Invisible Search and Online Search Engines

Front matter

INVISIBLE SEARCH AND ONLINE SEARCH ENGINES

THE UBIQUITY OF SEARCH IN EVERYDAY LIFE

JUTTA HAIDER AND OLOF SUNDIN N

Conclusion

Index vi

100 122 139

ACKNOWLEDGMENTS

This book is the result of years of research involving three projects that have been carried out in collaboration with many people whose input we could not have been without, but who at the same time cannot be blamed for anything written in the book.

The book is primarily an outcome of the project “Knowledge in a Digital World: Trust, Credibility & Relevance on the Web” (The Swedish Research Council, 2012-05632). We would, therefore, first of all like to thank all our colleagues on this project. Particularly, we want to thank those colleagues who worked together with us in carrying out the focus group discussions that we use as illustrative examples throughout the book: Cecilia Andersson, Hanna Carlsson, and Sara Kjellberg. It has been a pleasure to work with you. Without our discussions and written communications, this book would never been produced. Having said that, we of course would also like to thank all participants in our focus groups who generously shared their experiences of and ideas about searching in everyday life. The inside knowledge of your daily routines has been crucial for our project. We would also like to address a special thank you to Erik J. Olsson, who co-led the project, and to his colleagues in the philosophy LUIQ research group. You have provided us with many valuable comments that have challenged us to develop our argumentation, at the same time as the uniqueness of our contribution from information science/studies has become clearer. To some extent, the book also relies on the findings from our projects “Open Data — The Materiality and Fragmentation of Facts” (Erik Philip Sörensen Foundation) as well as “Algorithms and Literacies: Young People’s Understanding and Society's Expectations” (The Swedish Research Council, 2017-03631).

Earlier analyses of the focus group discussions we draw on in the book, as well as other related work, have been published by us in the following publications: “Invisible search: Information literacy in the Swedish Curriculum for Compulsory

Acknowledgments vii

Schools” in Nordic Journal of Digital Literacy, 2015, 10(4), 193-209; “Outsourcing trust to the information infrastructure in schools: How search engines order knowledge in education practices” in Journal of Documentation, 2016, 72(6), 990-1007 (together with Hanna Carlsson); “The structuring of information through search: Sorting waste with Google” in Aslib Journal of Information Management, 2016, 68(4), 390-406; “Searching for delegated knowledge in elementary schools” in Proceedings of ISIC, the Information Behaviour Conference, Zadar, Croatia, 20-23 September, 2016: Part 2. Information Research, 2017, 22(1), paper isic1618 (together with Hanna Carlsson), “Researchers’ online visibility: Tensions of visibility, trust and reputation”, in press at Online Information Review (together with Sara Kjellberg). In particular, thoughts developed in two publications have found their way into the book, where they were further developed and deepened, namely: “The search-ification of everyday life and the mundane-ification of search” in Journal of Documentation, 2017, 73(2), 224-243 (together with Cecilia Andersson, Hanna Carlsson, and Sara Kjellberg), and “Controlling the urge to search: Studying the informational texture of practices by exploring the missing element” in Information Research, 2017, 22(1), CoLIS paper 1613.

Drafts of individual chapters or parts of chapters have been presented at various conferences and seminars. The panel (organised by Olof Sundin and Trine Schreiber) “Is there a need for re-claiming search in information literacy practices research?” at CoLIS9 in Uppsala 2016 provided us with an early opportunity to test one of the leading hypotheses of the book — namely that the practice turn in information behaviour and literacy research has made search and search engines less visible. We are grateful to Cecilia Andersson, Andrew Cox, Helena Francke, Camilla Moring, Ola Pilerot, and Trine Schreiber, who all contributed to this thought-provoking debate then and on other occasions. This theme appears in particular in Chapters Two and Six of the book. The paper “Controlling the urge to search: Studying the informational texture of practices by exploring the missing element” was presented at the same conference by Jutta Haider, and ideas from this paper have been developed and interwoven particularly into Chapter Four. The conference paper “Critical algorithm literacies: An emerging framework” was presented at the ECREA Digital Culture and Communication Section Conference in Brighton in the autumn of 2016. The conference paper “Facts, fake and information literacy: A conceptual discussion of critical evaluation of information” was presented at ISIC 2018 in Krakow the autumn of 2018. These two papers, both presented by Olof Sundin, have primarily been developed into Chapter Five of this book as have ideas tested by Jutta Haider at a research seminar at The Swedish School of Library and Information Science, University of Boras. Thank you to Jan Nolin, Louise Limberg, and Annemaree Lloyd and the other seminar participants for perceptive and encouraging comments. Ideas that found their way into Chapters Three and Four have also been presented at the Department of Science and Technology Studies, Alpen-Adria-Universitat in Klagenfurt, Austria, by Jutta Haider. Thank you to Daniel Barben and Nils Matzner for commenting and asking thought-provoking input. In sum, the feedback and many constructive comments we were viii Acknowledgments given on all these occasions have been very important and we appreciate and are grateful for the valuable input of all participants.

