| As per Mart's and my post in the Comments and Suggestions thread as well as several others scattered throughout the 4th page of it, there's a problem with the way the current value of discussion is calculated, and to make it short, the problem is how heavily that value incentivizes manipulation: both by posting more and by actively going on the forum to report other posts. It effectively prevents many of us FAL players from also going to the forum to discuss things in earnest to avoid this being seen as manipulation. It really is a big problem and it's ruining our FAL experience. As Kineta suggested, I made this thread specifically to suggest and discuss potential solutions that remove the problematic incentive. My proposals below are focused on both capping the potential damage from singular users and trying to improve the mechanic as a whole to benefit the game. 1. Keep the formula intact but introduce hard limits to scaling. You can limit the total contribution per account to, say, 300 points per title per counting period. So, for instance, a user seeking to boost their anime through discussion in week 6 would can make at most 4 posts that actually count toward the show's total, or 2 posts in week 13. It averages out to two posts per title per week, which I think is quite reasonable in most cases. In addition, you could only count posts made from accounts over a certain age, e.g. at least 16 weeks since creation (total FAL period with registration included). It won't completely prevent abuse, but it would heavily disincentivize compulsive behavior and perhaps make it easier for admins to track premeditated boosting setups. Pros: Effectively limits abuse, penalizes bigger shows to some extent since they get most of the first posts from new users. Cons: Clunky, penalizes a solid chunk of legitimate posts, limits smaller shows' scaling potential. 2. Change the formula to introduce diminishing returns after a threshold. Right now the formula strongly benefits bigger shows proportionally to how big they are. You could make it so they see weaker returns compared to smaller, niche shows, and combat abuse at the same time by applying the following formula when the show hits its 50th post: sqrt(num_post)×530. Here's a table of how the values would look like and a graph to show how it changes over time:   As you can see, post value drops to 1/2 at ~200 posts, and to 1/3 at ~450. This is the kind of scaling many games (notably ARPGs because of their mechanical complexity) employ to combat abuse, and it makes sense because the solution scales itself down quite well and penalizes natural outliers, giving smaller shows a chance to bridge the gap. This would be my preference if we absolutely have to keep discussion in the game. However, it may require some additional limits (softer than in #1) on the amount of posts per account per title to truly be effective against abuse in the principal post ranges for top-10 shows (100..250). Pros: Balances the discussion contribution better, makes boosting much less valuable as it goes on. Cons: May require other kinds of limiting to be more effective against abuse. 3. Remove discussions. Use episode votes instead. Episode votes have one important immutable property: you can only vote for each episode once, so the total contribution per account per show is 13 votes or less. And since the number of votes total is a lot bigger than the number of forum posts (usually 5 to 10 times more), individual contribution finally becomes on par with the other metrics, making extra accounts not worth it whatsoever since you'd only get a fraction of the returns enabled by the current system. On the other hand, if there's any merit to the current system, it's how it sometimes helps smaller shows with well-written, highly complex plots stay afloat in this economy as the ongoing discussion effectively helps them catch up to the much bigger shows (both seasons of The Witch from Mercury are a great example). Moving to votes may penalize them more than the bigger shows. Pros: Simple and bulletproof, more predictable if you don't like chaos. Cons: Worse if you do like chaos, could be an unintentional nerf to the smaller shows. Comments? Suggestions? I think my preference would lie with a combination of the diminishing returns formula, but with an extra limit on the amount of posts per account to further disincentivize boosting which is relaxed enough to avoid penalizing bingers and legitimate discussion too much (so 8–10 posts per account per title in a counting period). I think this would retain the merits of the current system and limit abuse to the level where we probably won't have to think about it or actively monitor for it. In the worst case we can just play with limit values and let the root formula take care of the rest. If any of that feels too complicated, episode votes are a good fallback option that doesn't have that problem. |
| I found the idea thread :) A lot of people were suggesting using episode votes instead of discussion, since it's harder to abuse (1 vote per account per episode). I had a few ideas how this could be done: 1. Count the votes as the value the user gives the episodes. So if the user votes an episode a scire of 5/5, it's worth 5 points when counted, and if the user votes 1/5, it's worth 1 point. If the values are too low then they could all be multiplied by a factor of 10 ir 20 or something. 2. Just take the amount of votes and multiply it by a factor of 75 or something, so each vote is worth the same, regardless of if thr user liked it or not, like how discussion is worth now. 3. You could also instead treat it sinilar to score, by just multiplying the mean vote score (for example, 4.6 or something) by a huge factor like 10000. In all these cases I still imagine them being counted every 2 weeks like discussion is now. I didn't do any balancing or anything with this, just wanted to give some ideas :) |
 Maloween 2024                x16 |
