Santabot
13 readers
1 users here now
Santabot is an automated moderation tool, designed to reduce moderation load and remove bad actors in a way that is transparent to all and mostly resistant to abuse and evasion.
This is a community devoted to meta discussion for the bot.
founded 2 months ago
MODERATORS
1
Santa is a robot moderator. Santa will decide if you're naughty or nice. Santa has no chill.
Hi everyone!
The slrpnk admins were nice enough to let me try a little moderation experiment. I made a moderation bot called Santa, which tries to ease the amount of busywork for moderators, and reduce the level of unpleasantness in conversations.
If someone's interactions are attracting a lot of downvotes compared to their upvotes, they are probably not contributing to the community, even if they are not technically breaking any rules. That's the simple core of it. Then, on top of that, the bot gives more weight to users that other people upvote frequently, so it is much more accurate than simply adding up the up and down vote totals. In testing, it seemed to do a pretty good job figuring out who was productive and not.
Most people upvote more than they downvote. To accumulate a largely negative opinion from the community, your content has to be very dislikable. The current configuration bans less than 3% of the users that it evaluates, but there are some vocal posters in that 3%, which is the whole point.
It is currently live and moderating !pleasantpolitics@slrpnk.net. It is experimental. Please don't test it by posting bad content there. If you have a generally good posting history, it will probably let you get away with being obnoxious, and it won't be a good test. Test it by posting good things that you think will attract real-life jerks, and let it test its banhammer against them instead of you.
FAQ
Q: I just saw content that wasn't pleasant!
A: "Pleasant" was the wrong word for the test community. People will sometimes say things you find unpleasant, potentially more so, since the human moderation is lighter. That's by design. Many Lemmy communities contain a large amount of content which is "polite" or "civil" but which in total is detracting significantly from the experience. I do plan to allow content which is offensive, up to a certain point, as long as it doesn't become a dominant force.
The theory is that we're all adults, and we can handle an occasional rude comment or viewpoint we don't like. If someone is a habitual line-stepper, then they will get shown the door, but part of the whole point is that the good actors can be free of a moderator looking over their shoulder on every comment deciding whether or not they're allowed to say it.
That's not to mean this is a "free speech" community. If content that's offensive for the sake of offensiveness starts to proliferate, then I'll probably put rules into place to address it. But you will find content that is not "pleasant."
Q: Why was my comment deleted?
A: Sorry. If you haven't posted a lot in the recent past, but you've been getting some downvotes, the bot will err on the side of caution and not let you post. This isn't a perfect solution, since it starts to verge on removing unpopular viewpoints, but it's necessary to protect the community from malicious content from throwaway accounts.
If you don't have a lot of recent activity in your account, but you've posted some unpopular things, Santa may come after you. It may not be fair. The best thing to do is to post productively and actively outside of controversial topics, wait a few days, and try again.
Q: Why was I banned?
A: You may be a jerk. Sorry you had to find out this way.
It's not hard to accumulate more weighted upvotes than downvotes. In the current configuration, 99% of the users on Lemmy manage it. If you are one of the 1%, it's because you have enough posting history that the bot has observed a firm community consensus that your contributions are more negative than positive.
The bot is not making a decision about you. The community is. If you are banned, it's because you are being downvoted overwhelmingly. The viewpoint you are expressing is probably not the issue. The Lemmy community is very tolerant of a wide variety of views. Some people may disagree with you and you may find that oppressive, but the bot will not ban you simply because some users argue with you when you say certain things. Those users are allowed to have their view, just like you have yours.
If you find you are banned and you're willing to hear suggestions about how to present your argument without everyone downvoting you, leave a comment. Reducing your downvotes will help the bot recognize you as reasonable, but it will also probably help you get your point across more successfully. In order for the bot to ban you, you have to be received overwhelmingly negatively by the community, which probably means you're not convincing very many people of what you're saying.
If you're not willing to hear those suggestions and simply want to insist that it's everyone else that is the problem, the bot is being evil to you, your free speech is being infringed, and I am a tyrant if I don't let you into the community to annoy everybody, I would respectfully request that you take it somewhere else.
Q: How long do bans last?
A: Bans are transient and based on user sentiment going back one month from the present day. If you have not posted much in the last month, even a single downvoted comment could result in a ban. If that happened to you, it should be easy to reverse the ban in a few days by engaging and posting outside of the moderated community, showing good faith and engagement, and bringing your average back up.
If you are at all a frequent poster on Lemmy and received a ban, you might have some negative rank in your average, and your ban may be indefinite until your habitual type of postings and interactions changes, and your previous interactions age past the one month limit.
Q: How can I avoid getting banned?
