I've seen a lot of people marvel at how it seems like the YouTube algorithm seems to favor right wing content and pushes it relentlessly when you stumble into certain topics or watch one too many funny compilation videos. Often this is treated as something bordering a Google conspiracy or there's some trick the reactionaries are exploiting - but I'm here to explain that it's the inevitable result of how the algorithm works in the late 2010s (note it does shift somewhat regularly). I am hoping that with the information I'm going to share people will stop talking conspiratorially about THE ALGORITHM, and that lefttube will look to use deliberately what some of what the alt-right is using accidentally.

Sorry. This is really long. But there are two cool diagrams in here for you to enjoy! Took me ages to make 'em!

Firstly: Who am I to know or claim these things? I'm a guy with a small gaming YouTube channel who follows the knowledge dropped by people like Derral Eves (and others) closely. Derral is a consultant and trainer for YouTubers and does a ton of A:B testing and talks to Google a lot. Anyone who wants to understand how YouTube works (to the best of what anyone can figure out) and what behaviors it rewards should go watch everything he's put out (yes, there's hours of it). I also talk daily to multiple channels in the 500k-5m sub range. These are primarily gaming channels but they have contacts and talk about industry trends so I hear some insider baseball now and then from people whose livelihoods depend on the algorithm.

Ok, pre-amble out of the way. Let's get on with it. What's the secret?

YouTube rewards watch time (actual time, not percent of video watched as is often claimed). YouTube wants people to watch a video and see the ads. And then they want people to watch another video, and another, and so-on. Videos do that encourage this behavior from viewers YouTube will reward with better suggestion and placement on home pages and side-bars. Videos that don't encourage this behavior (eg, by driving views away from YouTube to another website or by being short or boring) will not be pushed by YouTube to the hallowed grounds of the Home page.

So now let's look at the alt-right on YouTube. They are unconsciously doing a few things that the YouTube algorithm really likes. They also benefit from having an audience that YouTube likes. The first thing these channels do is produce long videos - this allows watch time to flourish if viewers will stick around. Longer videos means more chances for the creator to put ads in, and more chance for YouTube to profit (Note, typically ads must be at least 7 minutes apart, and having a poke around a few alt-righters they do use 3 or more ads where they can. Not as many seem monetize as I would have thought, however, so I'm actually wondering if a few are banned from doing so).

The next thing that these 'Tubers enjoy are dedicated followers who, I'm expecting,watch most everything these people put out. I'm picking that their audience of young, reactionary males looking to get angry are the ideal YouTube ad consumer, 20s guys with some disposable cash, time on their hands, and an emotional itch to scratch. Further, the algorithm has probably discovered that you can keep serving up "X destroys liberal cucks!" videos (and so on) in the sidebar and this audience will keep watching them, one after the other, across the channels that put this stuff out.

End result is that the YouTube algorithm sees the videos put out by righttube as valuable due to the high total session watch time and, therefore, revenue that they drive Google's way.

In addition to the way these channels benefit from their audience, these channels also do another interesting thing the algorithmically benefits them - they're often quick to jump on topical matters featuring in the news. They're able to do this because they videos are generally poor - as we know - typically without time-consuming things like "facts", "research", and "editing." An alt-right YouTuber can slop out 40 minutes on topical-news-item X while the news is still fresh and doing the rounds in the press. This, combined with the perceived "value" placed on righttube community videos by the YouTube algorithm means that the alt-right can ride the "fresh" and "trending" wave more easily than anyone who has to do research - especially in any current affairs topic.

The alt-right - with their limited range of talking points - also tend to cover the same ground as each other. This leads to the YouTube algorithm identifying a relationship between the channels in this community and thus one vid on topical item X from channel Y will inevitably lead you to being recommended all the other tripe on the same topic from other reactionary channels.