| While I like the idea of using episode votes, I think points from votes is too different from points of actual ep discussions. As of now, while not ideal, discussion points make the game flow less predictable, which I consider a big plus for FAL. If the show is interesting enough, even with a small fanbase and popularity, it can generate a lot of discussions which leads to more points and higher place all in all. So what about votes? I really doubt they will have the same effect as discussions. In the long run they will be pretty much the same as "people watching" just with less points. The more popular show the more ep votes it has, there are no exceptions. But discussions are not always this way and like I've already said, it adds a bit of unpredictability. I also think that "limits" may be a good idea. But I'm not sure whether it's even possible to implement to limit the amount of posts from one user. |
moozooh said: I think we should go with number 3 out of all of your suggestions. Yes im aware that would make the game more predictable but I think we should switch to this system until the FAL moderators can come up with an implement a new system for discussions that wont be easily abusable.As per Mart's and my post in the Comments and Suggestions thread as well as several others scattered throughout the 4th page of it, there's a problem with the way the current value of discussion is calculated, and to make it short, the problem is how heavily that value incentivizes manipulation: both by posting more and by actively going on the forum to report other posts. It effectively prevents many of us FAL players from also going to the forum to discuss things in earnest to avoid this being seen as manipulation. It really is a big problem and it's ruining our FAL experience. As Kineta suggested, I made this thread specifically to suggest and discuss potential solutions that remove the problematic incentive. My proposals below are focused on both capping the potential damage from singular users and trying to improve the mechanic as a whole to benefit the game. 1. Keep the formula intact but introduce hard limits to scaling. You can limit the total contribution per account to, say, 300 points per title per counting period. So, for instance, a user seeking to boost their anime through discussion in week 6 would can make at most 4 posts that actually count toward the show's total, or 2 posts in week 13. It averages out to two posts per title per week, which I think is quite reasonable in most cases. In addition, you could only count posts made from accounts over a certain age, e.g. at least 16 weeks since creation (total FAL period with registration included). It won't completely prevent abuse, but it would heavily disincentivize compulsive behavior and perhaps make it easier for admins to track premeditated boosting setups. Pros: Effectively limits abuse, penalizes bigger shows to some extent since they get most of the first posts from new users. Cons: Clunky, penalizes a solid chunk of legitimate posts, limits smaller shows' scaling potential. 2. Change the formula to introduce diminishing returns after a threshold. Right now the formula strongly benefits bigger shows proportionally to how big they are. You could make it so they see weaker returns compared to smaller, niche shows, and combat abuse at the same time by applying the following formula when the show hits its 50th post: sqrt(num_post)×530. Here's a table of how the values would look like and a graph to show how it changes over time: As you can see, post value drops to 1/2 at ~200 posts, and to 1/3 at ~450. This is the kind of scaling many games (notably ARPGs because of their mechanical complexity) employ to combat abuse, and it makes sense because the solution scales itself down quite well and penalizes natural outliers, giving smaller shows a chance to bridge the gap. This would be my preference if we absolutely have to keep discussion in the game. However, it may require some additional limits (softer than in #1) on the amount of posts per account per title to truly be effective against abuse in the principal post ranges for top-10 shows (100..250). Pros: Balances the discussion contribution better, makes boosting much less valuable as it goes on. Cons: May require other kinds of limiting to be more effective against abuse. 3. Remove discussions. Use episode votes instead. Episode votes have one important immutable property: you can only vote for each episode once, so the total contribution per account per show is 13 votes or less. And since the number of votes total is a lot bigger than the number of forum posts (usually 5 to 10 times more), individual contribution finally becomes on par with the other metrics, making extra accounts not worth it whatsoever since you'd only get a fraction of the returns enabled by the current system. On the other hand, if there's any merit to the current system, it's how it sometimes helps smaller shows with well-written, highly complex plots stay afloat in this economy as the ongoing discussion effectively helps them catch up to the much bigger shows (both seasons of The Witch from Mercury are a great example). Moving to votes may penalize them more than the bigger shows. Pros: Simple and bulletproof, more predictable if you don't like chaos. Cons: Worse if you do like chaos, could be an unintentional nerf to the smaller shows. Comments? Suggestions? I think my preference would lie with a combination of the diminishing returns formula, but with an extra limit on the amount of posts per account to further disincentivize boosting which is relaxed enough to avoid penalizing bingers and legitimate discussion too much (so 8–10 posts per account per title in a counting period). I think this would retain the merits of the current system and limit abuse to the level where we probably won't have to think about it or actively monitor for it. In the worst case we can just play with limit values and let the root formula take care of the rest. If any of that feels too complicated, episode votes are a good fallback option that doesn't have that problem. |
  |
