Conceptual consolidation after fieldwork!

While moving forward in my endeavour to build complex and critical perspectives of data in society, I launched the concept of Data Culture.

In the last months, in conversation with several colleagues, I realised that I have been using the term throughout the Webinar Series and in some articles but there is still the need for a conceptual consolidation. An article will come for sure, but let’s start from this humble research blog, sharing ideas and hopefully engaging in a conversation with other colleagues to get a deeper vision of the concept.

Let’s begin with Wikipedia’s definition on DATA CULTURE – Data culture is the principle established in the process of social practice in both public and private sectors, which requires all staffs and decision-makers to focus on the information conveyed by the existing data, and to make decisions and changes according to these results instead of leading the development of the company based on experience in the particular field.

The truth is that I was impressed by the way Rahul Bhargava and Caterine D’Ignazio adopted the concept of data culture. Through their work at the MIT Media Lab, for several years they promoted the idea of citizen’s data literacy: data that are for all and must circulate and be appropriated according to the communities or collectives’ contextualised needs.

Basing on this concept, I started working on the concept of data cultures in Higher Education. As I said earlier, data in society and in higher education institutions are represented within people’s imaginaries in very different ways, with some being enthusiastic and others being extremely cautious about the power of data. I framed this problem using Stefania Milan and Lonneke van der Velden’s lens: people’s mindsets around data can be characterised as data epistemologies. In a continuum, one extreme relates proactive data epistemologies where we can place those prone to think about data in enthusiastic terms and to advocate for access, more skills to avoid the data divide, more public open data. At the other extreme, we will place reactive data epistemologies, which will refer to those cautious and critical about data extraction and usages, particularly connected to data monetisation by private platforms. By conceiving the nature of data as something that can be public and open or private and restricted, I reflected on the several relational spaces, placing in each of these spaces the data practices and discourses I came across in the last five years. Here it goes a draft on the several data practices, discourses and literacies connected to the heuristic tool I used.

Such a situation led me to discuss with my colleague and friend Bonnie Stewart the way educators could frame and understand data. She offered me the Cinefyn framework and we particularly agreed on the fact that data in society has to be considered as something eminently complex. We offered our ideas to the world here. Recently, we connected the dots of concepts, research fieldwork and educational development here.

It was clear that the connection between the idea of data epistemologies, the complexity as lens for data and the needed literacies was given by the context. I purposely used the word “culture” to refer to the differences in the discourses, practices and narratives that any university might embrace while building and transforming its institutional identity. These processes are reciprocal, from the institution as imposition or context of opportunity for the participants, to the participants in shaping success stories, attached values, celebrated or rejected strategies. A data culture might be completely focused on rankings, metrics, measurement and the idea of objectivity, whereas others might reject the imposition of national or global metrics, and might be concerned about participants’ agency and privacy by using their data.

I hence launched the utopia. I posited that a fair data culture should be open and transparent, encouraging stakeholders to take part in data practices and usages, promoting data ethics and justice.

Building fair data cultures: a complex and transformational perspective on data.

As I discussed with Bonnie Stewart, in approaching data literacy education and professional development as a complex issue, context matters.

Indeed, professional and institutional cultures vary, just as individuals within them, and they will bring different understandings and capacities to collaboration. Such cultures cannot be modified or go in only one direction, just because they are suggested to be an entirely replicable ideal of practice. In Bonnie’s words, quoting Haraway:

This approach reflects Haraway’s (1988) concept of situated knowledges, shaped by particular social locations, material realities, and power relations. The premise of situated knowledges is that material-semiotic actors – human and non-human – actively contribute to the production of contextual, situated knowledge as “active, meaning generating” (Haraway, 1988, p. 595) parts of specific assemblages. A fragmented approach to datafication, mostly technical, supports what Haraway (1988) would call an unmarked field of vision, or the detached “conquering gaze from nowhere” (p. 581) that science has traditionally valourized as signifying objectivity.

I would also add, as I did elsewhere in 2018, that data cultures, as part of an organisational culture, refer to the following ideas:

• Pierre Bourdieu’s concept of symbolic power (Bourdieu, 1986), addressing the generation of the politics and the aesthetics of dashboards as organized visualisations of psycho- physio-neuropedagogical concepts. These are allegedly objective, but they always encompass semiosis rooted on science as discourse of power. Data-driven technologies are hence deeply entangled with political and classist phenomena defining dominant discourses of “normality” in cognitive development, social and professional behavior connected to learning performance. Moreover, these dominant discourses lie behind the design of devices, apps and the algorithms predicting behaviors. As a matter of fact, in the case of early education and care and primary school, as Williamson denounced, there are forms of  “biocodification” of childhood (Williamson, 2016) in terms of structures that define learning objectives, acceptable moods and emotions, happiness and well-being indicators, based on a number of techniques of sentiment analysis. The positivistic and micro-level approach of neuropedagogies could end up in neglecting the effects of a mass psychological surveillance of childhood towards desirable learning outcomes. In the case of professional learning, the forms of surveillance are intertwined with the worker’s freedom to define their own time and productivity, with the risk of entering in neotaylorist structures that control an invisible pacing machine (Busch et al., 2015).
• The idea of assemblages as a collection of things that have been gathered together to make sense of a social process. The concept was coined by Gilles Deleuze and Félix Guattari (Deleuze & Guattari, 1987) as an ontological framework defining  the changing nature of social entities and their interconnectedness. In fact, the assemblages capture the idea that data are entities that operate jointly with the social definitions used to make sense of them, the forms in which data is collected, whether the system forces or ignores consent, and whether this data extracted supports political aspirations of administration and control through other entities such as dashboards, institutional reports, rankings, and so on.  These entities are mediated by algorithms, which are actionable mathematical conceptualizations. For instance, after a student obtains a score, or after a number of clicks on resources in an online environment, the statistical operations will lead to the prediction of learning outcomes, and an algorithm may trigger a webmessage or an AI tutor. The simple fact of selecting pedagogical support, or just informing the student that the learning outcomes will be probably negative at the end of a semester, encompasses strong pedagogical assumptions. On these basis, we can assume that data assemblages entail socio-material components, being the social components first practices, beliefs on effective learning, prejudices of the end user ability to adopt the system, myths on the systems’ effectiveness; and second, the required infrastructures, the statistical modelling, the computational code, and the digital interfaces.