The Department of Arts and Cultural Sciences at Lund University and all its staff provides an excellent environment for conducting research. Chapter One was presented at our Information Studies Research Seminar. Our thinking owes depth to the collective endeavour of our own research group, Information Practices: Communication, Culture and Society and we would, therefore, also like to mention, in addition to the names presented in the introduction of the acknowledgement, Lisa Dahlquist, Charlotte Hagström, Fredrik Hanell, Linnea Lindskold, Nora Schmidt, Björn Magnusson Staaf, Fredrik Åström, and Carin Graminius (thank you for your help with translating from Chinese). In particular we need to thank Johanna Rivano Eckerdal for her intellectual companionship and generosity. Furthermore, in our meetings with students on the Master’s Programme in Archival Studies, Library and Information Studies, and Museology (ALM) and in the bachelor program in Digital Cultures we on a daily basis develop our thinking through lectures, supervision, and other dialogues. The library facilities at our faculty are first-class, and we would especially like to mention Ulrika Karlsson who, in her capacity as subject librarian for information studies, has demonstrated what library support beyond facilitating access can be. We are also grateful to our editor Heidi Lowther and editorial assistant Elizabeth Risch for their support and responsiveness.

We have had the opportunity to discuss our research with wider society on a number of occasions, with journalists, teachers, and librarians around Sweden, and through our commissioned work at the Swedish National Agency for Education, the Swedish Schools Inspectorate, The Internet Foundation in Sweden and Government Offices of Sweden. Our outreach contacts have certainly made us better prepared to communicate the societal relevance of our research, which we hope shines through in the book.

Finally, but certainly not least, we would like to express our gratitude to the invited readers who we asked to comment on separate chapters or the entire manuscript. Dirk Lewandowski, Isto Huvila, Pamela McKenzie, and Nils Pharo have read one to two chapters and provided us with invaluable comments that helped us develop our argument. A final and special thank you goes to Ola Pilerot and Trine Schreiber who, despite a tight time frame, read through the whole manuscript. Your generous and insightful comments late in the process have developed the manuscript into something much better than it would have been without you.

Enjoy, Jutta Haider & Olof Sundin Lund, Sweden, Autumn 2018.

Introduction

This is a book about search. It is concerned with how search, searching, and with them search engines have become so widely used that we have stopped noticing them. It is thus more accurately a book about invisible search. One of society’s key infrastructures for knowing and becoming informed is computerised systems supporting the search for and locating of documents and information. The use of these systems, search engines, is curiously dispersed and centralised at the same time. It is dispersed across a vast array of social practices in which it has acquired close to naturalised positions (Hillis et al. 2013), while it is commercially and technically centralised and controlled by a handful of very dominant companies, especially one extremely powerful global player, Google. In the course of this, we can call it double movement, search has all but vanished from sight. Invisibility is often highlighted as one of the key features of an infrastructure (Star 1999) and clearly, in this sense it is safe to say that search engines are a fundamental information infrastructure. Yet, what does that mean more specifically? How do people deal with search engines? How do we research their use and which strands of previous research help us understand this all-encompassing, increasingly invisible information infrastructure? With digital tools now permeating most aspects of society, the use of search engines has become integral to everyday life on many levels. Searching for recipes or the bus timetable, for medical advice, for old neighbourhood photos or the weather report, searching for work-related documents, legal guidance or for the whereabouts of former colleagues, all this is done instantaneously on the same interface and using the same devices, thus inscribing itself into various parts of life.

