The Analytics of Hate

I was born color blind.

I truly am. Until I was twelve years old, I thought I was a genius at Rubik’s Cube.

I always get the same silly question, “What does black and white look like to you?”

I always make the joke, “It’s just fifty shades of gray.”

Growing up, I was a white kid surrounded by a mostly African American population in a poor Alabama city in the United States. All my best friends were black. My first girlfriend was black. I worked at a small local radio station and started the first New Jack Swing, hip hop live request show. I went on to go to a primarily African-American university in Atlanta, Georgia to study computer programming.

So often growing up, I was ashamed of being white.

I was mostly ashamed of the atrocities that my friends and their families had endured from people of my skin color. It was magnified when Hollywood filmed “Mississippi Burning” in my hometown. My father’s Piggly Wiggly is just out of frame from every shot of the court house.

When the Rodney King riots broke out across Los Angeles, Chicago, New York and Atlanta in 1992, I was in a car heading to Six Flags Over Georgia for Senior Skip Day to watch a hip hop concert. When we arrived and found that the concert had been cancelled due to security concerns, the white security guard looked at us and asked, “Why are you two together?”

My friend, Eric Bullard and I looked at each other and shrugged. “Because we are brothers.”

In May 2004, I arrived at OR Tambo International Airport in Johannesburg on the same day that Mandela came back triumphant after winning the right to host the World Cup in South Africa. I got to see him alive and at his happiest walking around the airport holding up the World Cup trophy to cheering, singing, and dancing. I went on to live in Cape Town for the next four years.

My girlfriend in South Africa at the time appeared in the movie opposite Samuel Jackson and Juliette Binoche in the film called “In My Country” that depicted the Truth and Reconciliation Commission hearings.

I personally do not distinguish between racism and hate. I believe they are one in the same. And being that I grew up always ashamed of the terrors of my skin’s past, I was always keenly supersensitive with my own actions, words that would make me perceived as racist.

While working in Islamabad, Pakistan for the Minster of Defense and using data science to determine why Al Qaeda was winning the conflict and prior to my Marriott Hotel being bombed, I had to face my own racist stereotypes head on. My co-workers in Pakistan were risking their lives by showing me around, protecting me, bringing me from meeting to meeting. They believed in their country and believed in a better life for their families. My friends and co-workers believed that analytics could change the direction of their country.

In Pakistan, initially I was afraid of the people on the street. But the more time I spent in Pakistan, and I met people, shared stories, laughed, and shared Halal meals, I realized something profound especially when it comes to data science and standup comedy: all people want to belong, all people want to be loved, and most of all everyone wants to laugh,

Which made me start thinking, as a data scientist, what is the pattern to hate?

Slate published an article around an effort last year trying to quantify racism using Big Data. The study was inconclusive. However, it found minor links between racist Google searches and black mortality from 2004 to 2009. According to the researchers, the study suggests that “racism shapes patterns in mortality and generates racial disparities in health.”

However, trying to find direct correlations to racism and hate speech by mining multiple Big Data sources: health records, poverty indices, mortality, prison records, Google searches around racial slurs including those for Asians and other minorities has proved to be not only a sensitive subject but hard to patternize.

But again, if a pattern did emerge, what can we do? How do we stop it? These are learned behaviors. Hate is an emotion that steers actions. And if we find those who fit the pattern aren’t we just stereotyping and not giving them the chance for redemption?

So maybe the analytics to hate is to find the pattern of forgiveness?

In South Africa, the Truth and Reconciliation Commission uncovered the patterns to hate where those who hurt, maimed, murdered, and brutalized faced their victims and asked them to forgive them. Those who purported the crimes faced their victims and asked them for forgiveness and were then given amnesty.

Hate online is not limited to racism. It is also directed toward sexual orientation.

Analytics and data collection around determining hate crimes towards sexual orientation is still in its infancy. But one positive step has been in the State of New York in the United States where it launched a coordinated, multi-agency effort to strengthen data collection regarding lesbian, gay, bisexual, transgender, and queer (LGBTQ) individuals in New York. Why?

LGBTQ individuals have unique health experiences and needs, yet companies, local, and national government agencies aren’t really sure what those needs are. What makes matters a little more difficult — any data collection is on a volunteer basis.

However, with self-reporting, LGBTQ individuals can offer inputs around verbal or emotional stress that directly links to physical violence, harassment, or online bullying.

As the US Presidential Elections are in full swing and other democracies around the world are also in election mode such as the Philippines, hate speech and racism has come center stage.

For example, in the Philippines elections Vice President Jejomar Binay is using his darker skin color to reach out to the provincial voters to make them feel they are being discriminated against in order to impel them to vote.

It’s interesting to overlay the data around hate speech, hashtag bullying to the geographic areas that are voting for candidates that condone such behavior. Fear drives votes. Also, hate drives votes.

If you investigate genocide or any world wide conflict and go back six months prior, you can see the escalation of hate online: YouTube videos, YouTube comments, Tweets, Facebook statues, and Facebook replies. Other factors include inflation and government rhtetoric. Another powerful indictator is to buy tickets to a live standup comedy show in any city and take special note of the jokes around race, hate, or gender inequality. Standup comedy is a real time sample size of it’s population.

For example in Afghanistan where social media wasn’t a factor, ranges of fruit prices in the villages near main roads and highways were the main indicator. When there were spikes in fruit prices, we could predict that a bombing or attack was immeninent because the people in the village might be too afraid to reveal that they know who is about to do an attack but they knew they had to make money quickly to survive through a conflict.

Although its important to note that all online hate is not hate — although it is seen as hurtful. Those that troll are also causing noise. These are people that stir up unnecessary online conflict by writing inflammatory statements — for fun. They do not believe it but they write it anyway to create anarchy.

Trolling causes data noise but also inadvertently causes guilliable viewers or consumers to believe the comments and amplify these messages.

The worry from a data science project is that these hate conversations will flaw machine learning algorithms and natural language processing applications. Because these algorthims are like newborns and have a hard time determining what is right and wrong, what is black and white, they learn only what they are taught.

As humans we can learn and re-learn based on our experiences, art, books, films — and most importantly friendships and relationships.

So just because I am color blind, doesn’t mean I don’t know the difference between black and white.

But I know this, there is no gray when it comes to hate.

GS Jackson, technologist, data scientist and a former standup comedian who tries to simplify the often difficult, befuddling world of Big Data and Data Science.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.