| Over the past few weeks, I have seen the scores of obscure anime increase significantly. I found a stack documenting the phenomenon, and I have observed similar increases in other titles not found on the stack. https://myanimelist.net/stacks/44765 As I check each title's stats, I see the new scores do not correspond to votes from new viewers. The whole thing is quite confusing. I previously posted this on another board, but I think the Anime DB board is more appropriate for a thread about the anime database. |
LucifrostMay 16, 4:34 PM
| その目だれの目？ |
| These anime have suddenly all returned to their low scores. |
| その目だれの目？ |
| There are two factors that could be at play here that may explain the changes to scores without the number of members/votes changing much : - MAL ruling out more accounts as "fake" (they'll show up in the overall stats, but won't be counted) - MAL tinkering with the hidden "minimum number of votes" parameter in score calculation - if you look at the formula, you'll see it pushes low-vote-count entries towards the overall mean. Since both of these factors are hidden from public view, we can only guess what's going on. |
| It would seem like shows with low number of votes got hit harder by this change, most shows with less than 10k votes lost between 0.05 and 0.1 while popular shows scores didn't change, just an observation |
Reply to Darkerplayer
The root cause was almost certainly due to botting with private lists. All shows majorly affected by this have had a steady and much quicker than usual growth in users since ca. mid April that coincided with their similarly steady score rises. Since these accounts do not show up in the recent updates, it's only really detectable by continuously tracking score and user count. The amount of bots is in the range of ca. 300, so it didn't have much of an effect on popular shows, but it heavily impacted obscure shows (particularly older ones, whoever is commanding the bots is using a set of lists that is primarily composed of old shows). The biggest change that I'm aware of was Game Tengoku, which jumped from 5.6 to 7.2, and a couple dozen shows had jumps in the range of 1 full rating point.
People have been tracking this data and have reported it to MAL a while ago, so I assume they've figured out a way to filter out these bots specifically.
People have been tracking this data and have reported it to MAL a while ago, so I assume they've figured out a way to filter out these bots specifically.
| @Darkerplayer primarily composed of old shows seasonal shows got just as affected after the recent change |
Tbh I think it is just another MAL's fcked up change. Look at that. Top 5 of all time. Is that novel really THAT goated? Would be better to leave the scores as they were instead of unnecessarily meddling. |
18927May 24, 5:15 AM
18927 said: Look at that. Top 5 of all time. Is that novel really THAT goated? He's almost unanimously loved by fans of Chinese stories (especially Wuxia and Boys Love), and the official English translations are best sellers, listed by the New York Times... Hate it or question it all you want, it somehow really is up there. |
 "Genius lives only one storey above madness." – Arthur Schopenhauer. "Stupidity is a talent for misconception." – Edgar Allan Poe. "I'm tired... and hungry." – Alexioos95. |
18927 said: And yet it was ranked #28 in March. Yeah, with less than 4.000 members. As the scores a weighted, such low popularity brings down the score down quite a lot (it was at 8.84). So MAL changed either the requirement of members for the calculation, or the script to detect bots. (Edit : It's said to be about the bots ?) |
Alexioos95May 24, 5:41 AM
"Genius lives only one storey above madness." – Arthur Schopenhauer. "Stupidity is a talent for misconception." – Edgar Allan Poe. "I'm tired... and hungry." – Alexioos95. |
Reply to Darkerplayer
The root cause was almost certainly due to botting with private lists. All shows majorly affected by this have had a steady and much quicker than usual growth in users since ca. mid April that coincided with their similarly steady score rises. Since these accounts do not show up in the recent updates, it's only really detectable by continuously tracking score and user count. The amount of bots is in the range of ca. 300, so it didn't have much of an effect on popular shows, but it heavily impacted obscure shows (particularly older ones, whoever is commanding the bots is using a set of lists that is primarily composed of old shows). The biggest change that I'm aware of was Game Tengoku, which jumped from 5.6 to 7.2, and a couple dozen shows had jumps in the range of 1 full rating point.
People have been tracking this data and have reported it to MAL a while ago, so I assume they've figured out a way to filter out these bots specifically.
People have been tracking this data and have reported it to MAL a while ago, so I assume they've figured out a way to filter out these bots specifically.
| @Darkerplayer Thank you so much for providing a plausible explanation! A mod locked the last thread I made to draw attention to this issue, and I worried nothing else would be done. |
| その目だれの目？ |
More topics from this board
Sticky: » Anime DB - Post Modification Requests Here ( 1 2 3 4 5 ... Last Page )cyruz - May 15, 2016 |
8759 |
by Barbaduk
»»
Yesterday, 5:56 AM |
|
» GIRLS’F1ST!!!!GTGinIonSui - May 25 |
7 |
by GinIonSui
»»
May 30, 4:47 PM |
|
» Flipnote warriors/Ugomemo SenshiSilverlight41 - May 28 |
3 |
by Silverlight41
»»
May 30, 12:00 PM |
|
» jojo addEldencheese - May 30 |
0 |
by Eldencheese
»»
May 30, 11:10 AM |
|
Sticky: » Anime DB - Post Genre/Theme Requests Here ( 1 2 3 4 5 ... Last Page )Kineta - Mar 28, 2022 |
711 |
by Stormy_77
»»
May 28, 9:12 PM |