| On the last day of Week 4 of Summer 2023, we've had at least one spammer, if not more, spamming the forums of Kanojo (Rent a Girlfriend S3) and racking up thousands of points with completely useless messages. Judging from one of them, moderators might've deleted previous comments before, but they came back and kept posting. Here's some examples of those posts I took as screenshots, but there was more. This isn't a exhaustive image list. I'm posting this before the deadline, so these posts may well be removed in time, but I thought this would be a good example to talk about the specific dangers of this discussion spam, because the points that Kanojo gets isn't actually the primary problem. That can be potentially bandaged up, so to speak, by subtracting the spammed posts away from W5-6 should they not be dealt with in time. The primary issue is that this spam allows Kanojo to outscore Stray Dogs Season in Week 4. If someone does not have Kekkon (My Happy Marriage), or swaps it out, this would allow them to ace Kanojo. This is important to them because it's a week where Masamune is behind Kanojo, so if they have both, this is a prime chance to ace Kanojo, but Stray Dogs is in the way. (Thank you Ebo for pointing out about Masamune <3). They spam to solve this by strategically spamming discussions to allow for aces in a discussion week. Weeks 4 and 6 are especially vulnerable to this, and Week 8 as well to a slightly lesser extent. If someone only plays week to week then they wouldn't really focus on this, but those of us who do seriously attempt to win every season will be already planning the entire season's aceing strategies right now, if they haven't already. I myself have about half a dozen different plans depending on when a certain anime caps, or the performance of a second one in Weeks 8 or 11. This planning goes on weeks and months in advance, and sometimes anime are picked with specific weeks in mind, like a favourites week or Week 13. Discussion spam not only jeopardies that artificially, it massively reduces one of the only areas of the game after team picking that requires any skill - ace timing. Especially in the last few seasons, where that hasn't been straight forward. We had a similar thing happen in Fall 2022, only this time it did nearly have a critical impact. Fuufu Ijou, Koibito Miman. became aceable due to manipulation of the forums with spam, and was able to be aced over Bocchi the Rock by barely outscoring it. This had an enormous impact, as this was the only week when Fuufu outscored Bocchi at all. Bocchi capped at the very end of Week 10, and Fuufu capped in Week 11. So if Bocchi had capped even a day later, Fuufu would've been unaceable through Bocchi, and the top 10 would've been completely different. I myself nearly got top 10 because of happening to have Fuufu in the team Week 6 and getting that ace through. So even if you aren't a spammer, you will end up being impacted. In that season, the results even inside the top 10 could've been decided by spam. Given that MAL provides prizes, I do feel like the Team would not like the fact that spamming could have such an impact. And it's only discussion that can do this, because it's unlimited. Because it's so high in points. I can post once in one discussion forum in Week 2, and it provides the same points as having an anime favourited for every single favourites week combined. Both 75 points. But I can only favourite once. I can, even legitimately, post at least a dozen times without spamming. A user in the Fall 2022 season provided Mobile Suit Gundam: The Witch From Mercury Season 1 what we believe to be ~10,000 points through legitimate posting alone, as they were arguing with other users. That had an impact too, in allowing that show to outscore Bocchi the Rock and get aced early in the season as well. But whilst some would applaud that as adding some unexpected variety, you can't have that without the potential for spamming at the same time. So overall, I think this just shows that Discussion is awkward as a metric and needs to be tackled. Because Discussion is an unlimited metric it's prone to being inflated, through spamming most of the time, but it also allows shows like say the Witch From Mercury, or perhaps (although it's far too early to say this definitively) Undead Girl Murder Farce to be competitive, so it's also not good to just remove it all together. Personally I think something along the lines of a unique post limit or diminishing returns is needed - there'd need to be a balance between allowing discussions to make certain smaller shows with great or interesting plots to be competitive, but also preventing the spam risk. It's an awkward metric. But I don't think it's sustainable as is, because not only does it provide those points, the impact it can have on aceing is strong. It's nearly dictated entire seasons through certain weeks, and I feel like it's best to do something about it before it could end up deciding one. |
CutieZiaJul 30, 2023 1:38 PM
 |
| I assume there is not an efficient way to simply not count discussion posts from user accounts created after the FAL season starts? It seems like an obvious first step if it was something that was possible. I'm guessing there is not a way to filter these out when tallying though. |
 |
moozooh said: KingYoshi said: I assume there is not an efficient way to simply not count discussion posts from user accounts created after the FAL season starts? It seems like an obvious first step if it was something that was possible. I'm guessing there is not a way to filter these out when tallying though. From what I gather, the current implementation doesn't have any significant coding or any sort of separate post/user database to track this; rather, it looks like it just takes the number of replies from the corresponding threads. So any solution that involves filtering or limiting counts on the user level will have to have all of that created pretty much from scratch. However, the problem with limiting just the FAL accounts is that it still leaves the system vulnerable to alts, and we know boosters use alts to avoid exposing their main account. So regardless of the solution, it needs to be global. Yeah, no doubt it wouldn't solve the overall problem. I just notice that a good chunk of the spam are from accounts made only a handful of days ago. So this would at least eliminate all of those posts from contention. However, it seems like it isn't something easy to do, so probably not worth the work for such a minor fix. |
