First of all, thanks for the explanation.
By the way, I was not claiming/implying the game in its current status has no issues, I was just wondering - but again, this is an honest uncertainty on my part, with little to no specific background knowledge on the matter - how accurate would be analyzing *that* kind of source.
For example, do we know (and most of all, would it make a difference) whether Nintendo just called a day and uploaded the twitch stream on Youtube or did they actually recorded video material from their own production control room?
(I am assuming you used the former as source)
Well, I believe it to be accurate based on the fact that we see completely even frame pacing from the 720p60 source. When a source is ruined by improper uploading and encoding, that's almost always the first thing to go. I'm 100% certain that their footage is accurate and represents what was playable as I saw exactly the same performance characteristics in person.
Truth be told, the only time compression becomes a real issue for analysis is when dealing with lots of tearing. That usually requires hours of manual tear insertion and examination of the heat-map, among other things.
That's the thing with all of this, while our tools do a very good job, we also check frames individually and can view each frame using ten or so different views.
Basically, what I'm saying is that what you're seeing there is representative of the game's performance in this case and I wouldn't have published it if it were not.
We'll have direct-feed at some point, hopefully, and will be able to do further testing.