Former New York Gov. Andrew Cuomo said in a video released Monday that he will run an independent campaign for New York City mayor, weeks after losing the Democratic primary to Zohran Mamdani.
so can u tell me more about this project wen u have time? I previously commented to you that I think it’s sick. but are you using it to attack users or to find out bad mod actions and proof of mods and admins and people like @PhilipTheBucket@quokk.au who obsess and witchhunt people they disagree with?
so can u tell me more about this project wen u have time?
Certainly! I can also give you code you can work and play around with yourself. I’m more than happy to send you some boiler plate you can play around with yourself.
but are you using it to attack users or to find out bad mod actions and proof of mods and admins and people like @PhilipTheBucket@quokk.au who obsess and witchhunt people they disagree with?
Well, not really any of the above. I’ve tried with some mild success to build a “troll detection” system, but it needs far more work. Also, in the months since my initial work on this matter, I’ve found some far better approaches and would want to implement them. So my old work isn’t reflective of the new direction I’m planning to take.
Fundamentally, I’m interested in these things from an academic perspective. How do conversations (debate) online work? What governs them? There are obviously rules (you can read them on the side bar), but there are also “rules” that aren’t written on the side bar. What are the unwritten rules?
What does it take to “change” someones mind? Or, more broadly, what does it take to change a communities mind? How do power dynamics play into that? For example, you’ve probably read my thread with @PhilipTheBucket@quokk.au at this point about the power mods have to steer the direction of a community.
Part of the reason I’m interested in this is because its part of my lived experience. I watch how myself and others were, frankly, absolutely brigaided against, for months, years on end, for holding critical but important positions on the Democratic party. It was very, very bad. We had opinions that were not very popular at the time and suffered as a result. But things changed, and time has shown we were right in our criticism. So what dynamics were at play that resulted in those perspectives being at first oppressed/ suppressed, to then become the dominant narratives? How does that work? What is its function?
Beyond that, I’m a quantitative person. I want a number at the end of the day; as shocking as it might be with all my discussion of narrative, at the end of the day I want a number, something solid I can stand on. So doing this kind of work in my own way, I want to find a way to quantify these things. Its not enough for me to simply look at an encode the stories; I want to put a number on it. That means building things up to be reproducible and automated to support large, if not census, level samples.
Finally, I really like doing network analysis. Its something I do professionally, but its just something I think is neat. Taking all of those previous questions and putting them into the context of a social network, thats something that sounds really challenging and fun to me.
And in regard to your other question:
this sounds gross and very much like witch-hunting and stalking. yes, all comments are public but u coming up w a tool just to find someones and analyze and make judgement on them seems gross and out of line. think of how fasicst and controling that sounds. what if some republican was doing that and bragged about it? think about that
I mean, you do realize that anyone, quite literally anyone, could form a 1 person instance and vacuum up all comment, post, etc… data from every other instance? I don’t have issue with it, in the same manner that I don’t have issue someone going through all 9k of my comments and reading them. If I didn’t want them to be made public, I wouldn’t make those comments. There are things I don’t say because I don’t want that information to be made public. Fundamentally these things are about power dynamics. For something to be fascistic or controlling I would have to have power over someone or something. I don’t. I have no secret access to any secret information, I have no power over any one or any thing. I’m simply working with and observing what is present.
This is a somewhat famous thread here that I recommend you read all of the comments of. Its the one @PhillipTheBucket@quokk.au and I are discussing. It also highlights the dynamic I’m interested in illustrating. Here is the link: https://lemmy.world/post/16224102?sort=Top
I recommend sorting by “Top” and reading through the first couple comment threads from top to bottom. Then scroll to the very bottom and read the comment threads in reverse order, basically most down voted. This should give you an idea for the type of dynamic I’m identifying, and the research I’m interested in conducting is how this dynamic shifted within our community. These days you would see an inversion of which narratives are being upvoted and which ones are being down-voted. So how did that come to be?
u are right that mods banend pepole for talking against the dem talk here. lemmy def tries to stifle voices. it’s better now, but befor eelection, it was so blatent. Anyone who supported thrid party were attacked. but yoru tracking program still sounds icky.
Well, not really any of the above. I’ve tried with some mild success to build a “troll detection” system, but it needs far more work. Also, in the months since my initial work on this matter, I’ve found some far better approaches and would want to implement them. So my old work isn’t reflective of the new direction I’m planning to take.
I’ve actually done a version of this and a couple of other various ideas about it. The current WIP idea works totally differently to what you are talking about, I actually got as far as making a community for it, but then abandoned the effort because I couldn’t figure out a way to deploy it that would be in any way productive.
I’m going to say it knowing ahead of time that roughly 100% of the people reading are going to think it’s a terrible idea: It is an LLM-based moderator that watches the conversation and can pick out bad faith types on conduct in the conversation. I actually 100% agree with you about political conversation online being almost exclusively a big waste of time (including because of the way moderation happens and people trying to deliberately distort the narrative). This was just my idea to try to help it.