Search permeates myriads of social practices and everyday life at all levels, but it often remains invisible. It appears to be simple and is done effortlessly. Yet, this effortless simplicity with which online search intersects with everyday life in so many different situations conceals an astounding complexity. Accordingly, various strands of research traditions have for a long time been dealing with different aspects of search and search engines. In fact, an entire set of considerations comes to the fore relating to issues such as the ways in which social practices are integrated with technology, with understanding what information might be and do in a certain situation, how to make sense of it in a specific context in relation to search technology, or in which ways to trust or question it. Furthermore, being searchable is today not only often seen as a feature of information, but information is also moulded to fit the shape provided by the tools used for searching for it, and, more often than not, this is a web search engine (Gillespie 2017; Haider 2014; Kallinikos et al. 2010). Inversely, this also means that information that is not produced in conformity with the rules laid down by dominant search engines gets buried and is made less visible (Mulligan & Griffin 2018, pp. 569-570). Ultimately this — we can call it search-ification — of everyday life relates to the ways in which an increasingly invisible information infrastructure is entangled across culture and its practices and to what means we have at our disposal for understanding and making sense of these entanglements (see also Sundin et al. 2017).

To look for, find, or retrieve information has always been one of the central concerns of libraries, as for information science, and at least since the 1940s and 1950s information science has been concerned with search in computerised systems more specifically. In a narrow sense and when the focus is on technical systems, the field is called information retrieval. In a broader sense and with a focus on people, it is often referred to as information behaviour. More colloquially, talk is of (online) searching or searching for information. Accordingly, these days society’s most important information retrieval systems are simply called search engines. Looking for mediated information is mostly done online and arbitrated by the various tools and devices that people carry with them on a daily basis. In addition, various algorithms and not least economic interests organise search. This way, search engines contribute to structuring private as much as professional lives and public and personal memories in ways that might not be immediately obvious.

This search-ification of everyday life is also connected to the fact that contemporary general-purpose web search engines are by most measures easy to use and return in most cases results that are experienced as useful. What used to be complicated-to-use information retrieval systems, integrated with expensive (for users) databases designed for specific professional groups and frequently queried by information professionals or librarians, have for many turned into an unassuming query box or are integrated into a mobile phone by voice recognition via virtual personal assistants. Full-text searches across an enormous and ever-growing index are carried out at a speed that makes them appear to be almost instantaneous. Neither professional education, knowledge of query languages, nor controlled vocabularies are needed to query a general-purpose web search engine and to do so in a way that is adequate for most purposes. No subscriptions are required to use them. In fact, they are increasingly seen to replace specialised retrieval systems or these have begun to emulate general-purpose search engines in order to appear user-friendly. As using search engines and doing so proficiently became feasible for the general public, searching became not only inserted into all kinds of social practices, but was also de-professionalised. We propose to call this the mundane-ification of search (Sundin et al. 2017).

Using search engines is now just another ordinary activity, hard to discern and increasingly difficult to study and also teach. This is not least due to the way that the technical workings of the system have become increasingly opaque to users the simpler to use a search engine appears. This is amplified by the fact that control over this crucial information infrastructure is in the hands of commercial organisations whose business model is based on their having full control over their index (database), their algorithms and their user data and whose “core unit of exchange” is traffic (van Couvering 2008, p. 177).

Everyday life and social practices

This book is framed around the notion of everyday life. A focus on everyday life is quite common in order to subsume all kinds of practices, behaviours, and periods that are not related either to education or the workplace. Obviously, this can pertain to extremely different situations and thus play out differently when related to search. Everyday life is a fundamentally temporal notion (Adam 1995). Its emergence and meaning is closely connected to industrialisation and to the structuring of production in capitalist society and in particular of the welfare state (Nowotny 1994). Everything is tightly structured, negotiated, and controlled by the rules of the market and of work, in terms of holidays, protected working time per week or day, shifts, overlay time, weekends, and so on. The advent of the notion of everyday life is also part of a shift from a focus on production to one on consumption, which makes possible and also requires non-working time conceptualised as leisure time or quality time and importantly also as time for consumption. Only when we think of time as something that can be subject to commodification, which it is in capitalist society, does the concept of everyday life make sense. “Everyday life has become the bracket combining work and so-called free time; the private ‘spending of time’ and the public spending form a new combination in everyday life”, writes Helga Nowotny (1994, p. 103) and she continues: “The great public institutions of the state and the economy, and their temporal perspectives, are confronted with the temporal perspectives of the citizens and employees, the economic subjects”. In information science, time is an undertheorised notion and Reijo Savolainen called already in 2006 (p. 124) for the “need to develop a research agenda in order to approach the temporal issues of information seeking more systematically”. Search engines are one of today’s most important information and communication technologies and clearly their enormous presence has implications not only for how we think of time but also for how we think, experience, and practice time.

