TikTok: Algorithms & bias

Joseph Reagle

Our question(s)

How do platform algorithms affect (marginalized) communities?

Overlooking Black creators

Renegade: Harmon; D’amelio & Rae

ziggitylyer’s frustrations

ziggityler #greenscreenvideo I’m going live in 30 minutes to answer questions. Y’all need to get this message out. Please. #fypシ #fyp #wrong #justice ♬ original sound - Ziggi

Issues

  • biased content warning (e.g., “white supremacy” is accepted “black lives matters” prompts a warning)
  • “for you” algorithm privileges white creators and appropriation (Mitchell 2021)
  • posts being taken down, not able to go live, or shadow banned

What is cultural, what is technical, what is purposeful?

Algorithmic bias

Bias and fairness

For you

Simpson and Semaan (2021)

  • goal?

    • how does the algorithm “construct contradictory identity spaces that at once support LGBTQ+ identitywork and reaffirm LGBTQ+ identity, while also transgressing and violating individual user identity intersections”

  • method?

    • 16 interviews

Findings?

algorithmic exclusion … construct and reconstruct exclusionary structures within a bounded sociotechnical system, or more broadly across societal structures. (SimpsonSemaan 2021, p. 24)

How?

  • “by removing or not sharing their content”
  • “by showing unwelcome stereotypes or personalities” (e.g., “ending up on wrong side of TikTok”)

“Ontological security”

In [Giddens’] view, the ability to both routinely enact and assume an identity while also seeing aspects of one’s identity visually represented in others provides individuals a sense of security about their existence. (SimpsonSemaan 2021, p. 3)

  • Ought people be separated for a sense of safety?
  • Does this then lead to echo chambers… or trite stereotypes?

Being seen

made our participants feel as though their identities are not as worthy of being seen by others within the community they imagine themselves a part of on TikTok. (SimpsonSemaan 2021, p. 21)

“Resistance strategies”?

Subcommunities on…

  • Reddit

    • subreddits; problems of “brigading”

  • Mastodon

    • federation and block mechanisms

  • Twitter

    • (weakly) hashtags; problems of “tag-crashing”

  • BlueSky

    • hashtags; custom feeds; shared mod lists

QICs

Other questions, insights, or connections?

Conclusion

Wrap up

ex. Design alternative??

Propose 5 (bullet-point) recommendations for systems and algorithms that support healthy and diverse online communities?

. . .

Review

What were your recommendations for healthy and diverse online communities?