Surface Level
To a person behind the screen at a company coding algorithms, to the massive computers filled with user data, and to advertisers paying to push their products onto social media platforms, I am nothing more than what information they can gather from me and what I reveal about myself online. This is the information that gets collected, stored, and used to create a computer-based model for me that can predict my behaviors and my personality more accurately than humans can (Youyou et al. 2015). As an avid social media user, especially Instagram, it is absolutely certain that with the eight hours I spend on that app per week, my computer-based model becomes more accurate and nuanced to who I am as a person. But how much of me do they know?
I have three Instagram accounts, each account catering towards a different group of people. My main Instagram account is what I want people to see about myself–my followers include high school friends, college friends, acquaintances, and people I spoke with once or twice and never again. I have a spam account for my close friends where I “photodump” my silly photos or repost funny posts that I see on my feed. And finally, I have an art account where I post my drawings under an online alias. I remain completely anonymous to my followers and only post when I have new completed art. Together, these three online identities contribute to my computer-based model, about who I am to these companies, brands, and developers.
In signing up for my Instagram accounts, I provided basic information about myself to the app. My name is Alicia Zhang, my birthday is February 2nd, 2003. That makes me nineteen years old. Through the people I follow and the photos I’m tagged in on my main account, they know that I lived in Neenah, Wisconsin my whole life, and attended Neenah High School. In my own posts where I tag my own location, the app data also knows that I live in the Bay Area in California, and one glance at my bio shows that I am a student at UC Berkeley, graduating in 2025. Living in the Midwest, I primarily grew up with caucasians–an immediate look at my profile picture shows that I’m chinese. Taking a glance at the list of people I follow, the algorithms can deduce even more information about me. Besides my friends and acquaintances, I follow influencers, my favorite bands and singers, actresses, artists, as well as science and Christian pages. My hobbies include art of all kinds, including digital art, painting, embroidery, and music, and recently I have been interested in ‘broken humor’ posts, skincare, and jewelry (Appendix Fig 1).
The Instagram “explore” tab is how the app targets me as a user to show me content they think I would enjoy. In a video made by Adam Mosseri, the head of Instagram, he explains that the algorithm looks at information about a certain post and your interactions with the poster, as well as your activity and the content you interact with. The algorithm then creates a guess on your interest levels based on how likely you are to like, save, and share the post. Based on these guesses, the app arranges various posts into the explore feed (Adam Mosseri). Across all three of my accounts, my explore page matches with the type of content I interact most with on each account. My main account’s explore page shows various Christianity posts, viral reels from Asian creators, fan posts about musicians I enjoy, artwork, and a variety of funny posts. On my spam account, I save a lot of posts about ear piercings and skincare and follow humor accounts, so my explore page is entirely filled with photos of earrings, milky skincare routines, and reposts of reddit posts and tweets. My art account only interacts with art-related posts, so my explore page is arts and craft content (Appendix Fig. 1). Even the Instagram shop tabs across all accounts are curated to my account’s interests, from jewelry to clothing to sticker art. These examples all demonstrate who I am to an algorithm and as a user of a platform.
The thing about social media is that, while an algorithm is capable of predicting my interests, they cannot pick up on the subtleties of myself and the deeper things I value. These parts of me are not aspects that can be marketed to, and have so much more meaning and impact to me than the surface-level demographics that apps such as Instagram see daily. I am Alicia Zhang–a student, scientist, musician, artist. As a fairly mellow individual, it’s the little things that make me happy: the sun coming out after a cloudy morning, seeing stray cats on the street, when my Spotify shuffle plays my favorite songs back-to-back. I am an introvert and a homebody, meaning meeting new people is always difficult for me since I dislike small talk. But in a crowd of my favorite people, I’m one of the loudest.
It is difficult to get to know me on a very personal level, but for the few people that do, I will do anything for them. [Close here]
Beneath my composed appearance are the deep thoughts and experiences that shape me to be who I am. I value achievement, belonging, dependability, efficiency, and happiness the most. Being surrounded by people who accept me and whom I can depend on is what contributes to my happiness the most. And my value for success comes packaged with my enneagram–a three. While my strive for excellence has certainly paid off in terms of receiving recognition (screenshot of NHS news?), the trade-off is the immense pressure I put on myself. Beginning in middle school, I worked hard to achieve good grades and now that I’m in college, that pressure to do well has evolved into a fear of failure, but I also have changed to be mentally well-prepared to handle that pressure as well. Because I am not an emotional person and do not get overwhelmed easily, it allows me to think clearly and logically about my problems so that I can find a solution or adapt to the issue. Not every decision gets made with clear logic of course, thanks to my cognitive biases. In the list of cognitive biases that neuropsychiatrist Dilip V. Jeste includes in his article “Wiser”, the ones I relate most to are loss aversion and escalation of commitment, both of which play into my value of efficiency. Loss aversion is when you perceive loss as more severe than gains, and escalation commitment is the tendency to stick with a decision because you have already put too much work into something (Jeste 2020). In order to make the most of my time, I try to avoid failure as much as I can (loss aversion), since that means I have to start over and try again. Likewise, and sometimes contradictory to loss aversion, if I have invested time into a project, even if it may be on its way to failure, I continue working and attempt to ‘save’ the project because it feels too late to start over.
