The abnormal viewership trends was indeed due to duplicated data, but recorded at different intervals.
I don't know if the numbers are fake as fuck, the coding was terrible, or just both.
This makes sense, kind of, it appears the dataset is updated whenever there's a new subscriber, so it doesn't actually correlate to peak viewers, just the viewers at the time of a new subscription?
I suppose storing viewer count based on new subscriptions does add randomness to the data, making it workable for me. This also brings some new opportunities, such as subscriber gain.
Sad, this platform is worthless. Rage pig's subscriber gain. Started off with nearly 1000 at inception.
Each peak in subscriber gain seem to correlate with peaking viewership. The most Ralph ever got was 2000 viewers.
I have a method to definitively prove the existence of this view multiplier, I don't doubt it at all now. I just need to see if it works.