Big Data Bra Size [...]

Big Data is seen as the End of Theory by some, and as the dawn of a humanistic personalization by others. Which is why we find this a useful example of Big Data: Alibaba, the largest online store in the world (centered in China) has correlated bra size to spending habits. After the quote we explore the weird intersection of Big Data, sexism, racism, and any other ism’s you might want to fold in.

Alibaba vice president Joseph Tsai talked to Quartz about findings that 65% of women with a B cup fell into the “low” spending category, while those with a C cup or higher were in the “middle” and “high” demographics.

“We’ve only seen the tip of the iceberg,” he said of the company’s data-dive. “We really haven’t done even 5% of leveraging that data to really make our operations more efficient, consumers more satisfied.” (Source)

What do we make of this? It makes us feel (hopefully) a bit icky. Why is that? It’s probably worth exploring.

It’s also interesting that if why knew WHY this correlation applied we’d feel (I think) a bit less icky. Is age a confounding factor (older women have bigger bra sizes, teenagers have smaller bra sizes)? Is cup size a sign of affluence in China? A proxy for city dwelling? Leaving the correlation at this basic level feels unfinished, dirty, and exploitative.

Suppose we extend special deals to C-cup women. (and certainly this is happening — otherwise why collect this data?). Is this wrong? What if such things break down by race?

When we look at this we realize that a world without theory is also a world that discards intent. And intent means something. Big Data Doesn’t Care why your coupon works, but you do. There is a big difference between offering sweet deals to large-breasted women and offering sweet deals to people in their twenties or to more frequent shoppers. Somewhere along the line we are going to have to come to terms with that.


For a more serious case of why the why matters, consider Big Data and OxyContin

Apart from issues of why, Big Data’s correlations are often ephemeral: Google Flu Trends succeeded until it failed.