Although Amazon’s same-day service is available to most addresses in Boston and reaches almost to New Hampshire, the centrally located neighborhood of Roxbury, with a population that’s about 59 percent black and 15 percent white, is excluded. The predomin
In same- day cities Amazon hasn’t yet sur- rounded with warehouses, the company must decide which neighborhoods are worth the cost of service and which aren’t. That’s where things get complicated.
Some excluded ZIP codes correspond with higher crime rates. Amazon won’t say whether concerns about stolen packages or the safety of drivers figure into its decisions about where to deliver, saying only “the safety of our employees is a top priority.”
Income inequality may also play a part. Many excluded areas have average household incomes below the national average. And households with Prime memberships skew wealthier—not surprising given the $99 membership fee. An April study of families with teenagers by investment bank Piper Jaffray estimates 70 percent of such U. S. households with incomes of $112,000 per year or more now have a Prime membership, compared with 43 percent for households with incomes of $21,000 to $41,000.
Income differences alone don’t explain the gaps in service, however. In Chicago, New York, Boston, Atlanta, and other cities, some areas that are excluded have household incomes as high or higher than ZIP codes Amazon does cover.
Berman points to cities where some black ZIP codes get same-day service and some white ones don’t. In Los Angeles, black and Hispanic communities south of downtown have same-day service, but mostly white Malibu, on the far side of the traffic-clogged Route 27 and Pacific Coast Highway, doesn’t. Overall, though, in cities where sameday service doesn’t extend to most residents, those left out are disproportionately black. (In the six cities with disparities, Asians, on average, are as likely as whites to live in an area with coverage; Hispanics are less likely than whites to live in same-day ZIP codes, but more likely than blacks.)
“As soon as you try to represent something as complex as a neighborhood with a spreadsheet based on a few variables, you’ve made some generalizations and assumptions that may not be true, and they may not affect all people equally,” says Sorelle Friedler, a computer science professor at Haverford College who studies data bias. “There is so much systemic bias with respect to race. If you aren’t purposefully trying to