|
| Hello! I finally made some time to post in this thread. Given the abusable nature of the discussion metric, most hardcore FAL players (including me) would rather have discussions removed than kept in the current state. Many players are suggesting to replace the discussion metric with something else, like episode votes. I find the discussion metric interesting and valuable, and don't want it removed. In my opinion, there are three key changes needed to make discussions less abusable. 1. Nerf value of single posts A single post is worth more points than anything else a player can do to boost an anime. If they post a single time to every episode thread of an anime, they contribute almost one thousand points in total to it. That is way too much power in the hands of individual actors. 2. Limit discussions to only three weeks + week 13. Discussion is the only metric (other than people watching) that is counted more than four times during the season. Counting discussions every other week is stupid and gives players more opportunities to manipulate FAL. All discussion posts made since the last discussion week would be counted instead of only the last two weeks. 3. Count only one post per user per thread. People here said that this might be hard to implement, but it is, in my opinion, absolutely necessary to prevent abuse of the forums. This could be implemented through "hard limits to scaling," as per the first suggestion in moozooh's original post. It doesn't need to be implemented immediately, but the sooner the better. To summarize how it could look if all of my suggestions were implemented: -Discussions would be counted only in weeks 2, 6, 10, and 13. -A single post would count for 15 points during weeks 2, 6, and 10, and 30 points during week 13. -Cheaters would need to make multiple new accounts to boost their anime. I do not like the second proposal in moozooh's original post to "change the formula to introduce diminishing returns after a threshold." Many casual players are already struggling to understand the rules, and this would make it even more confusing. Only changing values for forum posts instead of the entire formula will keep the rules page looking neat. 👍 Thank you for listening to players' opinions. I hope we can resolve this issue to everyone's satisfaction. |
Counting discussion votes seem like a good solution against spamming from one account, and people would have to resort to making multiple accounts to see any significant change in points. However, I still feel it's going to be a problem, since then we won't be able to tell if the vote is legit or not. I'd prefer if instead we could see the number of users discussing the episode, like so: But all of this would still lean on popularity, and any real discussion happening between multiple users won't be counted. So I think anti-spamming measures in episode threads is the way to go, and nerfing value of points + reducing the amount of times discussion is counted (same as WSlav's no.1 and 2 suggestions). WSlav said: I do not like the second proposal in moozooh's original post to "change the formula to introduce diminishing returns after a threshold." Many casual players are already struggling to understand the rules, and this would make it even more confusing. Only changing values for forum posts instead of the entire formula will keep the rules page looking neat. 👍 I don't think it's going to be that much of a problem if the rules are more clear. Just like the score formula (17,500*(score - 6.00) points), it's just using the square root * a fixed number (sqrt(post)*530). Getting rid of the 50 posts limit should make it easier to understand. For anti-spam measures, users should be limited to only a couple of posts per day in discussion threads, and posts need to be longer than the current character limit. If the account is new or hasn't posted anything for a few months, they shouldn't be able to post more than once a day. I know some of my suggestion require feature changes, but I'd really like if MAL can do something about it 😓 |
SamiiAug 14, 2023 4:47 AM
| Reporting from battlefield. Currently helck forum posts are being boosted. Situation on monday
It is currently w7 score week, but this is being done in preparation to make sure helck will be aceable on w8 discussion week. Rent a girlfriend also had lot of spam in w6 but some of those posts were removed before w6 deadline. Something needs to be done about discussion mechanic. |
| For anyone who may be browsing this thread, but doesn't understand the issue. Hard evidence visual example here: (I can't even fit them all in a screenshot because there are 12 of them in a row. Episode 1 of Helck, mind you. This is not ok. This metric needs fixed.) |
KingYoshiAug 14, 2023 10:21 AM
|
| Thanks for your feedback, everyone. For Fall season, we will be switching from # of replies per thread, to # of unique users per thread. Both unique user posts and user votes are susceptible to duplicate account spamming, but the former is more visible (since you cannot easily see which accounts are voting). We'll be monitoring the accounts creating posts to look for suspicious activity, as well. Let's try it for one season and see how it goes. I have an idea for another scaling metric (kind of like the score one) which we can try afterwards if unique user isn't sufficient :nyathinking: |
| An idea to prevent spamming all threads at once: I thought of limiting threads to get points only for next 7 days after airing time of an episode. so when we have a discussion week, we will get whatever points are accumulated/left for the last 2-3 episodes Like that users will only be able to spam one comment at most at a given time, as it's impossible for two episodes to give points at the same time :) |
 |
Reply to Kineta
Thanks for your feedback, everyone. For Fall season, we will be switching from # of replies per thread, to # of unique users per thread.
Both unique user posts and user votes are susceptible to duplicate account spamming, but the former is more visible (since you cannot easily see which accounts are voting). We'll be monitoring the accounts creating posts to look for suspicious activity, as well.
