I’m new to the fediverse, and so far, I have been seeing some situations from lemmy instances rejecting users to defederating from others, but I ask myself: what can be done if trolls or bots come from self-hosted single-user instances?
They get defederated wherever they get consistently banned…and then they can federate with their friends and make their community of shit and yell at each other or something.
Can it be practical if several instances get created programmatically?
Well, if we had a DDOS-like barrage of troll instances, probably a solution would be to institute a “pending federation request” for any new communities.
A reputation/trustworthiness rating for instances might be helpful too, something akin to karma but for the instance as a whole. More vulnerable communities would be able to set a minimum trust requirement for unapproved participation.
Matrix was working on something similar a whole back, although it may be easier to do on Lemmy
I like that idea. It would be good to put numbers on positive community traits like trust, helpfulness, politeness, inclusiveness, innovation, etc. so that amoral analysts have something more to go on than Nusers, connections, clicks and replies when deciding what/how to monetize things. A bit like biodiversity protection…or a freedom index or HDI.
Looks interestsing.
Would it be also possible to ask for a captcha on the first interactions users from new instances make?
No, because they would have to solve captchas on every instance they’re federated with. You’d be solving like 29 captchas per post
This should be the default in the next Lemmy version because otherwise it’s only going to get easier for people to launch new instances and it’s going to get too overwhelming
I believe there is a whitelist mode, that isn’t currently enabled but in that mode new instances are defederated by default. That would be a pain to administer for all the small instances joining, so they may set some size or reputation rules to apply for federation (??). I imagine this is inevitable if automated instance spamming becomes a problem, unless some type of RBL comes in vogue for managing federation first. It may become impossible to manage manually.
E: oh the whitelist thing may only be relevant to kbin federation, as that was the topic I saw it in. Sometimes I forget where I am :-/
E: oh the white list thing may only be relevant to kbin federation, as that was the topic I saw it in. Sometimes I forget where I am :-/
No, you were correct - Lemmy’s admin settings has an
Allowed Instancessetting, which acts as an allowlist for federation. If you put an instance (or more, separated by,'s) then your instance will only federate with the ones specified in that list.
I don’t usually make a habit of being an ass, but I am extra careful about what I say because I choose to self host. Since any instance admin that decides not to take a liking to me could take one look at my instance and defederate me without a second thought. If anything, self-hosting makes you more vulnerable than just normally ban evading on an open instance.
I agree with Jamie. While it is certainly possible for a bad actor to spin up burner instances for the purposes of evading defederation, that’s a disproportionate amount of effort compared to just creating a new account somewhere that already exists.
Will we see it happen? Probably. But it honestly seems easier to deal with than if those bad actors were to hide themselves in established instances.
deleted by creator
Assuming you can block a top level domain and have it apply to all subdomains, doing so would force the troll to pay for another domain to get around the block. That could add up to quite the expense.
I could see that happening, exploitation by self-hosted bad guys. Is there any type of moderation within the federation framework to avoid that kind of thing?
Right now I think the federation is small and new enough to avoid that kind of exposure, but I’d be surprised it did not become a threat as the federation network becomes larger and more well known.
I’m not running an instance yet. If I do, I’d expect to drop any instance that seems to emit mostly junk, even if they sometimes emit non-junk.
It would be good to have tools that (for example) give instance admins an overview of what traffic they’re receiving from various other instances.
Don’t know if lemmy.buttgoats.com is good? You should be able to get a summary of all the traffic they have sent to others, and how much of it was removed by moderators.
Instance admins can blacklist & whitelist other instances from federating with theirs. If you have a lot of bot spam originating from a particular server, you can just add them to the blacklist to prevent them from interacting with yours.
