Update: According to the article danah boyd sent out, Twitter says:
We select each tweet using a variety of signals, including how popular it is and how people in your network are interacting with it.
As Jeff Nichols (@jwnichls on Twitter) pointed out, this can be a positive thing, and for some is what they use Twitter for. This strikes me as not anti-homophily, the way danah boyd suggests, but instead what Recommender Systems people call “serendipity”. It seems it can’t be anti-homphily if this is partially based on people in your network’s interactions. I’m not certain this changes the higher-level point of the original posting, but it does add some context. I’ve modified the original posting to reflect this context.
Twitter has been inserting tweets into other people’s timelines, based on the favoriting patterns of those tweets and how your own social network is interacting with these tweets. That is, it’s possible for me to favorite a tweet, and because I’ve done so, it will appear in others’ timelines.
I'm not surprised that people don't like being exposed to random tweets,
but I see this as an anti-homophily move:
— danah boyd (@zephoria)
What would it take for people to actually get excited by being exposed
to perspectives from strangers?
— danah boyd (@zephoria)
I find this argument very intriguing, because I have two orthogonal reactions to this idea. I am going break these ideas into sections, because I think they deserve their own treatment.
The role of the platform ought to be neutral
My first reaction to this concept was approximately “why does Twitter get to decide that I need less homophily in my timeline?”. I think this fair from an autonomy perspective, don’t people deserve to choose what they view? There’s a number of ways Twitter could also use this mechanism that isn’t a “pure” anti-homophily effort, from advertising to a curated “Twitter” experience designed to keep you engaged on their site. Under this framing, inserting others’ tweets into my timeline struck me as invasive, and the opposite of the reason I use Twitter. I like the process of choosing what shows in my timeline. I get to choose to be homophilic or not.
The platform ought to recognize the power that it has, and pursue “social good”.
However, the recent events in Ferguson are a stark reminder of the socialized, institutionalized racism that still occurs in the United States. How does society combat this? There are structural approaches, like oversight of police forces. However, there are a number of underlying mechanisms (historical, or legal structures) at play that lead to this sort of racialized response (let alone the shooting of Michael Brown that sparked these protests). One of these mechanisms has been said to be social distance, which leads to othering.
Twitter, as a medium, played a role in dissemination of information on the ground, discussion of race issues, and a more timely exposure to these events than other news media. It also served as a platform for people saying very bigoted, racist things, while supporting the actions of the police. Homophily allowed people to choose which side of this issue they were exposed to through Twitter. Anti-homophily could have served to expose those who were ambivalent, or supporting the police action, to an alternative view of these events (there are questions of safety and harassment that would need to be addressed, certainly).
As the platform owners, Twitter is in a unique position to make these kinds of decisions. Is being “exposed to perspectives from strangers” a positive thing? Conceivably, a move by Twitter to counteract homophily could serve to decrease social distance and reducing the othering effects.
This is an example of what anti-homophily could bring to bear on the Twitter platform, and danah boyd certainly seems to believe that non-homophily is a good thing (at least if people are excited by it on their own).
If she is right, and Twitter is using favorites as a way to counteract homophily on their platform, what obligations do they have? What is Twitter’s purpose behind anti-homophily measures? How do builders of sociotechnical systems choose the attributes that support “social good” (for whatever that means in a given context)?
If Twitter can make the argument that favorites-as-anti-homophily does decrease social distance among their users, or perhaps even supports users choosing to counteract homophily on their own, I find this argument compelling. It situates platforms as “social change” shephards, or perhaps in extreme cases, digital community organizers. Is this a role platform owners identify as one they’re taking on? How does one separate the algorithmic change imposed on their users, and the obligation to monitor and curate this change, even if the change is for business purposes?