Let's try it for one season and see how it goes. I have an idea for another scaling metric (kind of like the score one) which we can try afterwards if unique user isn't sufficient :nyathinking:
Both unique user posts and user votes are susceptible to duplicate account spamming, but the former is more visible (since you cannot easily see which accounts are voting). We'll be monitoring the accounts creating posts to look for suspicious activity, as well.
Let's try it for one season and see how it goes. I have an idea for another scaling metric (kind of like the score one) which we can try afterwards if unique user isn't sufficient :nyathinking:
@Kineta Unique accounts per thread does not stop spam. People can post in all previous threads waiting for optimal timing to post on their main account or making multiple accounts. People are already doing this tactic again to spam helck and ruin other peoples aces. Group of people asking their friends to post in threads.  Polls can be spammed too but since more people vote in poll points can be lowered from 75 to 7.5, which would mean you would have to make 10 times more accounts to spam poll, which I suppose is not impossible considering how persistent people are on making forum posts... and once you have those 100s of accounts nobody even knows you are using them to spam poll.. Forum situation is especially bad this season since only Zom 100 and Happy Marriage reached ace threshold, meaning it is harder to ace things and people can use discussion to manipulate points for aces. BUT do not be fooled, just because it is more apparent this season does not mean it is not happening in other seasons. Competition really brings out the worst in some people. Here is proof, suddenly nanatsu lost half of its poster once most people had aced it and removed it from active, so spamming it for weekly points regardless of acing was not worth it for lot of people. Considering how volatile forums are, it should not account for 15% of seasons points. It needs to be inline with other metrics like favorites/drops at 10%. Condensing forum weeks to be similar with fav/drop which are counted 4 times total, would help slightly since there would be slightly higher percentage of real posters too and less weeks to manipulate ace timing. Currently every 2 weeks it is possible to manipulate points for acing. |
NMx13Sep 10, 2023 8:00 AM
| There's most definitely a huge problem with discussions. I gave up last FAL cuz of all the spamming on Oshi no Ko. I reported it but nothing happened. I mean one person had maybe 3 posts per page and one post was nothing but exclamation points. There's only 2 possible solutions in my mind: 1- Heavier moderation on the top 10 shows. Mods need to be actively deleting spam posts. If you're not replying to someone, there's no reason to have 10 posts in 1 thread. NONE! 2- Get rid of discussions as a measure all together. Don't replace it. Just be rid of it. We already know that many users have multiple accounts. The screenshot someone posted above was probably 1 person. If you replace it with episode votes, such people will just make multiple accounts. There's no way to stop the crazy. People like that think they're smarter than everyone else and get some kind of gratification in thinking they're putting one over on everyone. It's a sad state of being but one that exists everywhere online. edit: What could happen re: spammers is a penalty. Remove any show with spamming comments from FAL completely. People can move a bench to their active but not add another show. Making them play with 7 instead of 8. Mods should then lock the discussions so the spamming comments can't be deleted effectively putting the culprits on blast letting the majority of players not spamming know who is to blame for the penalty. I'm thinking 1 or 2 seasons of having their favorite little shows removed from participation will be enough to drop the amount of spam replies dramatically. And it doesn't require rewriting any code. It just requires hardball. |
dragynfaerieSep 10, 2023 11:14 PM
moozooh said: A solution that could complement the unique posters system would be to only count posts from threads created within the counting period (+1 week for week 13). This would limit the damage from binge-spam since a unique account would only be able to contribute two posts' worth of points. +1 |
Reply to NMx13
@Kineta Unique accounts per thread does not stop spam. People can post in all previous threads waiting for optimal timing to post on their main account or making multiple accounts. People are already doing this tactic again to spam helck and ruin other peoples aces. Group of people asking their friends to post in threads.
Polls can be spammed too but since more people vote in poll points can be lowered from 75 to 7.5, which would mean you would have to make 10 times more accounts to spam poll, which I suppose is not impossible considering how persistent people are on making forum posts... and once you have those 100s of accounts nobody even knows you are using them to spam poll..
Forum situation is especially bad this season since only Zom 100 and Happy Marriage reached ace threshold, meaning it is harder to ace things and people can use discussion to manipulate points for aces. BUT do not be fooled, just because it is more apparent this season does not mean it is not happening in other seasons. Competition really brings out the worst in some people. Here is proof, suddenly nanatsu lost half of its poster once most people had aced it and removed it from active, so spamming it for weekly points regardless of acing was not worth it for lot of people.
Considering how volatile forums are, it should not account for 15% of seasons points. It needs to be inline with other metrics like favorites/drops at 10%. Condensing forum weeks to be similar with fav/drop which are counted 4 times total, would help slightly since there would be slightly higher percentage of real posters too and less weeks to manipulate ace timing. Currently every 2 weeks it is possible to manipulate points for acing.
