The results? The profiles with white-sounding names received positive responses to their inquiries 50 percent of the time, compared to only 42 percent of the time for profiles with black-sounding names -- a finding that held true even when variables such as age, sex, and property type were taken into consideration. In fact, some hosts were so determined not to host black people that when the researchers later followed up to see whether they were ever able to fill the room, over a third hadn't (losing them an average of $65 to $100 per night). Owners would rather lose money than rent their homes to fake black people.
Meanwhile, the only reliable indicator of whether someone was likely to respond positively to a black respondent was whether or not they'd previously hosted other black people. This finding was later dubbed the world's first scientific proof that, yes, having black friends does mean you're less of a racist.
Although you could argue that racists gonna race regardless, it's Airbnb's system that makes this sort of discrimination possible. Other services, such as Expedia and Priceline, mask the identities of applicants, leaving hosts with the ability to choose guests based only on criteria such as reviews and profile ratings. Someone even made a browser plug-in that allows Airbnb hosts the ability to anonymize profiles and remove any unconscious bias, although it's hard to imagine this taking off among the "racist property tycoon" demographic.