Abstract
Content recommendations on big internet platforms such as YouTube are supported by crowd-based recommender systems. Users are presented with what other users in comparable situations spent time on. There are several examples, however, where sequences of suggested content quickly degenerated towards only slightly related, sometimes problematic content that we define as contextually inappropriate. We try to identify the basis of this effect from both the user and the technical side, and set up a simple simulation system in order to better understand the interactions. Simulation results provide evidence that autoplay is an especially problematic feature, but that completely preventing inappropriate suggestions is technically very hard if not infeasible because it is the nature of a recommendation system to take into account feedback from the user and adapt to it. We also propose some possible measures to mitigate the problem.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
- 2.
For a general idea on how embeddings work see: https://towardsdatascience.com/neural-network-embeddings-explained-4d028e6f0526.
- 3.
It may be difficult to imagine such a huge space since we experience the world around us as only three dimensional. In principle we could also break this high dimensional space into a large number of 3D spaces, in our case around 85 3D cubes would be necessary. A point (video in our case) would then have a 3D position in each of these cubes and all cubes together provide its position in 255 dimensions.
- 4.
References
Anderssen, R.S., Brent, R.P., Daley, D.J., Moran, P.A.P.: Concerning \(\int _0^1 \cdots \int _0^1 {(x_1^2 + \cdots + x_k^2)} ^{{1 / 2}} dx_1 \cdots, dx_k \) and a Taylor series method. SIAM J. Appl. Math. 30(1), 22–30 (1976)
Bridle, J.: Something is wrong on the internet. Medium (2017). https://medium.com/@jamesbridle/something-is-wrong-on-the-internet-c39c471271d2
Chaslot, G.: Youtube’s A.I. was divisive in the us presidential election. Medium (2016). https://medium.com/the-graph/youtubes-ai-is-neutral-towards-clicks-but-is-biased-towards-people-and-ideas-3a2f643dea9a
Covington, P., Adams, J., Sargin, E.: Deep neural networks for YouTube recommendations. In: Proceedings of the 10th ACM Conference on Recommender Systems, New York, NY, USA (2016)
Fisher, M., Taub, A.: On YouTube’s digital playground, an open gate for pedophiles. The New York Times (2019). https://www.nytimes.com/2019/06/03/world/americas/youtube-pedophiles.html
Google. How google fights disinformation (2019). https://www.blog.google/documents/37/How_Google_Fights_Disinformation.pdf
Kaiser, J., Rauchfleisch, A.: The implications of venturing down the rabbit hole. Internet Policy Review, June 2019. https://policyreview.info/articles/news/implications-venturing-down-rabbit-hole/1406
Khan, M.L.: Social media engagement: what motivates user participation and consumption on YouTube? Comput. Hum. Behav. 66, 236–247 (2017)
Maheshwari, S.: Chase had ads on 400,000 sites. Then on just 5,000. Same results. The New York Times, March 2017
Maheshwari, S.: On YouTube kids, startling videos slip past filters. The New York Times (2017). https://www.nytimes.com/2017/11/04/business/media/youtube-kids-paw-patrol.html
Maheshwari, S.: Publishers retreat from the risks of Google-YouTube advertising. The New York Times (2017)
Maheshwari, S., Wakabayashi, D.: AT&T and Johnson & Johnson Pull Ads From YouTube. The New York Times (2017)
Martineau, P.: YouTube has kid troubles because kids are a core audience. Wired (2019). https://www.wired.com/story/youtube-kid-troubles-kids-core-audience/
McQuail, D.: Mass Communication Theory: An Introduction. Sage Publications, London (1983)
Meyerson, E.: YouTube now: why we focus on watch time (2012). https://youtube-creators.googleblog.com/2012/08/youtube-now-why-we-focus-on-watch-time.html
Naughton, J.: How Peppa Pig knock-offs bring home the bacon for Google. The Guardian (2017). https://www.theguardian.com/commentisfree/2017/nov/12/content-google-youtube-kids-not-always-suitable-for-children-peppa-pig-brings-home-bacon
Smith, A., Toor, S., van Kessel, P.: Many turn to YouTube for children’s content, news, how-to lessons. pewresearch.org (2018). https://www.pewresearch.org/internet/2018/11/07/many-turn-to-youtube-for-childrens-content-news-how-to-lessons/
Solsman, J.E.: Ever get caught in an unexpected hourlong YouTube binge? Thank YouTube AI for that. cnet.com (2018). https://www.cnet.com/news/youtube-ces-2018-neal-mohan/
Grimme, C., Preuss, M., Takes, F.W., Waldherr, A. (eds.): MISDOOM 2019. LNCS, vol. 12021. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-39627-5
Tufekci, Z.: Opinion—YouTube, the great radicalizer. The New York Times (2018). https://www.nytimes.com/2018/03/10/opinion/sunday/youtube-politics-radical.html
Stöcker, C., Böhm, M.: YouTube: Werbung vor hetzvideos gegen flüchtlinge. Der Spiegel (2015). https://www.spiegel.de/netzwelt/netzpolitik/youtube-werbung-vor-hetzvideos-gegen-fluechtlinge-a-1058254.html
Wakabayashi, D., Maheshwari, S.: YouTube advertiser exodus highlights perils of online ads. The New York Times (2017)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Stöcker, C., Preuss, M. (2020). Riding the Wave of Misclassification: How We End up with Extreme YouTube Content. In: Meiselwitz, G. (eds) Social Computing and Social Media. Design, Ethics, User Behavior, and Social Network Analysis. HCII 2020. Lecture Notes in Computer Science(), vol 12194. Springer, Cham. https://doi.org/10.1007/978-3-030-49570-1_25
Download citation
DOI: https://doi.org/10.1007/978-3-030-49570-1_25
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-49569-5
Online ISBN: 978-3-030-49570-1
eBook Packages: Computer ScienceComputer Science (R0)