|
Top-level
 As others have pointed out, more sophisticated misinfo bot operations will be tried with the Fediverse. If they can, they will try to distribute the bots across multiple servers in good standing. I’ve already seen some of this happen with Russian propaganda bots. Servers admins have thus far been quick to respond. 15 comments
@mohaneds Most admins don’t intend on scaling to 100,000 users per instance. That gets expensive. It’s much easier to moderate for small servers.  @atomicpoet I think that the vector I'm most concerned about is single-user instances - feels like if someone figures out a way to set up a one-bot-per-instance production line that scales, that could be a problem @RufusJCooter That’s expensive. If we assume $5/server, that comes out to $750,000 for a single server bot farm of 150,000 accounts.  @atomicpoet Do mastodon blocks work by IP address or domain name? Seems you don’t really need lots of servers, just lots of domain names pointing to a few servers that route traffic appropriately. @atomicpoet I don't disagree - it's probably not a practical or cost-effective strategy /now/. My hunch, tho', is that it will get easier & cheaper to spin up servers as the 'verse matures. A possible side-effect of that, I worry, would be making it easier for trolls to troll.  @atomicpoet @RufusJCooter But do you actually need to spin up an instance? Or do you just need to write a program that acts like a server enough to federate? And if my "fake" server software is only posting, not reading in any messages, how much server HP would it actually need? Seems lightweight to me. My laptop could probably run 100+ "instances/bots" @atomicpoet @RufusJCooter and that is ignoring the fact that they would also have to pay for different domains name if they wanted the bots to truly be effective and not just blocked by blocking one domain name. @atomicpoet The accounts that made me realize I should turn on account approval for my server were a pair with big Russian bot vibes -- two created right after each other with random names and IP addresses that were close together, both traceable to... I think it was Belgium? @atomicpoet I haven't given this much thought, so maybe it's a dumb question, but doesn't the lack of an algorithm also significantly reduce the reach? 🤔 Even if the bots show up on reputable instances, they need boost-happy followers to get broader exposure, right?  @atomicpoet I've had bots from different instances with East Asian casino spam. I come across them when moderating new hash tag trends. 10 of them are easy to report to the source servers. If that should scale, it will be a lot of work. Maybe a hashtag reporting option would help here?  please: can you help this fellow ? @atomicpoet I’m not too worried about this, but if they were smart (and I’m hoping they’re not) they’d be using this time in isolation to hone their bots effectiveness. Then they just need to wait for a gap that allows them to either create mass numbers of accounts on multiple servers, awaken the thousands of sleepers they already have, or subvert existing accounts. |
Gregory's Server
@atomicpoet Out of curiosity, how do you think Mastodon admins will handle hundreds of thousands of them? I can see a future where they start using mastodon.social, mastodon.online, mstdn.party ETC. Most server admins don't have the resources to deal with a state or corporate-sponsored disinfo campaign.