So here are the results of my viewbot hypothesis testing from earlier today. Similar to my previous post, green dot represents my stream being online and red dot represents buffering. My methodology was that I explicitly asked my viewers to remain in my stream for the 5 minute downtime period. Unfortunately, it took longer than expected for my OBS to re-establish connection with YouTube for some reason.
My stream:
View attachment 3242240
DSP's stream from earlier today:
View attachment 3242247
We see a similar dip in our streams, obviously he has over double the initial viewership but the overall shape seems to be mostly the same. My conclusion is that I would hesitate to use the fact that he drops "x viewers over a very short period" as evidence on it's own that he is botting.
The real question I suppose would be, "is there any possibility that the initial 800+ viewers consists of view bots?" I don't think it can be answered utilizing this one thing. Again, the only question I wanted answered was if under ideal conditions (i.e. viewers were told ahead of time NOT to leave) we could show some kind of major difference (i.e. the viewer count is able to hold steady) and that clearly was not the case here.
Questions like, "How does he have 800+ viewers and a dead chat" are probably more indicative of botting at the end of the day, although I don't think we can definitively prove if he doesn't just have a significantly higher percentage of lurkers compared to an average streamer (I suspect this is a true statement).