Aside from struggles connected to schoolwork or decision-making, I also wrestled with my identity as a Chinese American. According to the US 2020 Census for Wisconsin, Asians make up 6% of the population, while the percentage of white people is 61.6%. Within my county, Winnebago county, the percentage of Asians drops even further to 3.2%, and the percentage of white people increases to 84.8% (America Counts Staff 2021). Because I grew up with all my peers at school being caucasian, I felt ashamed to be Chinese. I was embarrassed by my last name, my mom’s heavy Chinese accent, and the leftovers I brought to school for cold lunch. Even though I grew up also attending a Chinese church with other Chinese kids my age, I felt like I was one with the white kids at school–American Chinese, not Chinese American (Appendix Fig. 3). It was not until high school when I made friends that were proud to be asian, that I finally began to embrace this integral part of my identity.
Something that I discovered as I unravel these two parts of myself–what algorithms see versus what I see–is that targeted products and social media are more geared, at least from my experience, towards things and items that I’m interested in which more or less are not affected by my demographic. The products and posts I see on social media are affected by who I follow and what content I interact with, but as I have done in the past, it is really easy to change what content I see, especially on Instagram. My interests change fairly rapidly, and as such, my Instagram explore has shifted to match those interests. Between my phases of obsession with bullet journaling, video games, nail art, friendship bracelets, and binging shows, my explore page at each interest looks like it could belong to somebody else. I find that even though Instagram has tracked all my basic information, especially location, race, beliefs, etc, the only targeted content I see is based on what I interact with the most and the accounts I follow, which typically just reflects my interests at the time. This simply demonstrates how easy it is to mislead the algorithm, and is a reminder that even though technology and machine learning is becoming more advanced, we can still outsmart [...]
I have three Instagram accounts, each account catering towards a different group of people. My main Instagram account is what I want people to see about myself–my followers include high school friends, college friends, acquaintances, and people I spoke with once or twice and never again. I have a spam account for my close friends where I “photodump” my silly photos or repost funny posts that I see on my feed. And finally, I have an art account where I post my drawings under an online alias. I remain completely anonymous to my followers and only post when I have new completed art. Together, these three online identities contribute to my computer-based model, about who I am to these companies, brands, and developers.
In signing up for my Instagram accounts, I provided basic information about myself to the app. My name is Alicia Zhang, my birthday is February 2nd, 2003. That makes me nineteen years old. Through the people I follow and the photos I’m tagged in on my main account, they know that I lived in Neenah, Wisconsin my whole life, and attended Neenah High School. In my own posts where I tag my own location, the app data also knows that I live in the Bay Area in California, and one glance at my bio shows that I am a student at UC Berkeley, graduating in 2025. Living in the Midwest, I primarily grew up with caucasians–an immediate look at my profile picture shows that I’m chinese. Taking a glance at the list of people I follow, the algorithms can deduce even more information about me. Besides my friends and acquaintances, I follow influencers, my favorite bands and singers, actresses, artists, as well as science and Christian pages. My hobbies include art of all kinds, including digital art, painting, embroidery, and music, and recently I have been interested in ‘broken humor’ posts, skincare, and jewelry (Appendix Fig 1).
The Instagram “explore” tab is how the app targets me as a user to show me content they think I would enjoy. In a video made by Adam Mosseri, the head of Instagram, he explains that the algorithm looks at information about a certain post and your interactions with the poster, as well as your activity and the content you interact with. The algorithm then creates a guess on your interest levels based on how likely you are to like, save, and share the post. Based on these guesses, the app arranges various posts into the explore feed (Adam Mosseri). Across all three of my accounts, my explore page matches with the type of content I interact most with on each account. My main account’s explore page shows various Christianity posts, viral reels from Asian creators, fan posts about musicians I enjoy, artwork, and a variety of funny posts. On my spam account, I save a lot of posts about ear piercings and skincare and follow humor accounts, so my explore page is entirely filled with photos of earrings, milky skincare routines, and reposts of reddit posts and tweets. My art account only interacts with art-related posts, so my explore page is arts and craft content (Appendix Fig. 1). Even the Instagram shop tabs across all accounts are curated to my account’s interests, from jewelry to clothing to sticker art. These examples all demonstrate who I am to an algorithm and as a user of a platform.
The thing about social media is that, while an algorithm is capable of predicting my interests, they cannot pick up on the subtleties of myself and the deeper things I value. These parts of me are not aspects that can be marketed to, and have so much more meaning and impact to me than the surface-level demographics that apps such as Instagram see daily. I am Alicia Zhang–a student, scientist, musician, artist. As a fairly mellow individual, it’s the little things that make me happy: the sun coming out after a cloudy morning, seeing stray cats on the street, when my Spotify shuffle plays my favorite songs back-to-back. I am an introvert and a homebody, meaning meeting new people is always difficult for me since I dislike small talk. But in a crowd of my favorite people, I’m one of the loudest.