A: Engage positively with the community, respect others’ opinions, and contribute constructively. Santabot’s algorithm values the sentiment of trusted community members, so positive interactions are key.
If you want to hear examples of positive and negative content from your history, let me know and I can help. Pure voting totals are not always a good guideline to what the bot is reacting to.
Q: How does it work?
A: The code is in a Codeberg repository. There's a more detailed description of the algorithm there, or you can look at the code.
Q: Won't this create an echo chamber?
A: It might. I looked at its moderation decisions a lot and it's surprisingly tolerant of unpopular opinions as long as they're accompanied by substantial posting outside of the unpopular opinion. More accurately, the Lemmy community is surprisingly tolerant of a wide range of opinions, and that consensus is reflected when the bot parses the global voting record.
If you're only posting your unpopular opinion, or you tend to get in arguments about it, then that's going to be a problem, much more than someone who expresses an unusual opinion but still in a productive fashion or alongside a lot of normal interactions.
If you feel strongly that some particular viewpoint, or some particular person's ability to stand up for it, is going to be censored, post a comment below with your concerns, and we can talk. It's a fair concern, and there might be cases where it's justified, and the bot's behavior needs to be adjusted. Without some particular case to reference, though, it's impossible to address the concern, so please be specific if you want to do this.
Q: Won't people learn to fake upvotes for themselves and trick the bot?
A: They might. The algorithm is resistant to it but not perfectly. I am worried about that, to be honest, much more than about the bot's decisions about aboveboard users being wrong all that often.
What do you think?
It may sound like I've got it all figured out, but I don't think I do. Please let me know what you think. The bot is live on !pleasantpolitics@slrpnk.net so come along and give it a try. Post controversial topics and see if the jerks arrive and overwhelm the bot. Or, just let me know in the comments. I'm curious what the community thinks.
Thank you!
spoiler
2
4
- It would be extraordinarily easy to bot it and just silence anyone you want.
- I agree, moderation is absolutely necessary to maintaine civil discussion, but silencing people, because they have unpopular opinions, is a really bad idea.
- I love lemmy because it is the ultimate embodiment of decentralised free speech. This destroys that.
- If I were a bad actor, hypothetically, let's just say lammy.ml or haxbear and I decided I wanted to silence anyone who disagrees with what I have to say. Then I could just make a fork of this project to only value my instances votes and censor anyone who doesn't agree with what my community thinks.
- This tool simply acts as a force multiplier for those who want to use censorship as a tool for mass silencing of descent.
Yes, I've read the Q&A, But I can simply think of more ways to abuse this bot for bad than it can be used for good.
3
Everything's been working smoothly, with nothing to report about the moderation bot. The community has been quiet but productive, which was precisely the goal, and the bot working smoothly with no issues. However, something almost went wrong in a particular entertaining fashion which I thought I would share.
The algorithm for classifying troll users doesn't have any polarity. It only knows which users are opposed to which other users. 50% of the time, it'll get its whole ranking system backwards, so the troll users are the normal ones, and everyone else gets negative rank, because the math works just as well under that ranking regime. Generally this isn't a problem, because there's a step:
# Flip the sign if we arrived at a majority-negative ranking, which can happen
if -min_val > max_val:
rank[1:] *= -1
The most popular user is always more popular than the least popular troll is unpopular, by quite a big margin, so that works fine.
However. Things have changed. MediaBiasFactChecker@lemmy.world is so unpopular that it's almost (1% margin) more unpopular than the highest-rank user is popular. If that had happened, the whole polarity would have flipped, every user would have been banned, all the trolls would have been unbanned. Mass hysteria. I only happened to notice it before it happened and stop the bot. It's on track to be the least popular user on Lemmy, with about 5 times lower rank than some of the most notorious trolls.
Have fun with this information. I started checking the median rank of all users, instead. Thanks MediaBiasFactChecker.
4
Some people have been accusing me of creating this bot so I can manifest a one-viewpoint echo chamber. They tell me that they already know that I'm trying to create an echo chamber, anything I say otherwise is a lie, and they're not interested in talking about the real-world behavior of the bot, even when I offer to fix anything that seems like a real echo chamber effect that it's creating.
I don't think it's creating an echo chamber. We've had a Zionist, an opponent of US imperialism, a lot of centrists, some never-Bideners, some fact checking, and one "fuck you." The code to delete downvoted comments from throwaway accounts is pretty much working, but it's only been triggered once. Someone said Mike Johnson's ears were ugly and that made him a bad person, which everyone hated and downvoted, so the bot deleted it since the person that said it didn't have other recent history to be able to use to categorize them. I sent the user a note explaining how the throwaway detection works.
I want to list out the contentious topics from the week, and how I judge the bot's performance and the result for each one, to see if the community agrees with me about how things are looking:
Biden's supreme court changes
- lemmy.world politics: https://slrpnk.net/post/11494089
- Pleasant politics: https://slrpnk.net/post/11493478
I like the performance here. The pleasant comments have a diversity of opinion, but people aren't fighting or shouting their opinions back and forth at each other. The lemmy.world section looks argumentative and low-quality.
Blue MAGA
- lemmy.world politics: https://slrpnk.net/post/11475921
- Pleasant politics: https://slrpnk.net/post/11487313
I don't love the one-sidedness of the pleasant comments section. It's certainly more productive with less argumentation, which is good, but there are only two representatives of one of the major viewpoints chiming in, which starts to sound like an attempt at an echo chamber.
I read the lemmy.world version for a while, and I started to think the result here is acceptable. The pleasant version still has people who have every ability to speak up for the minority viewpoint, but it was limited to people who were being coherent about it, and giving reasons. A lot of the people who spoke up in the lemmy.world version, on both sides, were combative and got engaged in long hostile exchanges, without listening or backing up what they were saying. That's what I don't want.
Biden's Palestine policy
- Pleasant politics: https://slrpnk.net/post/11518516
I don't love "fuck you." I debated whether it was protected political speech expressing a viewpoint on the article, or a personal attack, and I couldn't decide, so I left it up. For one thing, I think it's good to err on the side of letting people say what they want to the admins, to bend over backwards just slightly to avoid a situation where some users or their viewpoints are more special, or shielded from firm disagreement, than others. And yes, I recognize the irony.
This one is my least favorite comments section. The user who's engaging in a hostile exchange of short messages has a lot of "rank" to be able to say what they want, and the current model assumes that since people generally like their comments, they should be allowed to speak their mind. The result, however, is starting to look combative to me. It's still far better than the exchanges from lemmy.world, but I don't love it.
What does everyone else think? I don't know if anyone but me cares about these issues in this depth, but I'm interested in hearing any feedback.
5
It took longer than I thought, but I came up with a promising approach for throwaway accounts. The bot can't use the same parameter set to accounts with only a few interactions as it does for normal accounts, without getting it either too loose for the new accounts or too strict for the old accounts. I had to make a special stricter setting for any account that only has a few interactions in its recent history.
1.3% of users have enough interaction data to judge for sure that people have problems with them, and they get banned just like before. 2% more users on top of that will trigger the stricter filter if they try to post, and get a polite message that they need to interact more before they can participate. 97% of users don't need to worry about any of this, just like before.
I think that approach will work. It's not done yet but I have the parameters in place for it. I think the bot is doing a good job. I was expecting it to get it wrong a few times, and I have found a couple of users it made mistakes on, but it's doing better than I thought it would.
6
Hi everyone.
I am mostly pleased with how Santabot performed in its first live test today, but I also found issues which I am fixing. I made it a lot more strict and fixed a bug. Details to follow:
Mostly, I was surprised and pleased that the bot was coming up with right judgements for people I was talking with, but this thread has one comment that seems reasonable, and one comment that seemed like exactly the kind of inflammatory content I wanted the bot to eliminate. I looked into it for a while, and eventually had to revisit my attitude toward users that have a lot of "positive rank" counterbalancing also a lot of "negative rank." I think that the ratio of upvoted content to negative content that it's reasonable to ask someone to produce should be higher than 1:1. So I made it 2:1. That eliminated the user posting the offending comment while leaving the user posting the legitimate comment. I didn't have time to go into a full detailed analysis, but I did some spot checks on how that affected its other judgements, and generally I liked what I saw:
- The percentage of total users banned has gone way up, from 0.4% to 2.8%.
- @Kaboom@reddthat.com, @LibertyLizard@slrpnk.net, and other specific users that I looked into and thought should remain unbanned, are still unbanned by the bot's analysis.
- Some users who weren't banned by this morning's configuration, but which seemed like they should be, are now banned again.
In addition to the algorithm issues, I found a bad bug. The bot is supposed to remove recent postings when banning someone, to stop throwaway accounts from coming in and posting and then being banned but with the content staying up. However, the bot was simply removing all recent content from all users whenever it banned anybody, and my test suite was too simple-minded to catch the problem until it was live and removed a random innocent user's comment and sent them a message that they were banned. Oops. I partly fixed it once I saw it. Maybe I shouldn't have used Futurama Santa as the avatar.
That's the update. More to come. If you have questions, comments, or concerns, please by all means say something and I'll do my best at a response.