No, community posts aren't broken (from a data scientist's perspective)

In which I debunk Spiffing Brit's latest video about how community posts are supposedly a "glitch" that can be used to game Youtube's recommendation algorithm.

Follow me on twitter! https://twitter.com/vastava_

0:00 Intro
0:42 Would you expect that community posts only go to subscribers?
1:08 There are lots of people “in your community” who aren’t subscribed
1:53 “The algorithm” and misconceptions
2:42 Flawed assumption of how community posts are promoted
3:30 The algorithm takes into consideration user history
4:16 Which explains why community posts are showed to non-subscribers
4:59 Engagement Rate of community posts
5:23 “Engagement” is differently defined for community posts and videos
6:08 Impressions vs. Views
7:23 You also can’t compare likes on posts vs. videos
8:42 Uses are predisposed to interact with polls, not videos
9:11 Why does this mean the feature is “broken”?
9:43 Community posts are working as intended
10:32 Brief recap of my points thus far
10:52 Interaction on Youtube videos is rare
11:12 Community posts solve the “rare feedback” problem
12:00 Community posts drive interaction and engagement
12:32 Let’s talk about viewer demographics
13:22 Why I think the community posts worked so well for Spiffing Brit
13:44 These posts are not getting promoted to *everyone*
14:46 Let’s talk about sample sizes
15:19 This “hack” has more to do with the content, not the feature
16:10 The feature isn’t “broken”
16:34 Responding to pt. 2
17:50 If the feature was so easily “fixed”, was it ever broken?
18:17 I’m not a hater
19:00 Outro

Other videos you might enjoy:
The cultural impact of Content Cop *data analysis* https://www.youtube.com/watch?v=EJpJWYTdtPc
TikTok mansions during pandemic: whY *i made a map* https://www.youtube.com/watch?v=BlrsnoHYW_I&t=4s
Six months later: comparing Shane and Jenna's absences from Youtube | data analysis https://www.youtube.com/watch?v=hu2haO42bmc

#vastava #TheSpiffingBrit #algorithm

Nächste Episode

S01E17 - Did 2020 break Rotten Tomatoes? | data analysis