Olives @olives@qoto.org

@jdp23 https://privacy.thenexus.today/steps-towards-a-safer-fediverse/

I could go further at length more privately, but I'm becoming somewhat sceptical of blocklists, and even the idea of blocking instances.

In principle, it makes sense, someone might want to block harassment. Sounds good, right? But, it invites a certain degree of centralization, and you end up with the problem of the "list curator" (it seems to be a guy who is described by a few as "that guy who runs an instance with ten users") who might also have other values (i.e. their own streak of puritanism, which could even be described as a form of colonialism, even if they rationalize as reasonable within their own personal belief system), and it might also involve various shades of admin toxicity / vibes. It seems it might be more of a "list clique" than a "list curator". And in the end, it creates another centralization of power, when someone is *trying to get away from that*.

The opposite reaction I've seen to this from a few is to be more provocative and to give a finger to them entirely (I'm speaking metaphorically here). I don't really think that is sustainable though. It is quite vexing. Even in other cases, it is still unilateral, typically people don't talk to each other (or not federate a very narrow band of content), instead a "list clique" determines unilaterally that someone doesn't belong on the "fediverse" (which is almost seen as someone's patch of turf).

This isn't actually even much of a post about what someone should do, even, it just outlines the sort of problems it runs into.

When it comes to things like Matrix, to a large extent, the moderation is shifted to the groups which operate on an underlying "infrastructure" (which avoids many of the problems you get when the "community" is welded to the "infrastructure" with Mastodon). Accounts are also not dependent on these particular groups. I've seen a similar property with accounts from Nostr (which goes about it a different way), although I'm not sure I like that one. With Matrix (and Cwtch, I think), you can also require an "invite" before someone can communicate with you, and in fact, if you don't have someone's "address" / handle, then all bets are off entirely.

https://cactus.chat We've also seen the Matrix protocol even being used for things like comments. This might be a sign that the Matrix protocol could be used for use cases other than pure chats. If your goal is "privacy by design", then maybe, ActivityPub might not be the best tool for that. Though, I'm not pointing to Matrix in particular.

Invitations might be an interesting idea for DMs in Mastodon, the UX for current methods of limiting DMs are also not great (i.e. telling someone that someone else's DMs are on private, so they don't waste their time).

When it comes to spam, what gets tricky is that anti-spam is usually more effective at hitting humans, than it is at dealing with bots. Though, I've come up with a few ideas, such as perhaps allowing five identical posts, before rate limiting that for a bit, when it is a very new account. I've seen stranger ideas for a different system (cwtch.im, which is more about chat), such as Proof of Work (PoW) algorithms to make spam less economic [1] from security researcher, Sarah Jaime Lewis. This would consume resources though (and mobile devices in particular might already be lacking in resources), so I'm not that keen on it.

The idea of more conservative approaches to federation is an interesting one, though for some use cases (i.e. if personal safety is a high priority), I wonder if federation is a good approach. That said, solutions like "block this or that instance" are not really sustainable (it also seems someone doesn't understand 2251?). The past several years have made that clear enough.

In the end, when you're pursuing an open federated network, then fighting against the fundamental nature of that network in various ways is probably going to undermine it's utility, and create all kinds of issues.

1 https://openprivacy.ca/blog/2019/12/03/Incentivizing-Trustlessness-ZcashFoundation-Donation/