As suggested by the expanded version of the acronym title — Checking Assumptions aND promoting responsibility in smart Development — the CANDID project has set about scrutinising presumptions it is finding in the discourse about ‘smart’ technologies.
Led by University of Surrey, this part of the project (named Module 4) is applying the tools of ‘discourse analysis’ to policy debates and discussions on the subject, as well as the topics covered by the other Modules, to gain insight in terms of implications to society.
Laying bare any unfounded presumptions can then enable CANDID to propose how technologies such as Internet of Things, smart homes, smart care, smart grids, smart traffic control, and others could be developed and more appropriately consider the needs of citizens and society at large.
But what is discourse analysis and what practices does it reveal?
According to the team working on this module, unexamined assumptions about the future are often presented by policy makers and research leaders in hurried support of techno-optimistic visions.
On purpose or by habit, questions of responsibility are often left insufficiently examined.
Techno-optimism is described by sociologists as an example of ‘normative’ practice. CANDID project researchers have identified numerous examples in relevant communications, using discourse analytic techniques.
But, by the same token, Social and Human Science (SSH) practitioners shouldn’t assume broad understanding their methods and terminology.
So, for the benefit of those interested in or actually now wrestling with the uncertain implications of increasingly capable ‘smart’ systems, this article seeks to clarify this part of what CANDID is offering, for helping bring about a more societally responsible ‘smart’ new world.
A sociologist’s understanding of how ‘smart’ is argued
According to Surrey University’s Dr Kristrún Gunnarsdóttir, “CANDID has been tasked with suggesting improved methods of communication, better understanding of what ‘smart’ stands for, and to point to where areas of innovation are – and could be – going. But, first we need to understand how ‘smart’ is already argued.”
“The project will make suggestions for how communications could be improved between the tech-driven ICT sectors and social sciences and humanities scholarship, as well as with other stakeholders, e.g., from citizen projects. So, we want to focus first on the current discourse.”, added Kristrún.
“We’re looking at how ‘smart’ is discussed, argued, and articulated – including about such things as artificial intelligence, Internet of Things (IoT) and ubiquitous computing – and, in particular how these developments get talked about and what is being promised it’s promoted.”
This involves content, thematic and rhetorical analyses of the discourse about ‘smart’ developments, looking at communications, such as promotional videos, speeches, policy documents, industry literature and images. It also involves more nuanced analysis, referred to as Discourse Analysis (DA).
Clarifying for understanding
Module 4 of CANDID – working with Discourse Analysis – consists of two parts: 1) analysis of the themes emerging in the work of Modules 1, 2 and 3; and 2) analysis of the policy rhetoric, it’s performative and generative functions.
This research is conducted by Dr Maria Xenitidou together with Dr Kristrún Gunnarsdóttir.
CANDID’s discourse-analytic work centres on the notion of ‘seeing’, as described by James C. Scott of Yale University’s Institute of Public Policy studies (Seeing as a State, 1998).
Here, ’seeing’ is oriented towards conditioning, using projections intended to inspire and motivate.
The CANDID project starts from a position that such discourse can, if left unexamined, help bring about certain states of affairs rather than just describe them.
According to Maria Xenitidou, “… In that sense, discourse is constituted in but also constitutive of everyday realities. Sociologists can identify such assumptions and seek the functions they may have, such as to treat something as ‘normal’, ‘factual’, or with an inherent ‘essence’ and (being) unitary.”
”The discourse analytic work gets to the heart of what is different about CANDID, but also what will follow from the analyses in developing and delivering tools to assist in more effective communications about ‘smart’.”
Identifying and questioning the normative
During the process of interviewing EU policy makers, Maria described how this approach developed from initial expectations.
“Module 4 is doing two things: it’s providing insights from the data on the other three modules and insights of its own by analysing EU commission policy documents and related sources.”
As the team started to look at the discourse about ‘smart’, Maria described how the approach evolved.
“We decided to focus on normative discourse of ‘smart’ developments, specifically policy and institutional discourses focusing on EU policy and looking at core texts and Horizon 2020 documents such as calls texts and other guides to see how smart discourses are constituted”.
“So my task was to interview European Commission officials and advisors involved in writing these documents.”
