This is really interesting information about YouTube recommendations and their role in influencing peoples' political views. They set out to find data that supported 4 claims. First: Radical Bubbles, that recommendations create echo chambers, essentially, which was partially supported. Second: Right-Wing Advantage, that YouTube's recommendation algorithm preferred right wing content. This was not supported. Third: Radicalization Influence, that YouTube made people more radical than they would be. This was also not supported. Fourth: Right-Wing Radicalization Pathway, that... honestly, take a look at it yourself, it sounds like a rewording of earlier points to me, also not supported.

The study included a, "high level view of algorithmic advantages/disadvantages in recommendation impressions" that compared Left, Right and Center channels on YouTube. As with most of the study, it is measuring percentage of increase or decrease of sent vs received recommendations. It showed Left channels receiving +7%, Right receiving -10%, and Center receiving -5%.

The more detailed views that divide groups of channels more finely (they have over 15 of these), unsurprisingly suggest that there is a much larger bias in favor of establishment and corporate media (Partisan Left at +29% and Partisan Right at +27%) than in favor of anti establishment media (Conspiracy at -98% particularly).

Let me list those findings for everyone's ease of reference. They're listed in order, highest to lowest, by recommendations sent (not received, as the -98% received for Conspiracy skews it).

Center/Left MSM -2% Partisan Left +29% Partisan Right +27% Anti SJW -13% Social Justice -34% Provocative Anti-SJW -73% Conspiracy -98% Religious Conservative -62% Libertarian -11% Socialist -25% Anti-Theist -8% White Identitarian -94% MRA -68%

It's worth noting that this grouping of 13 categories is a little different than the grouping of 18 categories given descriptions earlier in the study, that I referred to earlier as "over 15". I find this really interesting and useful in a number of ways. Even just dividing up these political positions in this many ways is, alone, very helpful, I think.

I think a common trend here is for arguments between people whose views may be roughly Social Justice and others whose views may be roughly Anti-SJW to both simultaneously claim that the other side gets more support in the media, the other side is racist, the other side is bad, etc. It's understandable, in a way. Someone who is generally Social Justice is right that Anti-SJW is less silenced than Social Justice by YouTube. And yet, someone who is Anti-SJW is also right that the left on the whole is more supported than the right on the whole.