The thing that led me to never do anything with it was that I didn’t feel like anyone would ever buy into it enough to even take part in a conversation where it was deployed (even assuming it worked passably well which is not proven). If you care about these issues also, though, would you like to try the experiment of having the whole conversation we’re having with it observing the conversation and weighing in? I would actually like to, I’d be fine with continuing with the questions you were asking and continuing this whole debate about moderation and its impact on Lemmy, in that context. Let me know.
The thing that led me to never do anything with it was that I didn’t feel like anyone would ever buy into it enough to even take part in a conversation where it was deployed
Yeah I think its got to work for people to buy into it. And frankly my earliest implementations were “inconsistent” at best.
My thought right now is that the tool needs to do a first pass to encode the “meta-structure”, or perhaps… scaffolding(?) of a conversation… then proceed to encode the impressions/ leanings. I have tools that can do this in-part, but it needs to be… “bigger”… whatever that means. So there is sentiment analysis, easy enough. There is key phrase extraction. And thats fine for a single comment… but how do we encode the dynamic of a conversation? Well thats quite a bit more tricky.
still seems to me u guys are doing it for witchhunting. if someone doesn’t like someone they can just ban them. you two going on and on about writing a program and using ai to catch peopel you don’t like is icky. I’ll be one of the people voting against this if it ever goes wide on lemmy. no thanks. u all need to touch grass, ur way too caught up in lemmy
At least for the tool I was talking about, it wasn’t planning on banning anyone. I’ve been a moderator for a decently large collection of forums on Lemmy and I can’t even remember the last time I banned someone, although it did happen a handful of times months and months ago. The tool was planned as purely something to give input to the participants about elements of the other person’s point that they were getting carried away with their own stuff and not addressing.
Yeah, generally having it read the conversation (I think as JSON, maybe in markdown for the first pass, I can’t remember, it’s a little tricky to get the comments into a format where it’ll reliably grasp the structure and who said what, but it’s doable) and then do its output as JSON, and then have those JSON pieces given as input to further stages, seems like it works pretty well. It falls apart if you try to do too much at once. If I remember right, the passes I wound up doing were:
What are the core parts of each person’s argument?
How directly is the other person responding to each core part in turn?
Assign scores to each core part, based on how directly each user responded to it. If you responded to it, then you’re good, if you ignored it or just said your own thing, not-so-good, if you pretended it said something totally different so you could make a little tirade, then very bad.
And I think that was pretty much it. It can’t do all of that at once reliably, but it can do each piece pretty well and then pass the answers on to the next stage. Just what I’ve observed of political arguments on Lemmy, I think that would eliminate well over 50% of the bullshit though. There’s way too many people who are more excited about debunking some kind of strawman-concept they’ve got in their head, than they are with even understanding what the other person’s even saying. I feel like something like that would do a lot to counteract it.
The fly in the ointment is that people would have to consent to having their conversation judged by it, and I feel like there is probably quite a lot of overlap between the people who need it in order to have a productive interaction, and those who would never in a million years agree to have something like that involved in their interactions…
so can u tell me more about this project wen u have time? I previously commented to you that I think it’s sick. but are you using it to attack users or to find out bad mod actions and proof of mods and admins and people like @PhilipTheBucket@quokk.au who obsess and witchhunt people they disagree with?
Certainly! I can also give you code you can work and play around with yourself. I’m more than happy to send you some boiler plate you can play around with yourself.
Well, not really any of the above. I’ve tried with some mild success to build a “troll detection” system, but it needs far more work. Also, in the months since my initial work on this matter, I’ve found some far better approaches and would want to implement them. So my old work isn’t reflective of the new direction I’m planning to take.
Fundamentally, I’m interested in these things from an academic perspective. How do conversations (debate) online work? What governs them? There are obviously rules (you can read them on the side bar), but there are also “rules” that aren’t written on the side bar. What are the unwritten rules?
What does it take to “change” someones mind? Or, more broadly, what does it take to change a communities mind? How do power dynamics play into that? For example, you’ve probably read my thread with @PhilipTheBucket@quokk.au at this point about the power mods have to steer the direction of a community.
Part of the reason I’m interested in this is because its part of my lived experience. I watch how myself and others were, frankly, absolutely brigaided against, for months, years on end, for holding critical but important positions on the Democratic party. It was very, very bad. We had opinions that were not very popular at the time and suffered as a result. But things changed, and time has shown we were right in our criticism. So what dynamics were at play that resulted in those perspectives being at first oppressed/ suppressed, to then become the dominant narratives? How does that work? What is its function?
Beyond that, I’m a quantitative person. I want a number at the end of the day; as shocking as it might be with all my discussion of narrative, at the end of the day I want a number, something solid I can stand on. So doing this kind of work in my own way, I want to find a way to quantify these things. Its not enough for me to simply look at an encode the stories; I want to put a number on it. That means building things up to be reproducible and automated to support large, if not census, level samples.
