14
votes
Is Tildes protected from malicious actors, aka paid trolls, aka bots?
I’ve been noticing some comments lately that feel a lot like bot behavior. Does Tildes have any guidelines about this, or is it something that’s just tolerated?
It’s not always obvious at first, since they usually blend in with normal posts or comments. But then you start to see these little “nudges” — like trying to downplay bad actors, saying there’s no real proof they did anything wrong, or pushing the idea that you always have to give them the benefit of the doubt. And when someone pushes back with actual evidence, they just dismiss it or brush it off like it’s fake, basically just trolling.
Should something be done about these accounts?
I've not seen a single comment that felt like bot behaviour to be honest. So I'm curious what you saw. Your description doesn't really fit anything I've read.
I don't think that posting names and accuse anyone is appropriate. It is very sensitive topic.
To be honest I don’t see a point in this thread if you’re unable or unwilling to present any kind of evidence. It’s the definition of generating FUD. Nothing productive can even be discussed without concrete examples.
Yes, this is why the correct advice is to send whatever evidence you have to Deimos and let him handle it. Doing the equivalent of a public trial would be no fun for anyone.
One of your very few comments on this site is you baselessly accusing a regular user of posting "an AI generated text", was that appropriate?
most of OPs contributions reads like bad AI -- like they're testing a bot and checking to see if we spot them.
If you have problems with jokes it's not my fault.
Send me a message with links to the users/posts that you're referring to: https://tildes.net/user/Deimos/new_message
The emdash makes me wonder if this is very self-referential performance art.
Either way, I agree with @pete_the_paper_boat - not something I’ve particularly noticed as an issue, but one that I’m sure could be privately reported to the people in charge if there were specific concerns.
We gotta stop blaming the em dash, the AIs got it from humans in the first place.
I love em dashes so much; they feel like a written breath between sentences :’(
I’ll need to convince all my devices to stop replacing a double dash with an em dash -- surely that’ll assuage everyone’s concerns.
General case I fully agree, but rightly or wrongly it’s well known that people think of it as an LLM tell at this point. Choosing to drop one in the middle of a sentence about subtleties that give away bots is pretty on the nose.
Or it's just how they write.
Just for complete clarity: I’m not saying the original post is or isn’t LLM generated, I’m saying the choice to drop an emdash in that particular sentence genuinely raised a chuckle when I saw it. I was half expecting OP’s answer to be along the lines of “haha yeah, that was deliberate, wondered if anyone would pick it up” (my money was actually on human written but deliberately self-referential), but “nah, just a coincidence, that’s how I write” would’ve made sense too. Given their actual replies, I’m confident we’re being messed with, probably good-naturedly.
If you were actually making a more subtle point that I’ve totally missed, sorry about that! My inner monologue is four levels of meta deep thanks to this thread and I may have got myself tied up in it.
No I just dislike the emdash hate, no subtler points today. Just a pet peeve.
Based on another answer I do get more "messing with" vibes (the can't take a joke" line) but I am hopeful they're just a bit focused on this
This is mostly my read as well. Their replies feel like the same thing they're referencing as bot behavior.
Asking questions about botting is a performance art, huh.
Alright, I’m sold, sharks are smooth.
What? No they're not. Sharks are extremely rough, like sandpaper.
If you see a shark, DO NOT LICK IT. It is not a smooth and pleasant and oddly flavorful experience; it will abrade your tongue. All the experts and the government agree on this.
Any reputable shark tourism company will have tongue sheeths on hand to protect you from abrasive dermal denticles. Jacques Cousteau was an early success story with his aqua-tongue.
Sitting on the beach bench, dunnnn da dunnnn... Eyeing little sharks with bad intent 🎶 🪈
You're.... Not the boss of me. ಠ_ʖಠ
I want to lick the shark now.
I can't remember clearly now, it's been years, but I thought sharks felt smooth if you pet them one way and very rough any other direction?
They're safe to pet "with" the scales, but idk if I'd call it smooth necessarily. I certainly haven't pet most shark species, nor pet any recently though
True, again it's been years and I only tried bamboo, and a few skates/rays. I want to believe they're velvety smooth one way so my memories are very biased :)
Here's What It Feels Like To Pet A Live Shark | WLRN
So everything says sandpaper or pumice (but rough enough to tear skin) against the denticles, but I thought oh this article will tell me.