Polls can be spammed too but since more people vote in poll points can be lowered from 75 to 7.5, which would mean you would have to make 10 times more accounts to spam poll, which I suppose is not impossible considering how persistent people are on making forum posts... and once you have those 100s of accounts nobody even knows you are using them to spam poll..
Forum situation is especially bad this season since only Zom 100 and Happy Marriage reached ace threshold, meaning it is harder to ace things and people can use discussion to manipulate points for aces. BUT do not be fooled, just because it is more apparent this season does not mean it is not happening in other seasons. Competition really brings out the worst in some people. Here is proof, suddenly nanatsu lost half of its poster once most people had aced it and removed it from active, so spamming it for weekly points regardless of acing was not worth it for lot of people.
Considering how volatile forums are, it should not account for 15% of seasons points. It needs to be inline with other metrics like favorites/drops at 10%. Condensing forum weeks to be similar with fav/drop which are counted 4 times total, would help slightly since there would be slightly higher percentage of real posters too and less weeks to manipulate ace timing. Currently every 2 weeks it is possible to manipulate points for acing.
| @NMx13 Thank you for the screenshot. I see what you mean now about opportune timing. It's too late to adjust things for Fall season now, but I'll keep an eye out for it and consider ways we can combat this. |
Reply to dragynfaerie
There's most definitely a huge problem with discussions. I gave up last FAL cuz of all the spamming on Oshi no Ko. I reported it but nothing happened. I mean one person had maybe 3 posts per page and one post was nothing but exclamation points. There's only 2 possible solutions in my mind:
1- Heavier moderation on the top 10 shows. Mods need to be actively deleting spam posts. If you're not replying to someone, there's no reason to have 10 posts in 1 thread. NONE!
2- Get rid of discussions as a measure all together. Don't replace it. Just be rid of it. We already know that many users have multiple accounts. The screenshot someone posted above was probably 1 person. If you replace it with episode votes, such people will just make multiple accounts. There's no way to stop the crazy. People like that think they're smarter than everyone else and get some kind of gratification in thinking they're putting one over on everyone. It's a sad state of being but one that exists everywhere online.
edit: What could happen re: spammers is a penalty. Remove any show with spamming comments from FAL completely. People can move a bench to their active but not add another show. Making them play with 7 instead of 8. Mods should then lock the discussions so the spamming comments can't be deleted effectively putting the culprits on blast letting the majority of players not spamming know who is to blame for the penalty. I'm thinking 1 or 2 seasons of having their favorite little shows removed from participation will be enough to drop the amount of spam replies dramatically. And it doesn't require rewriting any code. It just requires hardball.
1- Heavier moderation on the top 10 shows. Mods need to be actively deleting spam posts. If you're not replying to someone, there's no reason to have 10 posts in 1 thread. NONE!
2- Get rid of discussions as a measure all together. Don't replace it. Just be rid of it. We already know that many users have multiple accounts. The screenshot someone posted above was probably 1 person. If you replace it with episode votes, such people will just make multiple accounts. There's no way to stop the crazy. People like that think they're smarter than everyone else and get some kind of gratification in thinking they're putting one over on everyone. It's a sad state of being but one that exists everywhere online.
edit: What could happen re: spammers is a penalty. Remove any show with spamming comments from FAL completely. People can move a bench to their active but not add another show. Making them play with 7 instead of 8. Mods should then lock the discussions so the spamming comments can't be deleted effectively putting the culprits on blast letting the majority of players not spamming know who is to blame for the penalty. I'm thinking 1 or 2 seasons of having their favorite little shows removed from participation will be enough to drop the amount of spam replies dramatically. And it doesn't require rewriting any code. It just requires hardball.