Talking of different temporal perspectives that meet and collide, opens up for an understanding of everyday life that is more attuned to the intermingling of different structures of control, of various allegiances and demands than to a predominantly chronological notion, where one period follows after the other.

Rather, the illusive concept of everyday life can be understood as something that is part of all life in different ways and not necessarily clearly distinguished from that which is “not everyday life”. We can think of it as reaching into situations and periods of, for instance, work or education or participation in civic life. Helga Nowotny (1994) talks of the increasing blurring of the boundaries between public and private time in the phenomenon of everyday life. Interestingly for our investigation, she also identifies a related change in human perceptions of time that she discusses in relation to the ubiquity and dominance of communication technology in contemporary society. While work is sometimes conceptually distinct from everyday life as its negative and defining anti-thesis, it is also part of people’s everyday life in a more common-sense understanding. Interestingly, for many, today it is precisely the use of ICTs and general-purpose search engines that in part ties work and non-work together (see also Rosa 2015). This could be considered in terms of a context collapse, where the same tool is used in different situations, e.g. work and private life or family and friends, thus intermingling not just expectations and ways of doing things, but also time scales and temporal perspectives.

Everyday life also denotes the normal, that which is not exceptional. In this sense, everyday life is connected to habits, to repetition, all deeply temporal notions. Our account of online search in everyday life foregrounds precisely this perception of online search as the usual, the normal, as an activity that has established itself as a central part of various routine practices, that is, as one of the taken-for-granted elements making up the practice in question. Social practices are generally theorised in terms of arrangements of activities that connect in specific ways and which in these specific arrangements are observable across groups, that is, they are neither individual nor exceptional. Considerable attention, empirical and conceptual, has been paid to the intricate relationships between information and social practices. We will return to some of this work in Chapter Two. However, less attention has been given to the way in which information technology is implicated in re-arranging practices and what this means for how we engage with and organise everyday life. Chuck Moran (2015, p. 299), drawing on the foundational work of Barbara Adams (e.g. 1995), maintains “[s]ocial practices, particularly those with significant technological components, are organizing temporality in exciting ways”. And he continues: “The passing of time is becoming less important than the accrual of possible events or of sequences that can be activated whenever. Indeed, these practices challenge the very meaning of temporal terms such as memory, event, and even the contemporary”. They also challenge us to rethink our understanding of routines in social practices and of how and when they are constituted. Hartmut Rosa (2015, p. 235) draws our attention to what he calls a “paradigm shift” regarding everyday time strategies and practices. He sees the classic modern everyday routine conceptually superseded by a “temporalized everyday time of late modernity”; the first one dominated by schedules and predefined time, the latter by flexibility and deadlines. Specifically mobile technology, Rosa convincingly develops, and its potential to support the negotiation of flexibility in everyday life is implicated in propelling such a late modern time regime, where “[t]ime orderings are /.../ individually and flexibly created within time itself’ (Rosa 2015, p. 235). However subtle, search engines constitute a considerable technological component, to draw on Moran’s expression, in an ever-increasing number of social practices and their ubiquity and dominance are irrevocably tied to mobile technology. Relatedly, the role of mediated information in everyday life is being reorganised on just about every level possible and the way in which search engines are implicated in this is what is at issue in this book.

Between systems and users, search engines and search

Web search engines are formed by their use. They are not static systems that always perform in exactly the same way; rather, they are supposed to adapt to different users’ needs or expectations and also as they are being used to train and improve themselves constantly. This happens on two — closely interlinked — levels, an individual level and a collective level. On the one hand, web search engines are personalised systems. They adapt search results to individual users, or actually more precisely to their user profiles. Mostly this is done in order to increase the chances of returning search results that are perceived as being relevant and useful by the person searching and ultimately this helps to improve the placing of adverts. Increasingly and tied to the growing importance of machine learning and artificial intelligence (AI), search engines, and specifically Google, have started to develop into “suggest engines”, where active searching is not even necessary as the system is supposed to anticipate what a user wants or needs without a search having to be carried out, at least not in the sense of a person entering a query.