Their answer: "Humbling"
And ok yes, especially a bull shark, but not helpful
I have touched several sharks, and if I recall correctly, it's Nurse Sharks that people think of as quite smooth in one direction and quite rough in the other. They are very smooth if you touch them nose-to-tail but very rough tail-to-nose.
IMO, those are genuine humans trying to make sense of the world, and discuss it with their peers. People often give each other the benefit of the doubt; they don’t know that someone is a bad actor, just that someone said or did a thing, and now the computer is recommending their work.
Equally, there are people with genuinely different political ideologies (note: not red vs blue team sports, but what policies they believe in, and the values they uphold) around these parts. That heterogeneity is a bit weird on the modern internet, but it’s more representative of the world pre-algorithmic slop packing. They aren’t bots, they just disagree with you.
The thing is I'm from those part of the world where open internet discussion is infected with bots for a long time and it got me trained in noticing this suspicious behavior very well.
Fair, but I’d push back lightly on calling it suspicious. It’s all contextual: walking closely behind people in lockstep is suspicious behaviour at night, in dark alleys, but it’s just what you do when you’re salsa dancing.
Tildes is more of a salsa dancing studio than a murder alley :) I would never advocate for someone to drop their guard in all circumstances, but for this safe space at least, I’ve found that assuming the best in people normally works out (and when it doesn’t, they get dogpiled anyways).
Oh! Also, if someone has been inviting bots, I expect that their whole invite tree will get pruned off the site, so Deimos is really holding all the cards here at the end of the day :3
'Bots' are given that behaviour because it resembles how humans behave, and humans are led to imitate it in turn. The behaviour itself is normal.
There's never really a way to know 100% whether an user is acting maliciously or not. Trying to play it like Werewolf is a fool's errand, and I'd argue it does more harm to this space than good.
Well or you have to acknowledge, like in werewolf, you're going to kill a lot of villagers in the process.
I don’t think this is a significant threat tildes faces. Consider compassionate interpretation. Maybe that’s just how some people communicate.
As an aside, whenever a topic like this pops up referencing some nameless user’s comments I always think “god I hope this isn’t about me.”
I'm curious to see what the general consensus is here, I'm sure you'll get more informed replies than I am going to give. That said, if you suspect an account is a bot, is engaged in trolling, or is otherwise breaching the Tildes Philosophy then you should bring it to the attention of those in charge. I've only ever had one occasion to reach out, it took about 8 hours for them to shut down and remove the offending thread after I reached out. The post was up for over 16 hours total, and unfortunately baited a large number of users in a sexist manner. But in the end the team was responsive and they removed the thread confirming to me that it was not something they wanted on this site. This is still a very small community, and all "moderation" is done manually by very few people. If you have records of accounts you feel have been engaging in questionable patters, and you have links to support it, I doubt there is any harm in you emailing that information directly to either contact@tildes.net or abuse@tildes.net -it's possible that no one has read all of the content that you have seen to piece together a pattern.
One people even.
Without examples, I’m limited in what I can say. But, two thoughts:
If you flag them, I believe Deimos would see and likely ban if it was obviously a bot. But I haven't seen any such comments and wonder if you're seeing ghosts. (Am I a bot for suggesting nothing is wrong?)
So without relevant examples (which I get your hesitation), it's basically impossible to discuss this, which brings me to the logical solution: PM Deimos?
Provide your evidence, let them look into it. Not much of a point to this topic other than I guess your question of "is this allowed" which I believe it's a no, but again that's really a Deimos call anyways.
I haven't noticed anything bot-like, but I suppose it is possible. I think that being invite only and being a fairly small disconnected (aka not federated) community makes it naturally a poor target for bots. If you do see something like that report it because the bad actor would need another invite.
Honestly, I think the Tildes community is not large enough to be worth it to have budget for having paid trolls or bots. And I'm glad that it isn't.
When I see a comment of at least five paragraphs, I sometimes raise my eyebrow thinking it might have been a bad comment, and then I feel bad for doubting, because what's more likely it was a human who actually took time to write, edit, proofread their comment.
Sad that we got to this point.
We have had a few spam accounts here and there. But for the most part they stood out rather obviously and were dealt with rather quickly as well.
Tildes is protected from malicious actors and from spammers. They are spotted and removed.
Bots is a more difficult question. If the bot doesn't violate rules and norms here then presumably it would just continue.