[ruminator1]

The goal of the YouTube algorithm is not to give you great videos but keep you mildly entertained for long periods of time.

In my experience the 'suggested videos' next to the video you're watching is more than good enough to get recs.

I use newpipe and put all the videos I like in a playlist. So everytime I want new videos I just scroll through the list, click on a video and just try something from suggested videos.

[charcircuit]

Being shown great videos that are relevant to you is correlated with increased time on the platform and satisfaction which are some of the metrics YouTube cares about.

>the 'suggested videos' next to the video you're watching is more than good enough to get recs.

Which is also an algorithmic feed which optimizes for the same metrics as the home page.

[brookst]

Let’s say you’ve got a tire you can’t seat on a wheel. Which of these is the YouTube algorithm going to optimize for:

1) a 60 second video showing how to use wd-40 and a lighter to set the bead, or

2) a 7 minute showing how beads can be broken, things you might try, an ad, some bad ideas, a funny guy yelling, another ad, and then how to set a bead with flame

Increased time on the platform is a proxy for more ads shown. YouTube optimizes for maximizing ads, not minimizing time on the platform with short, clear, informative content.

[charcircuit]

Let's do an informal experiment. In incognito I searched "how to seat a tire on a wheel"

The first result is 2:43 then 3:14 then a shorts drawer which are sub minute videos then 5:43 then 3:29.

It turns out that YouTube is not optimizing for showing 7 minute videos.

>YouTube optimizes for maximizing ads

While they do monitor for regressions of metrics like ad revenue improvements to the algorithm aim to improve the user experience as opposed to improving revenue.

[brookst]

That's funny. When I do exactly the same search on Youtube (incognito, but Chrome incognito so they still know my profile), I get videos of times 2:43, 10:20, 7:49, 7:37, and 7:42.