dragynfaerie said: I cant agree with this last suggestion here and here is why. Say there was a dark horse of the season that alot of people missed out on including spammers. That would create incentive for spammers to target that anime to remove it from the competition. In other words all it does is change the reason for spamming not getting rid of it itself.edit: What could happen re: spammers is a penalty. Remove any show with spamming comments from FAL completely. People can move a bench to their active but not add another show. Making them play with 7 instead of 8. Mods should then lock the discussions so the spamming comments can't be deleted effectively putting the culprits on blast letting the majority of players not spamming know who is to blame for the penalty. I'm thinking 1 or 2 seasons of having their favorite little shows removed from participation will be enough to drop the amount of spam replies dramatically. And it doesn't require rewriting any code. It just requires hardball. |
|
| @Kineta So it's been just over 1.5 seasons since the last forum point count change, and I think it's a good time to draw some conclusions and perhaps make adjustments for the coming season. I think discarding non-unique users has been a major step in the right direction, but it evidently didn't solve the issue to the extent where we can leave it unattended. Forum spam, especially coordinated with multiple people posting in every episode discussion thread during the same week, is still the single most lucrative attack vector and far above any other means of manipulating the results. I think we're reasonably close to mitigating this problem, but at least one more action is needed in my opinion to move it where we ideally want it to be. Right now each user's post is effectively counted once per discussion thread but without considering when it is made, meaning one could still strategically withhold their posts for the most opportune week to attack the scoring system, and bringing in more users still scales this effort very well. What I propose is to limit it instead by counting up to a fixed number of posts per user per timing period: not more than 2 per anime title (1 double-value post on week 13). In other words, no matter when you decide to post, and no matter the thread (you could post in the same thread for all you want), your contribution in a given time period is limited to the same value. Nothing you do can exceed this value, short of registering a dummy account—and even then it can only contribute up to that much within the period. To compensate for the potential reduction in post count (contribution from legitimate bingers is particularly penalized in this manner), I suggest increasing the value of a single post back to 100 (200 for week 13). I've made an illustration below comparing the two systems and assuming a user does everything to maximize the value of their posts in given weeks.  As you can see, the proposed system actively discourages coordinated boosting and usage of dummy accounts because the value of each new person/account contribution cannot exceed 200 per counting period, making it not worth the time and effort to do so. It also discourages withholding of posts for any reason. On the other hand, it actively encourages healthy discussion scenarios, as posting regularly will net more points for a given title compared to offloading all posts at the same time, meaning we can finally stop frowning upon FAL players discussing a show they're watching or feeling bad about ever posting in discussion threads even if we do want to discuss it ourselves. If nothing else, this is a step up in mental well-being. :D For balancing purposes, the overall value of forum discussion per title should slightly decrease as the result of this change, coming at the expense of discounted end-of-season bingers, but discussion in the first half of the season becomes more valuable in comparison. This will likely shake up the meta slightly in favor of shows with a high volume of discussion relative to score or viewership (like this season's Mahou Shoujo ni Akogarete and Metallic Rouge). I consider this a good thing. |
Reply to moozooh
@Kineta
So it's been just over 1.5 seasons since the last forum point count change, and I think it's a good time to draw some conclusions and perhaps make adjustments for the coming season.
I think discarding non-unique users has been a major step in the right direction, but it evidently didn't solve the issue to the extent where we can leave it unattended. Forum spam, especially coordinated with multiple people posting in every episode discussion thread during the same week, is still the single most lucrative attack vector and far above any other means of manipulating the results. I think we're reasonably close to mitigating this problem, but at least one more action is needed in my opinion to move it where we ideally want it to be.
Right now each user's post is effectively counted once per discussion thread but without considering when it is made, meaning one could still strategically withhold their posts for the most opportune week to attack the scoring system, and bringing in more users still scales this effort very well.
What I propose is to limit it instead by counting up to a fixed number of posts per user per timing period: not more than 2 per anime title (1 double-value post on week 13). In other words, no matter when you decide to post, and no matter the thread (you could post in the same thread for all you want), your contribution in a given time period is limited to the same value. Nothing you do can exceed this value, short of registering a dummy account—and even then it can only contribute up to that much within the period.
To compensate for the potential reduction in post count (contribution from legitimate bingers is particularly penalized in this manner), I suggest increasing the value of a single post back to 100 (200 for week 13). I've made an illustration below comparing the two systems and assuming a user does everything to maximize the value of their posts in given weeks.
As you can see, the proposed system actively discourages coordinated boosting and usage of dummy accounts because the value of each new person/account contribution cannot exceed 200 per counting period, making it not worth the time and effort to do so. It also discourages withholding of posts for any reason. On the other hand, it actively encourages healthy discussion scenarios, as posting regularly will net more points for a given title compared to offloading all posts at the same time, meaning we can finally stop frowning upon FAL players discussing a show they're watching or feeling bad about ever posting in discussion threads even if we do want to discuss it ourselves. If nothing else, this is a step up in mental well-being. :D
For balancing purposes, the overall value of forum discussion per title should slightly decrease as the result of this change, coming at the expense of discounted end-of-season bingers, but discussion in the first half of the season becomes more valuable in comparison. This will likely shake up the meta slightly in favor of shows with a high volume of discussion relative to score or viewership (like this season's Mahou Shoujo ni Akogarete and Metallic Rouge). I consider this a good thing.
So it's been just over 1.5 seasons since the last forum point count change, and I think it's a good time to draw some conclusions and perhaps make adjustments for the coming season.
I think discarding non-unique users has been a major step in the right direction, but it evidently didn't solve the issue to the extent where we can leave it unattended. Forum spam, especially coordinated with multiple people posting in every episode discussion thread during the same week, is still the single most lucrative attack vector and far above any other means of manipulating the results. I think we're reasonably close to mitigating this problem, but at least one more action is needed in my opinion to move it where we ideally want it to be.
