[Edit: I see the problem, even with a self-hosted instance of 1, when you comment on posts in other instances that data is no longer held on your server, so you don’t own it and can’t control it directly, is that right?]
So as I understand it the big “advantage” of ATProtocol is the account portability via DID, however this is at the sacrifice of actually hosting an ATProtocol being extremely data heavy.
This has made me very curious about self-hosting ActivityPub (meaning an instance of 1 user), it would seem like focusing development on a client that makes it as close to as easy to self-host an instance as it is to join one would solve the issue of accountability portability, as you literally own all the data and rights when you self host. The Major challenge I see there is security, where experienced admins for larger instances should have some level of cybersecuroty expertise while the average use may have little to none. But then focusing group effort on auto-updating the client and the default settings of the client to maximize security would solve that issue it would seem?
So what am I missing? Other than hosting costs, what else is deferring a self-hosted-first development approach for the Fediverse?
Is it actually that AP development fundamentally believe moderation should be handled at the admin/instance level, and self-hosting makes moderation more difficult and less directly authority based?
Publicly shared blacklists and whitelists would seem the natural fit for a self-hosted-first network, akin to adblock and horizontal.
Edit: I see the problem, even with a self-hosted instance of 1, when you comment on posts in other instances that data is no longer held on your server, so you don’t own it and can’t control it directly, is that right?
The “problem” is also going the other way around. If a large instance is sharing data with your single user instance, they ultimately cannot control what you do with it.
Isn’t that a non-problem? I mean, afaik, you cannot control any data after it’s sent. If I send you an email with an attachment, I cannot stop you from forwarding it to everyone you know, nor can I stop you from editing it before forwarding. I guess it becomes a problem of reliability, then, which is only a problem if the original source is unavailable - which is not that uncommon in the fediverse.
It is a “problem” if your goal is full ownership and control over your user interactions and data footprint. ATProtocol in principle achieves this, you can delete or move your account and all your user interactions will be deleted or follow you (meaning edit permissions, full access to all the posts/comments, follows, updoots, etc.). The cost of this is a very high data load for the host, though I’m still not clear on exactly why it is so much more data intensive, is it the size of lookup instructions between hosts in addition to the actual markdown?
[Edit: I see the problem, even with a self-hosted instance of 1, when you comment on posts in other instances that data is no longer held on your server, so you don’t own it and can’t control it directly, is that right?]
So as I understand it the big “advantage” of ATProtocol is the account portability via DID, however this is at the sacrifice of actually hosting an ATProtocol being extremely data heavy.
This has made me very curious about self-hosting ActivityPub (meaning an instance of 1 user), it would seem like focusing development on a client that makes it as close to as easy to self-host an instance as it is to join one would solve the issue of accountability portability, as you literally own all the data and rights when you self host. The Major challenge I see there is security, where experienced admins for larger instances should have some level of cybersecuroty expertise while the average use may have little to none. But then focusing group effort on auto-updating the client and the default settings of the client to maximize security would solve that issue it would seem?
So what am I missing? Other than hosting costs, what else is deferring a self-hosted-first development approach for the Fediverse?
Is it actually that AP development fundamentally believe moderation should be handled at the admin/instance level, and self-hosting makes moderation more difficult and less directly authority based?
Publicly shared blacklists and whitelists would seem the natural fit for a self-hosted-first network, akin to adblock and horizontal.
The “problem” is also going the other way around. If a large instance is sharing data with your single user instance, they ultimately cannot control what you do with it.
The same is true if you just scrape the website.
Isn’t that a non-problem? I mean, afaik, you cannot control any data after it’s sent. If I send you an email with an attachment, I cannot stop you from forwarding it to everyone you know, nor can I stop you from editing it before forwarding. I guess it becomes a problem of reliability, then, which is only a problem if the original source is unavailable - which is not that uncommon in the fediverse.
It is a “problem” if your goal is full ownership and control over your user interactions and data footprint. ATProtocol in principle achieves this, you can delete or move your account and all your user interactions will be deleted or follow you (meaning edit permissions, full access to all the posts/comments, follows, updoots, etc.). The cost of this is a very high data load for the host, though I’m still not clear on exactly why it is so much more data intensive, is it the size of lookup instructions between hosts in addition to the actual markdown?