Those posters do way more than than 240. I think rimu did some napkin math so it doesn’t affect +99% of users. But its still just a bad solution. Its like saying the way to stop billionaires is to limit how much one person can spend.
In PJs case its just him upvoting every positive comment he receives. He is sort of like fossilique in just posting normal stuff 99% of the time. Its when he actually gives his opinion that he’s a problem.
Oh sure, but what I’m saying is, for those 10 users, they can’t possibly be doing 60%+ of the activity for all posts. That would mean they would need to be the only ones posting and commenting 60% of the time. It tells me either the organic local activities on the instance are DEAD or there is something wrong with the math. On a thread to thread bases of local post activity they can’t possibly make up 60%+ of the activity.
Right? Like, if what he did was take the total upvotes on the instance for a 24 hour period, then uses ranked by total upvotes cast for a 24 hour period, and then totaled their votes and found what percentage that represents from the whole, its not really representative of the activity on a per community or per thread bases.
Its also strange that this impacts federation too. I mean its very in character but still strange, because the pool of active users only grows when you include remote users, so for these 10 users to make up 60% of the activity, the remote instances would have to be effectively dead, which we know isn’t true.
Let’s assume the site gets 1000 upvote activity a day. That would mean these 10 people represent 600 upvotes a day. But that also implies that there were 600 posts and comments for them to upvote. This could be offset by making their own posts and commets though which always automatically provided 1 vote for themselves. Let’s say there are 1000 users. That would mean rest of the users, all 990 of them only make 400 vote activities a day.
Per user that means the 10 power users create 60 vote activities a day (votes, posts, or comments). That means the other 990 users create 2.47 vote activities a day. That’s 3 posts, 3 comments, 3 votes, or a mix there of a day per user. Considering how easy it is to vote, that seems wildly under what I would assume only casual users would generate. Just scrolling the main page you easily make more then 3 votes.
If this is including federated content that means that the other instances are only every providing 3 voted per user per day in effect. If the power users are a mix of users across instances that means at a per instance level, those 10 users are basically the only high activity users on the network.
I mean, its an interesting puzzle in some respects. However I’d be more worried about the problem that reddit has, where something like 80% of million+ subscriber subreddits were historically moderated by the same group of users, or at least a very narrow group of users. Which, is only an issue if you allow for user created communities instead of administration created communities.
Like what I’m understanding here is that pieces.social doesn’t have “power users” it has low activity? That’s what it seems to me anyway. This obviously is based on only what you know which is just second hand.
This data is really funny. There is so much data not being reported here:
Where are these users from? Local? Remote?
What communities do they vote in?
How much of the votes go to comments vs posts?
How many of the votes are from their own posts and comments? AKA how much do they post and comment?
How many of those votes are cast on front page content vs new content? What is the delta between the time the post/comment was made vs when the vote was cast? Do votes have timestamps?
How does piefed track “active” users? Is this post including only “active” users? Does it track them as active that day, week, month? Is it worth counting people who do not qualify as active that month in the count?
I feel like #5 is important given the framing in the post, “do you think these 147 people should control what you see?”
Because if they don’t vote on “new” content they are not really boosting a post that much.
Its also important to note that they only have 147 votes to cast on each item. So where are they voting? Its also important to know, how many votable items were created that month.
Because you could then identify what the spread is. What is their impact on the total voteable items?
Even with that though. What is the overlap between their interests and the communities interests? If a vote is cast and I’m not subscribed to see it, was a vote even cast at all?
There is so little being done to prove that these 147 people have an outsizes impact on the broader community of content.
What’s also amusing is, this change is pitched in the patch notes as a network stability issue. That implies that the level of votes cast are in some way clogging up the works over at piefed! How could this be? Shouldn’t that be the least impactfull interaction?
A rate limit on voting would solve the network issue, but then its stated in the comments that people would still be able to make “thousands and thousands of votes” even with a rate limit. What is actually the issue here? The rate of votes or the quantity of votes?
I can’t remember of the vote timestamp is exposed via the API or not. But you would think that given the nature of the investigation you would want to know when a vote was cast. Like, adding 147 votes to a post that’s already on the front page means very little.
I don’t think it is in lemmy. Though IDK about piefed, and I can’t be arsed to check. But then, rimu runs piefed.social, he can just go into the database and get info.
ⓘ This user is suspected of being a cat. Please report any suspicious behavior.
