Is the Internet Changing Our Perception of Names?
For years, I've talked about how the Internet has affected the process of choosing a baby name. It has helped shift our baseline for assessing popularity and "uniqueness" from internal ("How does this name strike me? Have I met a lot of people with that name?") to external ("How many Google results does this name return?")
A baby name choice, though, is just the starting point of the lifelong name experience. The more we live online, the more the first impressions we make are via our names. And given the reality of online life, that means that search results can shape the impressions we make on strangers...even if the results have nothing to do with us as individuals.
A new study demonstrates one pernicious example of this. Latanya Sweeney, a Harvard Professor of Government and Technology, discovered that web searches for her name frequently yielded advertising results with texts like "Latanya Sweeney, Arrested?" and "Check Latanya Sweeney's Arrests." As Dr. Sweeney had no arrest history, it seemed likely that the advertisers were keying on her name itself -- and perhaps on her distinctly African-American given name.
Dr. Sweeney generated lists of characteristically white and characteristically black first-last name pairs. She then ran searches on these names, and found that the "black" names were significantly more likely to generate ads suggestive of an arrest than the "white" names. The arrest-focused ads appeared regardless of whether the background-search firm presenting the ad actually had any arrest records for that name. (Edited for clarity per communication from Dr. Sweeney.)
It's important to note that this discriminatory effect does not necessarily imply a discriminatory intent on the advertisers' part. It's an unfortunate reality that African-Americans are arrested at a disproportionately high rate. An algorithm based off of actual searches for arrest records, or even name distributions in an arrest record database, could yield such an imbalance in ad presentations. But regardless of intent, the effect could have serious consequences for African-Americans. As Dr. Sweeney writes:
"Perhaps you are in competition for an award, an appointment, a promotion or a new job...perhaps you are completing a rental application, selling goods, applying for a loan, making new friends, dating, or engaged in any one of hundreds of circumstances for which an online searcher seeks to learn more about you. Appearing alongside your list of accomplishments is an advertisement implying you may have a criminal record, whether you actually have one or not. Worse, the ads don't appear for your competitors."
I'd suggest that this advertising result is just the tip of the iceberg of internet name search effects. I believe that global searches for names will tend to reinforce stereotypes, and even create new stereotypes where none previously existed. The reason is that they present us with instant aggregate impressions of people bearing a particular name.
Whereas one-on-one experience with individuals tends to break down stereotypes, an aggregate presentation can tend to encourage them, as the most sensational and negative examples catch our attention. Even a small number of such examples can be perceived as a trend. Worse yet, it's in the nature of web search results, particularly image searches, to overrepresent sensational extremes such as arrest mug shots and pornography. So any name-by-name differences in those arenas will be further exaggerated.
As a demonstration, I decided to look within a name demographic. (Like "male vs. female," "black vs. white" is just a broad starting point in analyzing names. Two names can have similar racial distributions but very different associations in terms of age, culture, socioeconomic status and more.) I ran image searches for 20 names from a "blackest names" list that are borne mostly by young women, paired with each of the five most common surnames in America. I then looked at the top 20 image results for each name pair.
Each set was a mix of imagery: graduation portraits, sports action photos, beauty queens, formal career headshots, teenagers smiling goofily into their cell cameras, highly sexualized photos, mug shots. The mix, though, varied dramatically from name to name. The mug-shot percentage ranged from near zero for names like Nia and Jada, to 20% for names like Tiara and Tierra.
There's nothing objectionable about any of these names. They're all borne by thousands of fine women, and I suspect that few employers, loan officers or rental agents would start off with much prejudgment of a woman named, say, Tierra. Yet if they ran image searches hoping to find a picture of an applicant, could they completely shake the impression made by a mug shot-laden result? Could you?
Below are the top chunks of image results for two of the names I just mentioned, paired with the same common surname.
Imagine typing in the name of a job applicant and getting one of these two results. How would the result affect your feelings toward the résumé in your hands? It's not a purely hypothetical question; this happens countless times every day. More than ever before, the people we share our names with will influence the impression we make on the world...and one bad apple may well harm the whole name bunch.