Blackbird in the Coal Mine — Are Technology Platforms Stifling Black Community? (And Others Too?) — Part 1

On August 30, 2018 the New York Times ran a piece by Farhad Manjoo entitled, “Here’s the Conversation We Really Need to Have About Bias at Google” which explained why Trump’s unsupported claims about search engine bias against conservative voices potentially undermines the conversation various communities and experts have been trying to have for a very long time about the hidden, pervasive and often unintended bias of the search engine results. The 60 Minutes piece “How Did Google Get So Big?” which aired on September 23, 2018, includes discussion of the harmful impacts of the bias of search engines. Even while Google and others strive to modify their algorithms to produce results that are free from the hidden or unintended bias and also more capable of producing results individually relevant, I would suggest there is still the need to create communities of content in ways that don’t silo us as people, but provide an additive overlay.

Mr. Manjoo’s recent article caused me to reflect on how algorithms have become our collective ‘mainstream,’ challenging minority groups seeking to ensure their voices are heard. Properly understood, we are all a part of a minority group or a part of various affinity groups, so it is my belief this problem is relevant to everyone. Ideally, voices of diverse communities and affinity groups will breakthrough to be heard alongside all voices in the mainstream discourse, however, it is also valuable to have community-based platforms where members of those communities can reliably have access to these voices. It serves as a check on the system and is also the basis of shared community. The role back of the FCC’s Net Neutrality rules and the way in which the operators modify their platforms and business models, will also impact the degree to which diverse voices can be located and heard.

A little more than a decade ago, I asked myself the question of where one would find Black or African American (take your pick) content online if you were starting with no assumptions about trusted sources of content. I started plugging into search engine topics, such as “barbershop” to see if they would take you to discussions of the topic from a Black perspective or pull up any Black barbershops. While the search results generated were relevant to a mainstream version of the word “barbershop,” I was interested to see if any of the results were from Black authors, Black media outlets or Black bloggers (and yes, there were a lot of bloggers at the time). Content from African American sources did not appear in the search results until the 8th or 9th pages of the results, nor were there any Black barbershops listed. You had to go through roughly 100 results to get to the Black source. While that might not seem shocking to someone who is not looking for Black sources, it was profoundly troubling for those seeking Black sources and to those Black sources trying to find their audience. The results were the same across all search engines. We used this as an incentive to see if we could do better by designing tools that would help match people in specific communities with content they wanted.

Even if we started with search topics that seemed generic, like politicians (at the time I used Bill Clinton, George Bush, and Hillary Clinton) the search results were from major news outlets, Wikipedia, and a handful of mainstream blogs. Content from Black magazines, authors or blogs were way down on the list, and I feared that traffic played a big part in the algorithm. The question then became how could you get traffic to validate your content, if you were so far down in the search results that no one ever found your site. We even tried entertainers, and specifically Black entertainers, but still the same result. The search was less of a relevancy search, rather, at its core, we believed it was a popularity search. To crack the results and make it to the first page of results certainly could seem impossible to outside groups or content sources that catered to a minority audience.

This was obviously very different than the broadcast model, where content is on a channel and every channel essentially has the same chance of being selected, even if it has a higher channel number, and content is grouped according to a particular demographic it is trying to reach. With entertainment and content moving away from the broadcast model, how could we create tools to allow people to find Black bloggers and content. It wasn’t as if people were not creating content or there weren’t Black content platforms, like Radio One or Essence. Were there tools that could work that could allow content to thrive in other than the old “portal” model?

It is also worth pointing out that shortly after this time period many people started to notice that Google was producing some surprising results when keyword phrases “Black Girls” or “Black People” were entered, the first producing pornographic results and the second producing images of Gorillas, for which Google apologized. Mr. Manjoo points this out in his NYT articles and also cites “Algorithms of Oppression: How Search Engines Enforce Racism,” by Safia U. Noble, as a thoughtful examination of how humans play a role in defining how algorithms, including search engines are constructed. Our robots are only as smart and unbiased as the inputs we provide them, so we need to pay attention when creating and relying upon Artificial Intelligence based solutions to human problems.

A small group of us created a custom Google search engine and an internet browser from an open source kernel and started Blackbird. The theory was simple: create a set of tools that would function as a way to access Black content and help foster or create Black community. But, the idea was always that the tools could be used to access any “affinity” content — i.e. content potentially not in the mainstream, and create any community built around an identity or set of shared interests. It was the idea that there could be another way to experience the Internet, a lens that expanded and refined the viewer landscape. We added a news feed, comments, and an ecommerce landing page. It was clunky and had bugs, but it worked. The first couple of months of operation we saw 300,000 download the browser and use the services. Surprisingly, 40% of our users were from Brazil, where the social empowerment of a new generation of Afro-Brazilian users is taking place.

This initial success produced a surprising reaction that was emotional, political and largely divided along racial lines — essentially giving a glimpse into the underlying societal upheaval we are now openly realizing in the US and globally, as well. Many groups in America have an unquestioned presence (offline and online), for example, groups formed around a common political, ethnic, or geographical experience — but, we found that when the basis of an online community is focused on a Black cultural experience you run right into America’s racial divide. Part 2 explains how Blackbird reacted to an onslaught of criticism and found new opportunities to support Black community.