Because lemmy is federated, it’s not decentralized. Instances run on centralized servers, using DNS, they can get deplatformed at any time and delete your data. They effectively work just like regularly centralized websites, and can block each other. Whereas on plebbit, each community is a node that can’t get deplatformed (works like torrents, ie no domain/DNS/SSL) and users connect to it p2p. So, to run a lemmy instance, you have to run a whole site, whereas to run a plebbit node you just have to open the desktop app and browse the site with it. Creating a sub with your node is free, just like creating a torrent file.

old.reddit was open source, also seedit is fully open source under GPLv2 license and it can’t get taken down (it’s serverless, and it’s hosted on IPFS)

edit: also, for example, there’s old.lemmy.world

Yeah, we need better communities, most of them are poorly maintained at the moment. But some of them are cool, like technopleb.eth, plebpiracy.eth, plebmusic.eth, movies-and-anime.eth. You can check out the full list on https://seedit.app/#/communities/vote (this is a maintained list of default subs to show in the app, but you can connect p2p to any sub whatsoever if you know its address, just like you can download any torrent with a torrent client)

spam

each subplebbit has its own admins, who set up an anti-spam challenge which gets sent p2p to users when they publish to the sub. The cool thing is these challenges can be anything that can be code (anything: including PoW if they want to get spammed, or SMS auth, a captcha, a whitelist, a password, a time-based or usage-based challenge, biometrics to fight AI like worldcoin, whatever regularly centralized social media sites will end up using to fight spam)

csam

all data on plebbit is text-only, you cannot upload media. All media you see is embedded from centralized websites, with direct links, meaning if you post a link to csam from some site like imgur, imgur will ban you, take down the media (the embed returns 404, media disappears) and report your IP address to authorities. Plebbit is also not private, it works like torrents, your IP is in the swarm (even though the app and community can’t see it, authorities can track it and figure out what you seeded, just like with torrents)

Seedit is fully p2p, it’s not federated, meaning there’s no “instances” in the federated sense. You just download the app or go to seedit.app, connect directly to a community, using its address, and you publish to it. It’s like torrents, you’re also seeding back content you receive (that’s why it’s called seedit, but seeding is fully automated at the moment, maybe it will be possible to selectively seed communities in the future)

https://github.com/plebbit/plebbit-js

https://github.com/plebbit/plips

nothing prevents it, the sub owner can put a challenge that’s impossible to solve to troll people. it’s required that this be possible otherwise the sub owner wouldnt have full control over what the challenge is.

a lemmy instance could do the same thing so it’s not really an issue, the fix is just dont use subs / instances that dont work.

It’s stored in each plebbit node. Each subplebbit runs a custom IPFS node for plebbit, with its text-only database, which is the content you see in the app. Peers download it and seed it back.

You can’t encode base64 images on plebbit, each fiels has a character limit. Obviously centralized links, from which media is embedded, will be taken down by the relative centralized website.

there’s no 🍕 because ALL data on plebbit is text-only, you cannot upload media. We did this intentionally, so if you want to post media you must post a direct link to it (the interface embeds the media automatically), a link from centralized sites like imgur and stuff, who know your IP address, take down the media immediately (the embed 404’s) and report you to authorities. Further, plebbit works like torrents so your IP is already in the swarm, so you really shouldn’t use it for anything illegal or you’ll get caught.

nobody is running the matrix server at the moment, if you are interested in running it dm @estebanabaroa on telegram

The communities moderate themselves with their own admins, just like on reddit. The difference is, there’s no global admins that can censor communities or enforce global rules. However, the plebbit app developer can basically act like a global admin by blacklisting connections to certain communities. I predict the most popular plebbit apps won’t include such blacklisting functions.

Plebbit is like BitTorrent, there’s no global BitTorrent admin. You use a BitTorrent client (like uTorrent) to download torrents, and the client could technically blacklist your torrent. You use a plebbit client (like Seedit) to download a subplebbit, and the client could technically blacklist your subplebbit.

It’s entirely possible that more centralized plebbit clients will be created, to be published on app stores for example, and they will implement whitelists of safe communities to participate in, blocking any other community.

You need to have a structure for delegating moderation and such tasks to other people.

We actually have it: since there’s no central database of communities, who decides which ones appear in the homepage of the apps to first-time users? We use a “default list” of communities, which is effectively moderated (vetoed) by the app developer. This is the only “global admin” we basically have, but it’s only for the app itself, not the protocol, and it still doesn’t stop users from connecting p2p to the community (depending on the app, some plebbit client developers could implement blacklists).

I agree in general, just like the word “decentralized”. But in this case it’s legit, because it simply means it’s p2p. I’d call bitcoin “serverless” as well, so it’s BitTorrent and IPFS. Plebbit is exactly the same: you open the desktop app and it runs a p2p node automatically in the background, to run your subplebbit, and users connect to it peer to peer. Your p2p node is not really a “server”, because it doesn’t require any centralized domain to function, it uses transport protocols and peer discovery instead.