sucks. Let’s accept the criticism and get on fixing it
To piggyback on that Linux analogy: if ActivityPub becomes as essential of a backbone to the internet as we know it as Linux servers are then the transformation has happened IMO.
There’s a key difference. ActivityPub is a user-facing technology, in terms of the way data is exchanged.
Linux as the backbone of the internet is not user-facing.
The average user doesn’t care or even know if some web server is using Linux or not. The average user does care if it’s difficult for them to find and see interesting content on their content consumption platform.
We need to improve the new user experience and ways to introduce users to content they may care about. We need a way for instances to have awareness each other and what communities are available without user-initiated federation. Maybe we need some semi-federation by default. Or maybe we need some global index that all instances can share. Ideally, we should have some kind of recommendation system at the community level to help users find communities they may be interested in.
Yes, those are good points!
As long as we make it clear to users “you can interact with content on other servers independent of your choice” and recommend instances in a way to distribute load, I think we can manage.
Something like recommending instances based on location or main content focus could also help.
The author is directly saying that the more instances you have, the higher load on each of them. Because they’re all replicating and sending traffic to each other. Then again I don’t know enough to verify that, and Lemmy seems to be working just fine for me for the past week.
There are technical ways to solve this. If we can identify some way for instances to determine users’ interests and allow instances to query based on interests, that can reduce load. We can do all sorts of caching to reduce load even further.
It complicates the system, which is a risk, but it leads to a better user experience and better performance.
We just need the right people with enough time and desire to implement these kinds of technical solutions.
Guess it’s time for me read the ActivityPub spec.