The alt-right end up enhancing the algorithm's assessment of relatedness with all their cross-over videos where they interview each other or appear on each other's shows (richly populating title and tags with each others' names, etc). This seems to have built a "recommendation" community where the YouTube algorithm understands channels are related and will tries to cross-pollinate audiences in order to drive greater total session watch times. You can see YouTube making these connections by visiting any channel home page and seeing the "related channels" bit on the right hand side.

..and this is where I spent an hour making two diagrams to illustrate my point. Ugh. I knew this post would be long.. but heck.. Ok.

So. Let's look at the pictures first. Diagram 1: Lefttube. Diagram 2: Righttube. In both diagrams I started with a seed channel (Contrapoints and Sargon, respectively). I then looked at who was in the "related channels" list for the seeds (it's usually between 4 and 6 channels, by the way). With those noted down I then jumped to the related channels and looked at who was related to these channels, listing who they were linked to.

The results are interesting. In Lefttube there were a total of 27 entities, and 6 mutual (green) linkages made by YouTube. Let's call that a 22% interconnectedness factor. There's a proper technical term for this and anyone who has done graph theory - feel free to chip in - I've forgotten and this is just a bit of quick and dirty analysis.

On Righttube there is a much denser relationship between channels. Only 11 channels featured in the graph and yet there are 6 mutual connections, leading to a interconnectedness factor of 54%. Righttube is TWICE as interconnected as lefttube. Also interesting, righttubers were linked to Fox News (lol), which, I ponder, has an impact on their ability to be suggested alongside more "mainstream" content (ie, you can go from a Fox News article to Sargon talking about the news, to Nazis doing Nazism pretty easily via recommended/suggested videos).

In summary; the result of high watch times on news-responsive and quickly-produced long videos, the collaboration videos righttube likes, and an audience that will eat up this content endlessly means that the algorithm assesses righttube videos as being superb, ad-watch-driving material. This is why once you see one righttube video, you're going to have the entire corpus of that community suggested back to you.

Conversely, lefttube channels are more loosely related and aren't driving the algorithm to reliably recommend one left tube channel's videos to another lefttube channel's audience (well, not as reliably as my little sample of righttube). Anecdotally this fits for me. The mutually linked channels I see in my lefttube graph do get suggested to me a lot, but not much beyond those. For example - although I watch Contra, Shaun, and HBomb religiously - and have done so for months - I only recently got into Philosophy Tube because the other channels talked about him. The algorithm wasn't making the connection between Contra, Shaun and Hbomb reliably for me and so he wasn't popping onto my home page much. As you can see in my diagram he only has one "mutual" with the Contrapoints community.

Compare this with the righttube graph. Watch any Sargon video and you are reliably going to be suggested a lot of other videos forming a tight community of opinion. In fact, "The Thinkery", which is on the far left of that diagram, has related channels which are ALL already on the graph. The algorithm has assessed righttube as very highly related to each other and so once you're in that sphere you're going to find your recommended videos endlessly bounce around the echo chamber every time you open YouTube.

Should lefttube do anything about this? Maybe, maybe not. It might be advisable, if we're getting all strategic here, for lefttube to work on a few more collaboration videos across the various channels. Building strong networks for the algorithm to index would likely mean greater cross-recommendation. Another idea: If there are channels doing very topical, newsy content, they should perhaps be encouraged and supported and brought into the community so that everyone else can ride their recommendation coat-tails.

That being said, I do actually think lefttube is on the ascendancy as more channels take on this work and the high video quality drives much wider interest in the material. Let's compare: Contrapoint's "Incel's" video, now has 1.5m views and 3,890 Google hits for the URL. Meanwhile, none of Sargon's top-three most viewed videos tops more than a few hundred hits on Google for the URL, and his last video to top 1.5m views was two years ago. Righttube is getting repetitive and boring. Maybe?

Ok, I think I'm done. I hope that's helped a few people. And it's really late and my diagrams took too long and - yeah - my methodology is a bit rubbish (take it and do better!), and it's past my bed time. Night!