Internet users tend to navigate between websites in a racially segregated way despite pathways that provide equitable access to different sites, finds a new study.
The findings show that variables that have historically contributed to racial inequality offline -- such as segregated traffic patterns and destinations -- are present within the web's environment.
"We know that people do racist things on and using the internet, but looking beyond individual, interpersonal accounts of bigotry, how does systemic racial inequality form in the digital world?" asked Charlton McIlwain, Associate Professor of Media, Culture, and Communication at the New York University.
McIlwain designed a study, specifically looking at how users navigate the web's structure and how that structure influences users' navigational patterns.
McIlwain then created the architecture of the actual traffic patterns among and between racial and non-racial sites using a programme that employs a spatial algorithm to compare links between sites.
He found that web producers create hyperlink networks that do not steer audience traffic to other sites based on their racial or non-racial nature.
However, the opposite pattern emerged when looking at users going to and coming from sites in the network.
The team noticed that user navigation reflects a racially segregated traffic pattern, where visitors to non-racial sites visit other non-racial sites with greater frequency than what would be expected by chance, and visitors to racial sites visit other racial sites more than expected.
"The evidence suggests a tendency toward racially segregated site navigation. Web producers seem to build pathways providing equitable access to sites, without concern for the racial nature of the site," said McIlwain in a paper published online in the journal Information, Communication, and Society.
The results show how a race-based hierarchy might systematically emerge on the web in ways that exemplify disparate forms of value, influence and power that exist within the web environment.