YouTube's next algorithm update will weigh sidebar rank position to reduce 'implicit bias'

Source (technologyreview): https://www.technologyreview.com/s/614432/youtube-algorithm-gets-more-addictive/

Here's the full paper from Google: https://dl.acm.org/citation.cfm?id=3346997

How you can take advantage of this update:

There aren't too many ways in which this kind of update can be exploited.

The main thing you can do in preperation of this update is optimize your title and thumbnail for click through rate.

If you have a video that is ranking low in the sidebar of some popular videos, you can optimize for a higher CTR then your competitors relative to their sidebar ranking. When YouTube makes this update to their algorithm where they start weighing clicks based on ranking - your videos higher CTR with a low sidebar ranking will result in a sidebar rank increase meaning more targeted views (plus engagement and other signals) which can translate into higher search rankings for your target keywords and more recommendations.

Keep in mind, the users clicking your video need to actually interact with it in a way that positively influences the main ranking factors at least as much as viewers are doing so with your competitors videos. Essentially, to get value (in the forum of increased sidebar ranking) from a higher click through rate your video needs to be relevant to the audience and of the same or greater quality then your competitors videos that you are trying to outrank.

Anyways, this paper really isn't much but I thought your guys would find it interesting. We don't get too many serious papers that reveal info on how YouTube's recommendation system actually works, and this definitely provides some new (albeit minor) insights

Interesting.

Machine-learning and AI is definitely the future.

I'm sure this will be a permanent rollout with such a marginal increase in engagement being worth tens of millions of dollars.

So beside the recommendation .. youtube actually rank your video based on ctr???

ehhh... clickbait thumbnail + catchy tittle

they can tweak all day, it always comes down to this, yet they want you to not be clickbaity

This a really clever way to tackle that implicit bias problem, where users are likely to click the top recommended videos thereby creating a feedback loop within only the "top" content and making it harder for lesser videos to rise even if they should based on merit.

If I'm understanding this paper correctly and they actually implement these changes, it could be beneficial to smaller YouTubers working on getting 'traction' and the algo giving their videos a slightly better possibility of succeeding.

There is another change I'd love to see in the algo and that is the autoplay up next giving a greater weight towards selecting videos from the same channel. It's something I've felt like would encourage creators more to promote their work if the algo kept users they were putting an effort into bringing to YouTube (from IG, or wherever) instead of the autoplay algo immediately taking users to a current popular channel on YouTube. If small creators felt like their promo work had a greater impact on their channel growth, you'd get a nice feedback loop with that.

That's just me though.

Thanks for sharing the link to this paper.