Have you ever looked something up on Wikipedia or Google and failed to find much relevant or high-quality content? Every time that happens, it’s a subtle hint to us that the topic isn’t important or interesting. However, this implicit message is often biased, especially against diversity and in favor of the dominant culture and its language and trends.
A clever study (Hinnosaar, Hinnosaar, Kummer, & Slivko, 2019) used a controlled experiment on Wikipedia to estimate how much of an effect this has on a real-world outcome. They randomly assigned Spanish cities with low-quality articles on Italian, German, and French Wikipedias, and added relevant information (mostly by translating from Spanish and English Wikipedias) to 120 treatment group articles while leaving the 120 control group articles unchanged. They estimate that hotel stays by Italian, German, and French tourists are increased by an average of 9% in cities that had their articles improved in the language of the tourist’s country of origin compared to control cities.
The authors consider the paper’s implications from an economic perspective, discussing the failure of interested parties to respond to the apparent economic incentive to provide the missing information. But the paper also serves as a reminder that informational bias has real impacts. Given the prevalence of male editors on Wikipedia, the commercial interests of Google, and the lower representation of minority cultures and languages online generally, it’s no surprise that low-yield searches disproportionately affect people who are interested in minority or foreign cultures or notable women, BIPOC and/or LGBTQ figures, and those who seek information in non-English languages or in English about non-English-dominant topics. Regardless of their intent, whenever content producers neglect a topic, and whenever algorithms privilege popular topics at the expense of others, they risk exerting a substantial marginalizing effect.