Finally, I really like doing network analysis. Its something I do professionally, but its just something I think is neat. Taking all of those previous questions and putting them into the context of a social network, thats something that sounds really challenging and fun to me.
And in regard to your other question:
I mean, you do realize that anyone, quite literally anyone, could form a 1 person instance and vacuum up all comment, post, etc… data from every other instance? I don’t have issue with it, in the same manner that I don’t have issue someone going through all 9k of my comments and reading them. If I didn’t want them to be made public, I wouldn’t make those comments. There are things I don’t say because I don’t want that information to be made public. Fundamentally these things are about power dynamics. For something to be fascistic or controlling I would have to have power over someone or something. I don’t. I have no secret access to any secret information, I have no power over any one or any thing. I’m simply working with and observing what is present.
This is a somewhat famous thread here that I recommend you read all of the comments of. Its the one @PhillipTheBucket@quokk.au and I are discussing. It also highlights the dynamic I’m interested in illustrating. Here is the link: https://lemmy.world/post/16224102?sort=Top
I recommend sorting by “Top” and reading through the first couple comment threads from top to bottom. Then scroll to the very bottom and read the comment threads in reverse order, basically most down voted. This should give you an idea for the type of dynamic I’m identifying, and the research I’m interested in conducting is how this dynamic shifted within our community. These days you would see an inversion of which narratives are being upvoted and which ones are being down-voted. So how did that come to be?
u are right that mods banend pepole for talking against the dem talk here. lemmy def tries to stifle voices. it’s better now, but befor eelection, it was so blatent. Anyone who supported thrid party were attacked. but yoru tracking program still sounds icky.
I’ve actually done a version of this and a couple of other various ideas about it. The current WIP idea works totally differently to what you are talking about, I actually got as far as making a community for it, but then abandoned the effort because I couldn’t figure out a way to deploy it that would be in any way productive.
I’m going to say it knowing ahead of time that roughly 100% of the people reading are going to think it’s a terrible idea: It is an LLM-based moderator that watches the conversation and can pick out bad faith types on conduct in the conversation. I actually 100% agree with you about political conversation online being almost exclusively a big waste of time (including because of the way moderation happens and people trying to deliberately distort the narrative). This was just my idea to try to help it.
The thing that led me to never do anything with it was that I didn’t feel like anyone would ever buy into it enough to even take part in a conversation where it was deployed (even assuming it worked passably well which is not proven). If you care about these issues also, though, would you like to try the experiment of having the whole conversation we’re having with it observing the conversation and weighing in? I would actually like to, I’d be fine with continuing with the questions you were asking and continuing this whole debate about moderation and its impact on Lemmy, in that context. Let me know.
Yeah I think its got to work for people to buy into it. And frankly my earliest implementations were “inconsistent” at best.
My thought right now is that the tool needs to do a first pass to encode the “meta-structure”, or perhaps… scaffolding(?) of a conversation… then proceed to encode the impressions/ leanings. I have tools that can do this in-part, but it needs to be… “bigger”… whatever that means. So there is sentiment analysis, easy enough. There is key phrase extraction. And thats fine for a single comment… but how do we encode the dynamic of a conversation? Well thats quite a bit more tricky.
still seems to me u guys are doing it for witchhunting. if someone doesn’t like someone they can just ban them. you two going on and on about writing a program and using ai to catch peopel you don’t like is icky. I’ll be one of the people voting against this if it ever goes wide on lemmy. no thanks. u all need to touch grass, ur way too caught up in lemmy
At least for the tool I was talking about, it wasn’t planning on banning anyone. I’ve been a moderator for a decently large collection of forums on Lemmy and I can’t even remember the last time I banned someone, although it did happen a handful of times months and months ago. The tool was planned as purely something to give input to the participants about elements of the other person’s point that they were getting carried away with their own stuff and not addressing.
Yeah, generally having it read the conversation (I think as JSON, maybe in markdown for the first pass, I can’t remember, it’s a little tricky to get the comments into a format where it’ll reliably grasp the structure and who said what, but it’s doable) and then do its output as JSON, and then have those JSON pieces given as input to further stages, seems like it works pretty well. It falls apart if you try to do too much at once. If I remember right, the passes I wound up doing were:
And I think that was pretty much it. It can’t do all of that at once reliably, but it can do each piece pretty well and then pass the answers on to the next stage. Just what I’ve observed of political arguments on Lemmy, I think that would eliminate well over 50% of the bullshit though. There’s way too many people who are more excited about debunking some kind of strawman-concept they’ve got in their head, than they are with even understanding what the other person’s even saying. I feel like something like that would do a lot to counteract it.
The fly in the ointment is that people would have to consent to having their conversation judged by it, and I feel like there is probably quite a lot of overlap between the people who need it in order to have a productive interaction, and those who would never in a million years agree to have something like that involved in their interactions…