It is difficult to get to know me on a very personal level, but for the few people that do, I will do anything for them. [Close here]
Beneath my composed appearance are the deep thoughts and experiences that shape me to be who I am. I value achievement, belonging, dependability, efficiency, and happiness the most. Being surrounded by people who accept me and whom I can depend on is what contributes to my happiness the most. And my value for success comes packaged with my enneagram–a three. While my strive for excellence has certainly paid off in terms of receiving recognition (screenshot of NHS news?), the trade-off is the immense pressure I put on myself. Beginning in middle school, I worked hard to achieve good grades and now that I’m in college, that pressure to do well has evolved into a fear of failure, but I also have changed to be mentally well-prepared to handle that pressure as well. Because I am not an emotional person and do not get overwhelmed easily, it allows me to think clearly and logically about my problems so that I can find a solution or adapt to the issue. Not every decision gets made with clear logic of course, thanks to my cognitive biases. In the list of cognitive biases that neuropsychiatrist Dilip V. Jeste includes in his article “Wiser”, the ones I relate most to are loss aversion and escalation of commitment, both of which play into my value of efficiency. Loss aversion is when you perceive loss as more severe than gains, and escalation commitment is the tendency to stick with a decision because you have already put too much work into something (Jeste 2020). In order to make the most of my time, I try to avoid failure as much as I can (loss aversion), since that means I have to start over and try again. Likewise, and sometimes contradictory to loss aversion, if I have invested time into a project, even if it may be on its way to failure, I continue working and attempt to ‘save’ the project because it feels too late to start over.
Aside from struggles connected to schoolwork or decision-making, I also wrestled with my identity as a Chinese American. According to the US 2020 Census for Wisconsin, Asians make up 6% of the population, while the percentage of white people is 61.6%. Within my county, Winnebago county, the percentage of Asians drops even further to 3.2%, and the percentage of white people increases to 84.8% (America Counts Staff 2021). Because I grew up with all my peers at school being caucasian, I felt ashamed to be Chinese. I was embarrassed by my last name, my mom’s heavy Chinese accent, and the leftovers I brought to school for cold lunch. Even though I grew up also attending a Chinese church with other Chinese kids my age, I felt like I was one with the white kids at school–American Chinese, not Chinese American (Appendix Fig. 3). It was not until high school when I made friends that were proud to be asian, that I finally began to embrace this integral part of my identity.
Something that I discovered as I unravel these two parts of myself–what algorithms see versus what I see–is that targeted products and social media are more geared, at least from my experience, towards things and items that I’m interested in which more or less are not affected by my demographic. The products and posts I see on social media are affected by who I follow and what content I interact with, but as I have done in the past, it is really easy to change what content I see, especially on Instagram. My interests change fairly rapidly, and as such, my Instagram explore has shifted to match those interests. Between my phases of obsession with bullet journaling, video games, nail art, friendship bracelets, and binging shows, my explore page at each interest looks like it could belong to somebody else. I find that even though Instagram has tracked all my basic information, especially location, race, beliefs, etc, the only targeted content I see is based on what I interact with the most and the accounts I follow, which typically just reflects my interests at the time. This simply demonstrates how easy it is to mislead the algorithm, and is a reminder that even though technology and machine learning is becoming more advanced, we can still outsmart [...]
Works Cited
Adam Mosseri [@mosseri]. How the “Algorithm” Works. Instagram, 23 Jun. 2021, https://www.instagram.com/p/CQdxvdNJ_sC/?utm_source=ig_embed&utm_campaign=embed_video_watch_again
America Counts Staff. "Wisconsin Population Increased 3.6% Since 2010." https://www.census.gov/library/, 25 Aug. 2021, www.census.gov/library/stories/state-by-state/wisconsin-population-change-between-census-decade.html#:~:text=Race%20and%20ethnicity%20(White%20alone,or%20More%20Races%2010.2%25).
Jeste, [Excerpts on] Cognitive Biases in Wiser, 2020
Youyou, Wu, and Michal Kosinski. "Computer-based personality judgments are more accurate than those made by humans." Psychological and Cognitive Sciences, vol. 112, no. 4, 12 Jan. 2015, PNAS. www.pnas.org/doi/full/10.1073/pnas.1418o680112.
Zhang, Alicia. Various photographs.
America Counts Staff. "Wisconsin Population Increased 3.6% Since 2010." https://www.census.gov/library/, 25 Aug. 2021, www.census.gov/library/stories/state-by-state/wisconsin-population-change-between-census-decade.html#:~:text=Race%20and%20ethnicity%20(White%20alone,or%20More%20Races%2010.2%25).
Jeste, [Excerpts on] Cognitive Biases in Wiser, 2020
Youyou, Wu, and Michal Kosinski. "Computer-based personality judgments are more accurate than those made by humans." Psychological and Cognitive Sciences, vol. 112, no. 4, 12 Jan. 2015, PNAS. www.pnas.org/doi/full/10.1073/pnas.1418o680112.
Zhang, Alicia. Various photographs.