Andrea Jones-Rooy

homophily homogeneity of the echo chamber.

And so how do you look for that on the internet?

Well, in the sake of time, I'm not going to go through all the experiments, but I will say that basically what they do is a bunch of different types of experiments.

And again, none of these are perfect and they're pretty limited, frankly, because we don't have access to the full algorithm.

where they do things called an audit study, or they do something called a sock puppet study.

So there's a bunch of different ones.

And a sock puppet study is where you build an account, just like a brand new account with no history, no viewer history, and you sit it in front of videos and see what's recommended.

And so one of the experiments they did is they basically said, okay, we're going to take a whole bunch of participants.

We're going to create neutral, you know, no browser history, no nothing results.

And then we're going to create a toy little experiment playground where they start with this one of the same 24 videos.

And then they're going to click on that video.

And then we're going to see what YouTube...

gives them.

And for some of the people in the experiment, and they did a bunch of different versions of this, but the gist of it is it's some people in the experiment, they said, okay, start with these 24.

Everyone's gonna start with the same 24 videos to begin.

Click on whatever you want.

And that's where the attention, the interest, like just the user, you're on YouTube, click, click, click, click, click, see where it goes.

But the others, they said, okay, because we wanna disentangle user choice from what's actually kind of powering the algorithm,

We're going to give you different rules to try.

So sometimes it'll say, always click on the second video or always click on the video with more odd numbers or whatever sort of rule, right?