That makes little sense. That’s only 10 up votes per post or comment. That sounds like a very low traffic instance lol.
Those posters do way more than than 240. I think rimu did some napkin math so it doesn’t affect +99% of users. But its still just a bad solution. Its like saying the way to stop billionaires is to limit how much one person can spend.
In PJs case its just him upvoting every positive comment he receives. He is sort of like fossilique in just posting normal stuff 99% of the time. Its when he actually gives his opinion that he’s a problem.
Oh sure, but what I’m saying is, for those 10 users, they can’t possibly be doing 60%+ of the activity for all posts. That would mean they would need to be the only ones posting and commenting 60% of the time. It tells me either the organic local activities on the instance are DEAD or there is something wrong with the math. On a thread to thread bases of local post activity they can’t possibly make up 60%+ of the activity.
Right? Like, if what he did was take the total upvotes on the instance for a 24 hour period, then uses ranked by total upvotes cast for a 24 hour period, and then totaled their votes and found what percentage that represents from the whole, its not really representative of the activity on a per community or per thread bases.
Its also strange that this impacts federation too. I mean its very in character but still strange, because the pool of active users only grows when you include remote users, so for these 10 users to make up 60% of the activity, the remote instances would have to be effectively dead, which we know isn’t true.
Let’s assume the site gets 1000 upvote activity a day. That would mean these 10 people represent 600 upvotes a day. But that also implies that there were 600 posts and comments for them to upvote. This could be offset by making their own posts and commets though which always automatically provided 1 vote for themselves. Let’s say there are 1000 users. That would mean rest of the users, all 990 of them only make 400 vote activities a day.
Per user that means the 10 power users create 60 vote activities a day (votes, posts, or comments). That means the other 990 users create 2.47 vote activities a day. That’s 3 posts, 3 comments, 3 votes, or a mix there of a day per user. Considering how easy it is to vote, that seems wildly under what I would assume only casual users would generate. Just scrolling the main page you easily make more then 3 votes.
If this is including federated content that means that the other instances are only every providing 3 voted per user per day in effect. If the power users are a mix of users across instances that means at a per instance level, those 10 users are basically the only high activity users on the network.
I mean, its an interesting puzzle in some respects. However I’d be more worried about the problem that reddit has, where something like 80% of million+ subscriber subreddits were historically moderated by the same group of users, or at least a very narrow group of users. Which, is only an issue if you allow for user created communities instead of administration created communities.
Like what I’m understanding here is that pieces.social doesn’t have “power users” it has low activity? That’s what it seems to me anyway. This obviously is based on only what you know which is just second hand.
OK I found the post.
This data is really funny. There is so much data not being reported here:
I feel like #5 is important given the framing in the post, “do you think these 147 people should control what you see?”
Because if they don’t vote on “new” content they are not really boosting a post that much.
Its also important to note that they only have 147 votes to cast on each item. So where are they voting? Its also important to know, how many votable items were created that month.
Because you could then identify what the spread is. What is their impact on the total voteable items?
Even with that though. What is the overlap between their interests and the communities interests? If a vote is cast and I’m not subscribed to see it, was a vote even cast at all?
There is so little being done to prove that these 147 people have an outsizes impact on the broader community of content.
What’s also amusing is, this change is pitched in the patch notes as a network stability issue. That implies that the level of votes cast are in some way clogging up the works over at piefed! How could this be? Shouldn’t that be the least impactfull interaction?
A rate limit on voting would solve the network issue, but then its stated in the comments that people would still be able to make “thousands and thousands of votes” even with a rate limit. What is actually the issue here? The rate of votes or the quantity of votes?
Lmao what a nightmare.
With utmost respect, do not put this much effort into understanding any fediverse drama. Its not healthy.
Don’t yuck my yum, thanks 😇
Votes have timestamps in my database, though I am running lemmy.
ⓘ This user is suspected of being a cat. Please report any suspicious behavior.
I can’t remember of the vote timestamp is exposed via the API or not. But you would think that given the nature of the investigation you would want to know when a vote was cast. Like, adding 147 votes to a post that’s already on the front page means very little.
I don’t think it is in lemmy. Though IDK about piefed, and I can’t be arsed to check. But then, rimu runs piefed.social, he can just go into the database and get info.
ⓘ This user is suspected of being a cat. Please report any suspicious behavior.