These elements compose a combination of situated beliefs, dispositions, narratives, occasions to celebrate heroes and aesthetics, which lead to shape a technological infrastructure. The engagement with such a socio-technical system in the situated context, producing or reproducing data practices and narratives, is what we can denominate “a data culture”.

Let’s take the example of what happened with the decontextualised adoption of technological solutions during the pandemic where data monetization spread at the same fast pace of adopting such solutions (Williamson & Hogan, 2021; Williamson, Eynon and Porter, 2020). A contextualised approach would require local technological development fighting for technological sovereignity (as it has been the mantra for the Open Source movement). Impeding such a contextual technological approach (for example, through “low cost” political decisions on the usage of public funding to support private monopolic platforms instead of local open source developments) might come with a deleterious impact over the ability of a collective to self-determine its technological choices and future (I need to come back on this concept exploring many raising Latin American debates and production!)

From a perspective of situated knowledges, in fact, the neutral lens is an impossible fiction that simply reinforces status quo power relations. The idea of situated knowledges and their embedded, contextual, and relational approach reject the reductionism of unreflexive data science connected to automatization and to AI as key frontier of knowledge in order to focus on the complex cultural shifts that datafication in education and society has created.

A contextual, situated approach builds educators’ literacies not just in optimizing data management but understanding its impacts, risks, and long-term implications, namely, the symbolic power behind data infrastructures connected to the several assemblages required to produce data visualizations, recommender systems, dashboards or even the simple cirtculation of metrics and statistics. It integrates the technological sphere and the understanding of data infrastructures to a vision of what is expected as human collective, what can be changed, what requires negotiation.

The data practices and value assigned to data, the institutional story around data usage; the availability or inaccessibility of technological infrastructures; the way into which all is wrapped up in discourses, jointly considered, form what we have called a “data culture”.

This idea has a number of implications for educators. Firstly, they have to consider the institutions they work for and the stakeholders they work with. What are the meaning-making processes triggered by data? What are the consensus and the common ideas around data practices? Do the actors “fear” or believe that data is being extracted and used abusively? Do they feel entitled to hack abusive data procedures? Or do they consider themselves able of modifying, enhancing or being creative with the available data?

But, when can a data culture become “fair”? Only the visibility and the negotiations on the questions posed above, concerning data practices and discourses, generate the spaces to explore complexity and take the right decisions around complicatedness.

Stakeholders’ efforts to make visible their dispositions toward data allow them to explore, deconstruct and renegotiate data assemblages and materialities (Raffaghelli, Manca, Stewart, Prinsloo, Sangrà, 2020).  For instance, students and teachers’ participation in the selection of the metrics expressing the “quality” of teaching might impact the way the university rankings are considered and used to address the allocation of funds (Sangrà et al., 2019); the discussion about the ethical principles that learning analytics must respect leads to understand data as a synthesis of values and concepts, more than as an objective and incontestable source of information (Tsai & Gasevic, 2017); understanding the relationship between the digital and data infrastructures might lead to promote an open source movement supporting the principles of data sovereignty, as expressed in the Berlin Declaration on Digital Society and Value-Based Digital Government  (2020). Datafication is evolving also in such a way that it represents several diversified situations across the globe, with geopolitical areas that are more at risk of delivering data rather than extracting value from it and vice versa, as a sort of new forms of data colonialisms (Ricaurte, 2019), something that higher education institutions have not escaped from (Prinsloo, 2020).  Therefore, in shaping its own data culture, the university must recall the continuous tension between the goals of an educational neo-humanistic perspective based on intellectual autonomy; and the technocracy’s requirements highlighting the scholars’ attention to evolving contexts of social and economic innovation. The debate on data as wealth or data as part of surveillance brings to the fore such tensions.

Specifically, universities are equipped culturally and materially to blend advanced, interdisciplinary theoretical reflection with empirical research and practice about datafication. In such a space, as envisioned early on by Wilhelm von Humboldt in the 19th century, academics and students engage in a conversation that ultimately pushes the latter to take an active part in addressing the problem about data practices and cultures as reflective citizens and professionals (Pritchard, 2004). On these bases, the university is called on to mediate meaning making around the emerging data practices through activities such as collaboratories, workshops, professional development and quality evaluation exercises in addition to existing research activities.

Patterns and solutions may emerge, but they will not be predictable in advance. Rather than by experts, they are likely to be led by collaborations that represent various invested parties and that put them in relationship with each other, as wisely suggested by Caterine D’Ignazio and Laureen Klein in their marvelous book “Data Feminism“. A complex approach to educators’ data literacy will reject universal solutions and techniques, and it will acknowledge the professional cultures and institutional cultures in which faculty development takes place, as well as the individual perspectives that faculty and staff members bring to professional learning.

In short: fair data cultures are spaces of meaning making where to disentangle the knots of complex data in the society.