Not just 'For You': How the Algorithmic Crystal mediates communication and identity work on TikTok's FYP

This research highlights how TikTok’s “For You Page” (FYP) algorithm operates as a socially-infused system rather than a purely individualized content recommender. Through interviews and screen-sharing sessions with 27 users, we extend the algorithmic crystal framework to account for a social layer that emerged from our data, and demonstrate how people interpret FYP recommendations as meaningful social signals to communicate aspects of themselves (“this is me”), acknowledge others (“this is you”), and express shared identities (“this is us”).

Our findings also reveal a perceived feedback loop between users’ social behaviors and the algorithm’s subsequent recommendations. Participants described how repeatedly engaging with content connected to their relationships appeared to prompt the FYP to surface more of the same content, creating new opportunities to connect. This dynamic made the feed feel increasingly attuned to their interpersonal lives, reinforcing the sense that the algorithm was not simply reflecting individual preferences but was responsive to, and shaped by, their social interactions.

We argue that these interactions constitute a form of AI-mediated communication (AI-MC), in that the algorithm becomes an active participant in relational meaning-making, shaping how users perceive themselves, understand others, and communicate across their social networks.

Fig 1. An individual’s self-concept includes both personal identities and social identities, such as their social roles and the communities they are a part of. The algorithm is a reflection of users’ self-concept and allows people to recognize themselves (“This is me”) in its recommendations as well as their close ties (“This is you”) and their shared identities (“This is us”).

In our paper, we suggest that these dynamics invite us to rethink how people develop their sense of self in digital spaces, because identity work is increasingly shaped not only by social audiences but also by algorithmic interpretation. To capture this shift, we introduce the concept of a hybrid digital identity, the algorithmically-networked self that combines the relational intent of the networked self (where identity is shaped through interactions with social ties) with the curatorial mechanisms of the algorithmized self (where identity is shaped through interactions with the system).

In doing so, we suggest personalized algorithmic systems do more than reflect user identity; they are used to communicate that identity with others, shaping both the message and the medium of identity expression. Personalized algorithms shape the message by surfacing content that highlights, reinforces, or amplifies particular aspects of a user’s identity, signaling which attributes are most legible or valued within the system. At the same time, algorithms shape the medium by providing a repertoire of recommended formats, sounds, templates, and trends (materials not generated by the user but supplied by the system) that become the structures through which identity is performed and communicated to others.

On the FYP, identity performance, social ties, and algorithmic interpretation converge, producing a dynamic identity practice that is collaboratively constructed by the user, their social ties, and the algorithmic system. This reframing highlights algorithmic personalization as a social, relational, and communicative process, and positions socially-infused algorithms as increasingly central to contemporary notions of digital identity.

For more details and full description of the research context, methods, findings, and discussions, please refer to our full paper  by Zoë Natalia Cullen, Angela Y. Lee, Brenna Davidson, Jeffrey T. Hancock, & Nicole B. Ellison., archived and presented at the 28th ACM SIGCHI Conference on Computer-Supported Cooperative Work & Social Computing (CSCW) held at Bergen, Norway on October 18–22, 2025.

Read the full paper

Read the blog post