That's true if you look at the algorithm that is currently being used -- but to only look at the current algorithm is to ignore the many different iterations of the algorithm that preceded it, and the fact that Youtube deliberately altered their algorithm because developers inside the company noticed that the old one was radicalizing people. It's even an oversimplification for me to talk about an "old" versus "new" algorithm, because they are constantly changing -- so a single study is simply not enough to demonstrate or even to meaningfully test whether Youtube was pushing certain users toward radical alt-right content. We know from developers within the company who have now left and blown the whistle on it that this was, in fact, the case. NYT's Rabbit Hole podcast has some great interviews about this:

They/Them. Social Psychologist. My book is out now: