Living in an Algorithmic World

Two recent conferences, re:publica 18 in Berlin (May 2-4) and Theorising the Web 2018 in New York (April 27th/28th) have featured the influence of algorithms on today’s world.

This week ‘How an algorithmic world can be undermined’ was the title of danah boyd’s opening keynote for the re:publica 18 conference.

Algorithmic technologies that rely on data don’t necessarily support a social world that many of us want to live in. We must grapple with the biases embedded in and manipulation of these systems, particularly when so many parts of society are dependent on sociotechnical systems.

 

Over the course of an hour danah boyd covered:

  1. Agenda settings – the safety of the internet and how this can be manipulated by online groups.
  2. Algorithmic influence – most systems are shaped by algorithms in the belief that algorithms are the solution to everything. Boyd asks how we can challenge this and how these systems can be made accountable. How are these systems manipulated at scale (e.g. political campaigns)? She says we are starting to see a whole new ecosystem unfold.
  3. Manipulating the media – there are plenty of examples of how this can be done to gain attention and amplify messages. Who is to blame for this? Twitter, journalists, news organisations, Wikipedia, reporters? We need to think about the moral responsibility of being an amplifier. What is it? She asks what does strategic silence look like and says – if you can’t be strategic be silent. The process of amplification can cause harm, e.g. reporting on suicide can increase suicide numbers.
  4. Epistemological warfare – how doubt is fed into the system about how we produce knowledge. This is destabilising knowledge in a systematic way, creating false equivalencies that media will pick up on. “We are not living through a crisis of what’s true. We’re living through a crisis of how we know whether something is true” (Cory Doctorow).
  5. Bias everywhere. What biases are built into our systems and how are they amplified? Bias is everywhere and in algorithms. Society’s prejudices are built into the system. Machine learning systems are set up to discriminate, to segment information and create clusters. They are laden with prejudice and bias.
  6. Hateful red pills. Gaming problems and data voids. Red pills are meant to entice you into something more – radicalising people. Where does this fit into broader sets of contexts?
  7. The more power the technical system has, the more that people are intent on abusing systems. We have to try and understand the dynamics of power and the alignments of context, particularly in relation to human judgement.
  8. The new bureaucracy. How do we think about accountability and responsibility? Who is setting the agendas and under what terms? There is growing concern about the power of tech companies and power dynamics in society. It is not simply about platforms. Algorithms are an extension of what we understand bureaucracy to be. Regulating bureaucracy has been difficult throughout history. It is not necessarily the intentions but the structure and configurations that cause the pain. Bureaucracy can be mundanely awful to horribly abusive. Algorithmic systems are similar, introducing a wide range of challenges. Technology is not the cause of fear and insecurity. It is the amplifier. It is a tool for both good and evil. We are seeing a new form of social vulnerability. It comes back to questions of power. Regulating the companies alone, isn’t going to get us what we want.

Similar themes were covered in the final keynote of the Theorising the Web conference

#K4 GOD VIEW – https://livestream.com/internetsociety/ttw18/videos/174107565

This keynote took the form of a panel discussion between John Cheney Lippold, Kate Crawford, Ingrid Burrington, Navneet Alang and Kade Crockford, and moderated by Ayesha A. Siddiqi. It was a fascinating discussion, interesting not only for the content, but also for the format and the lack of point scoring between panel members. Mariana Funes describes this well in her notes on hypothes.is where she writes “This [] felt like a chat after dinner, exploring the implications of the spread of AI systems ……”

The underlying message and final question from both keynotes was: What kind of a world do we want to live in?

Diversity is hard

complexity

Source of image

dana boyd has written a post in which she discusses why America is self-segregating and she comes up with a few suggestions such as the role of social media in segregating people into filter bubbles and echo chambers. But a key point she makes is that diversity, which is ‘often touted as highly desirable’ is hard – ‘uncomfortable, emotionally exhausting and downright frustrating’. So instead of using the many online tools we now have at our disposal to become diversely connected, we use them instead to find like-minded people who, as Kirschner wrote in 2015, ‘discuss, confirm, validate and strengthen the group’s position’ (p.622). In doing this we reduce diversity.

(This tendency to try to reduce diversity is not only evident in online networks. It can also be seen in ‘The Big Sort’ and geographical clustering that I mentioned in my last post, i.e. people physically move geographical location to live near those more like themselves.)

More than ten years ago in 2005 in his ‘Introduction to Connective Knowledge’ (revised in 2007) Stephen Downes wrote of diversity as a key principle of ‘knowing’ networks. Downes sees the fostering of diversity as the means to

 ‘counterbalance the tendency toward a cascade phenomenon in the realm of public knowledge’.  

(Information cascades occur when external information obtained from previous participants in an event overrides one’s own private signal, irrespective of the correctness of the former over the latter’ (Wikipedia ). Cascade phenomena can sweep through densely connected networks very rapidly).

Downes writes

the excesses made possible by an unrestrained scale-free network need to be counterbalanced through either one of two mechanisms: either a reduction in the number of connections afforded by the very few, or an increase in the density of the local network for individual entities’.

According to Downes, the only way to avoid information cascades is to ensure multiple viewpoints and alternative perspectives from observers with different sets of prior experiences, world views and interpretations.

Related to this, a couple of years later Downes wrote of the different affordances of groups and networks – Groups vs. Networks: The Class Struggle Begins – saying that a group is about what members have in common, whereas ‘a network is like an ecosystem where there is no requirement that all the entities be the same.’ If we accept this it follows that a group tends towards homogeneity, but a network to heterogeneity (see also my post on the hazards of group work). Diversity is therefore essential to a healthy network.

