[azemetre]

It might also give worse insights on how you interpret the data. I'd imagine that those with huge followers/events will have most people grit through the ads. I myself do this when watching esports tournaments. I don't mind the pre-roll because I'll be there for 30+ minutes.

But as you and the other poster mentioned, that wouldn't necessarily be true for smaller streamers. In fact I share the sentiments posted elsewhere; where if you want to check out a smaller channel but don't want to waste 1.5 mins in ads. I'll just immediately bounce.

It's a doom cycle were those with popularity will always out compete smaller streamers. Smaller streamers get cannibalized by site's practices.

The only thing missing is a competitor but as we see from past competitor's streamers/viewers seem to be a very fickle bunch. They like the platform not the streamers, but the platform is now hostile.

[jerojero]

If you do A/B testing you should be trying to control as many variables as possible.

It makes sense for twitch, given its skewed distribution, to compare the behaviour of users engaging with the different types of streamers.

Under this logic, I presume they might actually see that engagement is not substantially harmed for big time streamers while getting actually millions of views on their ads as opposed to engagement harmed for low viewership streamers who don't really give the platform much money anyway.

I think it's a flaw in your thinking to assume twitch might have the interests of small streamers in mind. They probably don't. Once a big streamer retires, another takes their place. The viewership remains.