Right now each user's post is effectively counted once per discussion thread but without considering when it is made, meaning one could still strategically withhold their posts for the most opportune week to attack the scoring system, and bringing in more users still scales this effort very well.
What I propose is to limit it instead by counting up to a fixed number of posts per user per timing period: not more than 2 per anime title (1 double-value post on week 13). In other words, no matter when you decide to post, and no matter the thread (you could post in the same thread for all you want), your contribution in a given time period is limited to the same value. Nothing you do can exceed this value, short of registering a dummy account—and even then it can only contribute up to that much within the period.
To compensate for the potential reduction in post count (contribution from legitimate bingers is particularly penalized in this manner), I suggest increasing the value of a single post back to 100 (200 for week 13). I've made an illustration below comparing the two systems and assuming a user does everything to maximize the value of their posts in given weeks.
As you can see, the proposed system actively discourages coordinated boosting and usage of dummy accounts because the value of each new person/account contribution cannot exceed 200 per counting period, making it not worth the time and effort to do so. It also discourages withholding of posts for any reason. On the other hand, it actively encourages healthy discussion scenarios, as posting regularly will net more points for a given title compared to offloading all posts at the same time, meaning we can finally stop frowning upon FAL players discussing a show they're watching or feeling bad about ever posting in discussion threads even if we do want to discuss it ourselves. If nothing else, this is a step up in mental well-being. :D
For balancing purposes, the overall value of forum discussion per title should slightly decrease as the result of this change, coming at the expense of discounted end-of-season bingers, but discussion in the first half of the season becomes more valuable in comparison. This will likely shake up the meta slightly in favor of shows with a high volume of discussion relative to score or viewership (like this season's Mahou Shoujo ni Akogarete and Metallic Rouge). I consider this a good thing.
| I also think limiting the number of posts per counting period is good idea, would help in reducing the amount of time spent on dealing with spam. moozooh said: To compensate for the potential reduction in post count (contribution from legitimate bingers is particularly penalized in this manner), I suggest increasing the value of a single post back to 100 (200 for week 13). Also yes to increasing the value if this idea is implemented. |
| First, I want to thank the MAL mod team for helping us to fix the issue. You have been amazing. I don't have any great ideas on my own so I will give my support for the one suggested by moozoh. Making posts count only during the "discussion week" period is very good for both the players and the mods. I think this is the easiest and least intrusive solution for now that keeps the system the same and limits the ability to hoard unique posts. However, it does not fix the elephant in the room which is bot accounts and "dummy" accounts. Making the site more durable against fake account registration would be the ultimate fix, but I'm nowhere near qualified enough to give suggestions on this matter and I don't think the FAL team can do smth about that either. Keep up the great work. Broken or not, we still love FAL |
 |
| We are currently still facing issues with users attempting to manipulate discussion points. This is very annoying for us FAL players, but I can imagine it’s even more frustrating and time consuming for the MAL mods. To maintain discussion as a metric we need to implement some sort of balancing. I think moozooh has given us a promising solution for this problem. Even though it won’t completely shut down the problem it should significantly reduce the impact. I think this is a big step in the right direction, so I'm supporting his idea. Additionally, I hope we can find a way to address the issue of alt/bot accounts being used for manipulation. For example 'Date A Live V' was affected by these burner accounts in the later weeks (and it definitely wasn't the only anime affected). I’m not really sure myself how we should resolve this though… At least moozoohs solution should make it less annoying to deal with for us players and the moderators too. Maybe someone else has an idea. |
 |
| I'm very thankful for all MAL mods who went around dealing with the spam when it was very bad but it must have been very time-consuming, continuing to deal with it in the same manner isn't very efficient. So any solution, even if just temporary is very appreciated. I've seen regular FAL players contemplating whether to stop playing because of all the alts and bots, too, and I wouldn't want to see them quit because of that. Also, I think the bot problem isn't just FAL-related as it also greatly affects how genuine MAL statistics are. Sorry that I don't have any valuable feedback but I just wanted to share concerns regarding the problem. |
More topics from this board
» Fall 2024 Registrations Now Open!Kineta - Sep 15 |
26 |
by Hattyson
»»
Oct 29, 2:33 AM |
|
» A comprehensive guide to Fantasy Anime Leaguemoozooh - Mar 25, 2023 |
12 |
by moozooh
»»
Oct 8, 6:04 AM |
|
Sticky: » Comments and Suggestions ( 1 2 3 4 5 )Naruleach - Mar 25, 2012 |
214 |
by Samii
»»
Aug 9, 7:10 AM |
|
» Spring 2024 Week 13: Ace/Swap Reminder & Discussion (final week)eplipswich - Jun 23 |
3 |
by badabass
»»
Jul 1, 6:52 AM |
|
» Spring 2024 Week 12: Ace/Swap Reminder & Discussioneplipswich - Jun 16 |
6 |
by kernel99
»»
Jun 27, 5:33 AM |