There is some obvious bot activity in Superman & Lois ratings

[TheUnsatisfiedConsum (TheUnsatisfiedConsumer)]

Problem

I just happened to be browsing through some of my past ratings and saw that the general consensus on Superman and Lois were abnormally lower than I remembered. When I went to check, the ratings are extremely skewed and irregular, specifically the season ratings.

I haven’t seen such voting spikes outside of mainstream “hyped” titles that tend to yield high positive scores. Even with popular mainstream titles, the votes are not as clearly CONCENTRATED on a specific score like they are here. Usually the ratings average out, even if most of the score weight sits in one general section. Here, the data is quite biased and abnormal.

Show Overall Ratings (Normal)

Season 1 (Irregular Distribution)

7 Star Vote Count (66%):

Season 2 (Irregular Distribution)

2 Star Vote Count (67%):

Season 3 (Irregular Distribution)

1 Star Vote Count (64%):

Season 4 (Irregular Distribution)

3 Star Vote Count (73%):

Concern

Is this a case of multiple bot accounts (social media farm)? Or is it a worse case, with accounts being compromised having them rate the show lower? Could this be an API endpoint being abused by an extension or third party plugin?

Has anyone seen any other skews like this in other Shows/Movie titles?

[CanePlayz (Jacob P.)]

Good catch, hope some investigation will shed light onto this

[Seerjo]

Welp there goes me thinking trakt ratings are better because no one would bother manipulating them.

[kcador (Kevin)]

Thanks for flagging this, and for the detailed breakdown.

We investigated this case and we can already confirm that this is not bot activity or intentional rating manipulation. The unusual season rating distributions were caused by duplicated season ratings created through one of our first-party app’s sync flow.

The issue that was generating those duplicated ratings has already been fixed, so the data affected is contained. Because, like you, we think ratings are only useful if they reflect real user activity, and incorrect data can affect general trust in Trakt, we’re now working through a cleanup phase: we are identifying the duplicated season ratings, keeping the original valid rating where possible, and removing the extra incorrect entries.

Thanks again for reporting it with such clear examples. That made it much easier to confirm what was going on.