But what is diversity?  Dictionaries, e.g. Cambridge dictionary, define diversity as being many different types of things or people, ideas or opinions, being included in something. I would add that in addition many different resources are needed to inform these ideas or opinions. In a paper that Carmen Tschofen and I published in 2012, Connectivism and Dimensions of Individual Experience, we also suggested that there is a need to recognise the importance of psychological diversity of online learners, the complexity of their human needs and connections, i.e. that diversity is not just an external manifestation of difference, but also internal to individuals. Each individual is unique. We argued that connectivity needs to be viewed not only in terms of the network but also in terms of individual characteristics and biases, further complicating an understanding of diversity.

But why is diversity ‘desirable’? dana boyd points to more diverse teams outperforming homogeneous teams and claims that diversity increases cognitive development. In my own field of research into learning in open online environments, this point of view is endorsed by the call for more interdisciplinary, multidisciplinary and cross global, international working (see for example Haywood, 2016 and Eynon et al., 2016).

However, Cilliers (2010) suggests that there are deeper reasons. These are related to viewing the world in which we live as a complex adaptive system. Complex systems are heterogeneous, asymmetrical and full of non-linear, unpredictable interactions, which means we cannot fully know or control them. Complex environments exhibit the following characteristics (and more!):

  • Distributed knowledge
  • Disequilibrium
  • Adaptive
  • Self-organisation
  • Unpredictable
  • Emergence
  • Connectedness
  • Diversity
  • Openness
  • Co-evolution
  • Interaction
  • Retrospective coherence

Cilliers tells us that diversity is a key characteristic of complex systems and is essential to the richness of the system, because it is difference not sameness that generates meaning.

An abundance of difference is not a convenience, it is a necessity. Complex systems cannot be what they are without it, and we cannot understand them without the making of profuse distinctions. Since the interactions in such systems are non-linear, their complexity cannot be reduced. The removal of relationships, i.e. the reduction of difference in the system, will distort our understanding of such systems. (Cilliers, 2010, p.58)

But this does not mean that ‘anything goes’. To get the most out of diversity and difference, complex systems require boundaries and constraints, negative, enabling constraints, ‘which determine what is not allowed to happen, rather than specifying what does have to happen’ (Williams, Karousou & Mackness, 2011, p.46). There needs to be an effective balance between openness and constraint, structure and agency.

And difference does not mean opposition. Meaningful relationships develop through difference (Cilliers, 2010), but achieving the right amount of difference to support this development, depends on ethical judgement and choice.

To make a responsible judgement—whether it be in law, science or art—would therefore involve at least the following components:

  • Respecting otherness and difference as values in themselves.
  • Gathering as much information on the issue as possible, notwithstanding the fact that it is impossible to gather all the information.
  • Considering as many of the possible consequences of the judgement, notwithstanding the fact that it is impossible to consider all the consequences.
  • Making sure that it is possible to revise the judgement as soon as it becomes clear that it has flaws, whether it be under specific circumstances, or in general. (Cilliers, 1998, p.139)

These points seem as relevant today, if not more so, than when they were written in 1998. Respect for differences and an understanding of diversity is a key ethical rule for complex systems and no amount of retreating into homogeneous groups will help us cope with living in an increasingly complex world.

As Stephen Downes wrote in 2005 when proposing connectivism as a new learning theory appropriate for living and learning in a digitally connected world:

‘Connective knowledge is no magic pill, no simple route to reliability and perhaps even more liable to error because it is so much more clearly dependent on interpretation.’

but

‘Freedom begins with living free, in sharing freely, in celebrating each other, and in letting others, too, to live free. Freedom begins when we understand of our own biases and our own prejudices; by embracing autonomy and diversity, interaction and openness….’

I agree with dana boyd – diversity is hard, but if as Cilliers (2010, p.56) says, ‘Difference is a necessary condition for meaning’ in a complex world, in order to learn we will need to embrace diversity and maintain, sustain and increase our global networks and connections.

References

Cilliers, P. (1998). Complexity and postmodernism. Understanding complex systems. London and New York, Routledge

Cilliers, P. (2010). Difference, Identity, and Complexity. Philosophy Today, 54(1), 55–65.

Downes, S. (2007). An Introduction to Connective Knowledge in Hug, Theo (Ed.) (2007): Media, Knowledge & Education – Exploring New Spaces, Relations and Dynamics in Digital Media Ecologies. Proceedings of the International Conference held on June 25-26, 2007. – http://www.downes.ca/post/33034

Eynon, R., Hjoth, I., Yasseri, T., & Gillani, N. (2016). Understanding Communication Patterns in MOOCs: Combining Data Mining and qualitative methods. In S. ElAtia, D. Ipperciel, and O. Zaïane (Eds.), Data Mining and Learning Analytics: Applications in Educational Research, Wiley.

Haywood, J. (2016). Learning from MOOCs: lessons for the future. In E. de Corte, L. Engwall, & U. Teichler (Eds.), From Books to MOOCs? Emerging Models of Learning and Teaching in Higher Education, p. 69-80. Oregon: Portland Press Limited.

Kirschner, P. A. (2015) ‘Facebook as learning platform: Argumentation superhighway or dead-end street?’ Computers in Human Behavior, vol. 53, December, pp. 621–625. Elsevier Ltd. [Online] Available at http://dx.doi.org/10.1016/j.chb.2015.03.011

Tschofen, C., & Mackness, J. (2012). Connectivism and Dimensions of Individual Experience. The International Review of Research in Open and Distance Learning, 13(1). http://www.irrodl.org/index.php/irrodl/article/view/1143

Williams, R., Karousou, R., & Mackness, J. (2011). Emergent Learning and Learning Ecologies in Web 2.0. The International Review of Research in Open and Distance Learning, 12(3). http://www.irrodl.org/index.php/irrodl/article/view/883