💾 Archived View for idiomdrottning.org › everyone-must-see-everything captured on 2024-08-18 at 19:30:17. Gemini links have been rewritten to link to archived content
-=-=-=-=-=-=-
Alex kindly replied to my confusing Usenet essay and then even more kindly allowed me to try to clarify by hosting my reply right there in his blog. Thank you for that. When I sent to him I said that I was also gonna post the last li’l bit of it here since I thought it had some ideas that I wanted to say even outside of the original context.
Alex’ 2024-07-02 reply: The importance of moderation
There are two kinds of protocols. Those where everyone must see everything (like Antenna and CAPCOM, and Twitter), and those where that’s not the case (like ActivityPub and email).
Here are three network layouts:
First we have a centralized network. One hub server and then every client is a spoke connected to that center core. The most efficient way to organize an “everyone-must-see-everything” protocol, but vulnerable to disasters like meteor strikes, nuclear explosions, billionaire takeovers, or earthquakes.
Then we have a decentralized network. Several hubs connected to each other, mirroring each other. Usenet worked like that. The hubs need to be just as beefy as the ones in a centralized network. This has some advantages like redundancy and robustnesss, and it’s more politically appealing to the anarchists that built the internet. (Not sure why since there's still two tiers and a mod group that rules everything for everyone.) It can be wasteful. It can handle everyone-must-see-everything protocols. It can also really shine when everyone doesn’t need to see everything, like IRC which is super efficient. IRC is a brilliant protocol design that, like a mailing list, only sends the messages to a server that that server needs.
Finally we have a distributed network. No hubs. It looks more like a mesh where every machine is connected to only a handful of “nearby” machines. This type of network is an absolute disaster for the everyone-must-see-everything scenario! Redundant to the point of wastefulness. That’s a future we must fight tooth and nail! And that’s exactly where the relay-centered version of ActivityPub was headed, and that’s what would be “needed” for hashtags-as-groups-and-discovery to “work”.
Here’s where we the grassroots federalized ancom degrowth community need to check ourselves before we wreck ourselves because even though energy and e-waste is underpriced, a capitalist for-profit silo company would never in a million years set up a system like this because it’s too wasteful ever for them, while we’re liable to do so since each instance op only sees their own costs, not the cost for the network as a whole. The megacorps are OK with centralization because for them, the power consolidation in the first two network types is a positive, while for us it’s a huge negative so we’re drawn to the third type, so we need to be really careful and really responsible here.
But as bad as this layout is for everyone-must-see-everything, it’s really awesome for protocols where everyone doesn’t need to see everything. It’s economical, robust, cozy. ActivityPub (without the relentless mirroring) is a good fit here. I can follow a handful of users and be connected to their instances and only their posts are sent to my instance, I don’t mirror the entire server. It’s actually pretty awesome! It’s good for moderation too because I only need to moderate my “neighborhood”. Yeah, a “neighborhood” is the perfect analogy for this; neighborhoods that can and overlap but are still small and managed. A system set up like that can remain smol and cozy even as it invites everyone on Earth. It’s great. ActivityPub is awesome.