“I’m now going through a list of invitees. The latest list is of advisors to a document that was the outcome of a workshop on responsible research and innovation in SSH and ICT projects”, said Maria.
Kristrún explained further, “Normativity is a really important concept we take from science and technology studies, especially considering the content we’re observing from innovators and research leaders, policy makers, government officials and others that bring to the table questions about R&D programmes.”
“What we’re observing is that the discourse they use to express themselves is ‘normative’, in the sense that it pre-supposes a world that’s configured in certain ways.”
“They may produce (or reproduce) certain ideas and ways of seeing the world without questioning it or stopping to ponder how these have developed the way they have and what the social-cultural implications are for seeing the world in these ways”, added Maria.
Kristrún cited a example from her email In-box. ”I was just looking at an invitation to a high-level workshop next month on ‘Acceptability and Value of the IoT in the Home’. When you read the text I see a host of normativities, as is common in these sorts of communications.”
“For example, the first sentence reads: ‘IoT applications ranging from smart energy, efficient homes, to elderly and home care, offer opportunities to address global challenges such as sustainability and life-long wellbeing’”.
”What is normative here is the suggestion that technology is in ‘the driving seat’, not the ‘back seat’ or the ‘navigation seat’, as it were.”
“That is normative: typically, what is seen as ‘business as usual’. Statements like that are repeated all over the place saying the same sort of thing over and over again, but without ever examining assumptions upon which they are based.”
“It’s not that technologies can’t be in the driving seat – but it shouldn’t be seen as self-evident”.
“We’re interested in such normative discourse, and breaking up the discourse and critically questioning it.”
A ‘codebook’ for discourse analysis
As detailed in the initial proposal, Module 4 examines the ideological underpinnings of policy documents and other texts, such as any common-sense assumptions, and identifies the respective functions and aims in relation to other texts. The macro-social implications of the content is also examined, identifying which discourse is dominant, and whether it’s normative or counter-normative (for example, coming out of citizen initiatives that differ markedly from industry or government projects).
The analysis also asks, ‘who are the agents implicated in the discourse?’; ‘what is their power status?’; ‘what are the modalities of how the discourse is communicated?’ and ‘what genres are employed?’ (such as academic discourse, journalistic articles, inspirational speech, promotional discourse, and more).
The luxury of asking the ‘why’ question
Ideas presented by those in authority (such as policy makers and even research leaders) carry the risk of gaining momentum that might not be truly justified.
According to Kristrún, “Practices that become normative – like how to identify a social or any type of problem and then how to orient towards finding solutions – emerge because the majority of those that get involved orientate towards similar types of problems, and finding similar solutions without being systematically challenged.”
The ‘why’ question is often seen as a luxury, as a bit of a ‘armchair exercise’ in challenging what is the purpose and the direction of technological development. There’s no time given to leaning back and speculating on the reasons why things are done in certain ways especially why the course we are on appears to be inevitable.”
“The main reason to take issue with seemingly self-evident assumptions built into techno-optimist visions of a ‘smart’ everything, is decades of ideological blind alleys and some notable misguidance, for example, the over-simplification of what is in practice a ‘home’, ‘energy consumption’ and ‘care’ in ‘smart home’ designs, including assistive living arrangements for the elderly.”
Sense making
Maria added, “We’re not analysing everything, we’re reflecting in discourse analytics terms on our data, including policy documents and communications with peers on smart developments.”
As Kristrún points out, this sort of analysis might be unfamiliar to those working in ICT.
“Discourse analysis is a rather specialised area, but these are very well known analytic methods looking at what people say and represent… in any media”
“I think of it as about understanding. It’s about analysing talk, as well as other media.”
“It’s a very particular, nuanced and fine grained analytic technique.”
Module 4 of the CANDID project is currently examining the discourse of European Commission policy documents, and gathering insights into debates and discussions connected to ‘smart’ technologies. It is, for example, looking at the tensions between top-down thinking common to institutions and enterprise, compared with the tendency toward bottom-up thinking favoured by DIY-hacker communities, enthusiasts, and other stakeholders.
The discourse analysis is unpicking assumptions and biases, and will help provide templates for developers of ‘smart’ technologies, for greater legitimacy across society and disciplines.
In a forthcoming article, we will provide further updates and news of early stage findings of Module 4.