Lifestream, Tweet: reply to @Digeded

Note: This post is more connected to our third block, algorithmic cultures.

Eli Pariser’s (2011) ‘filter bubble’ – or, the way that algorithms edit the content we are exposed to online, connecting us with similar views and exposing us to advertising news we are likely to be interested in – came back into the spotlight in the build-up to the 2016 US presidential elections*. However, danah boyd (in the article tweeted by Matthew and also taken up in various other forms, such as boyd’s blog and Data Society:Points) makes the point that people’s personal choices have a significant role in creating the silos for which personalisation algorithms are frequently blamed. 

Why does it matter? While boyd notes that more diverse teams are known to outperform more homogeneous teams, at the crux of her articles is a concern for democratic process: ‘If we want to develop a healthy democracy, we need a diverse and highly connected social fabric.’ For me, this links to Granovetter’s (1973) theory of strong and weak ties, which suggests that people within more diverse social networks (with weak ties between people of diverse backgrounds) are better able to read the opinion climate and are more likely to express opinions they believe to be in the minority.** In turn, a diverse network within which people speak out despite having minority-opinions helps to avoid what Sheehan (2015) termed a “spiral of silence”. Within a spiral of silence some opinions become artificially dominant because of the unwillingness of minority-view holders to speak out, through fear of isolation – hence diversity is necessary.

Recently (January 12, 2017), Jenny Mackness offered me two other ways of understanding the necessity of diversity, in the blog post to which my reply to Matthew links. Firstly, rather than a spiral of silence, Mackness refers to “information cascade” or “cascade phenomena” (Downes, 2005, 2007) in which external information overrides individual, private information signals despite possible contradictions between the two. This phenomena offers further insight into the artificial dominance that some opinions and acts gain, regardless of apparent lack of truth or lack of fit with proclaimed values. Secondly, Mackness raises Cillier’s (2010) proposal that the world is aptly viewed as a complex adaptive system, in which meaning is generated by difference, and meaning is distorted by a reduction in diversity. Both of these ideas will, I feel, enrich my understanding of the impact of algorithmic cultures in block 3 of #mscedc.

*for a demonstration of how conservative and liberal facebook feeds tend to differ, see , based on a study by Bakshy, Messing & Adamic (2015).

**I previously wrote (briefly) about weak and strong ties within the context of supporting open networked learning not just for its value for individuals but also society  here.  The paper as a whole is on vulnerability within socially networked learning, and the emergence of new student identity roles.