On the other hand, at a collective level, the aggregated queries and user logs are used to improve the system, also with the help of machine learning. Knowing common spelling mistakes, understanding which terms are often or rarely entered together, or which links are followed helps making reasonable assumptions about what others also might want to find (or be suggested). This is also shown to the users in the form of suggestions for further, related searches or through the autocomplete function, where search terms are completed as they are being typed. While this clearly has the effect of helping to formulate relevant searches and thus supports the users, there is also another side to it. Displaying common searches further re-enforces their very dominance, thus contributing to further bolstering their significance and cultural meaning. It is a kind of Matthew Effect, where the rich become richer or, in our case, the known becomes more known. In certain cases, this way — given the trust vested in search engines as neutral brokers of facts — truth and factuality can be established. As Helga Nowotny poignantly (2015, p. 58) notes, “[w]hen Google sought to gauge what people were thinking, it became what people were thinking”. The relationships between users, documents, and systems or between searchers, information, and engines are extremely intimate, in many ways they are constitutive of each other, and as search engines and search — 1. e. what we do with them — become increasingly invisible, it becomes increasingly obvious how agency here is not something reserved for humans, but a function also inherent in technology — a function distributed and in constant flux within and between different actors (Latour 2005).

Yet, despite the growing importance of search engines and their dominant role in how information is accessed and organised in society, and despite the long-standing concem of the discipline with information retrieval and information behaviour, the contributions of information science to understanding search and search engines are scattered across different subfields. Research on systems occurs on one side and research on users most often on the other. Both traditions thrive, yet, despite attempts (e.g. Ingwersen & Jarvelin 2005), they do not always connect. This is to the detriment of research on web search engines and their use in what is commonly called everyday life. One reason for this might lie with the different theoretical starting points of the two fields. The traditions simply see different things and they see things differently.

One strong strand of research is concerned with information systems: How they can be built, evaluated, and optimised and, within limits, how they are used. The literature and research produced in this area has a focus on technical issues. Accordingly, it is institutionally often situated in the universities’ various technical departments. Research in this field does consider behavioural aspects of search, that is how systems are used, yet these issues are mainly addressed in experimental laboratory studies and very rarely in everyday life situations (Jansen & Rieh 2010). On the other hand, and especially since the 1980s, there has been a reaction to the perceived dominance of this technical orientation in the field, and user studies emerged as a further strong area in information science. Here, the focus is on people and their various encounters with information in more natural, everyday life settings. Often, these are studied by drawing on general theories developed in the wider social sciences and in psychology (for a compilation, see Case & Given 2016). Here, the focus is on information behaviours and needs, cognitive factors, tasks, and literacies, and most recently — in line with the so-called practice turn that has occurred in most social sciences — information practices. Admittedly, there are sometimes vast theoretical and methodological divides and more or less subtle differences between the areas united in this user perspective. Yet, for all intents and purposes they are on “speaking terms”. What unites them is their interest in the use of information or engagements with information aside from dedicated information systems. These are positioned as just one factor amongst many, which people engage with in ways that can be captured as informational, and over the years information systems have moved further and further away from the primary locus of interest. Curiously, this has happened at the same time as dedicated information systems, which search engines after all are, have moved into ordinary everyday life with a force that has made them more or less ubiquitous. In Chapter Two, we will focus in greater detail on earlier information science research that has a bearing on the formative role of search engines in society and will present it alongside research from other related disciplines in an attempt to bring them together.

For now, as not least Ralph Schroeder (2015, 2018) has emphasised, while information science has contributed with foundational principles for how search engines work, as a discipline it has not yet contributed in any significant way to the understanding of searching and search in everyday life. Partly this can be explained by the divided nature of the discipline we have just discussed where relevant research is scattered across different subfields that are not often connected. However, search engines remind us how these two strands of research depend on each other and are intimately connected. We cannot study search without saying something about search engines and this cannot be done without acknowledging the foundational contributions made by the technical side of information science to how they function. Their technical functioning is no coincidence but is rooted in certain understandings of what they should deliver in terms of results and how these are achieved and presented to people. Conversely, we cannot study search engines without considering search, that is, how they are used. This cannot be done without acknowledging the important contributions of research to how information is looked for, stumbled upon or otherwise embedded in everyday life and its practices. As Wanda Orlikowski (2007, p. 1445) puts it, we see how the researcher’s Google search is constituted by the performativity of computers, networks, softwares, algorithms, directories, databases, and infrastructure, as these are enacted by the human agencies entailed in their design, construction and operation.

Together these actors form a “sociomaterial assemblage” (ibid.). We suggest that a sociomaterial theoretical understanding of search, which will be presented further in Chapter Three, has potential for bringing different research fields in information science together. In this sense, this book is an attempt at moving the study of information systems back into the centre of information science, yet fully acknowledging and continuing the work that has been carried out on the situational and material character of information and specifically on information in social practices.

How is a search result relevant?

In order to be meaningful, a search for information needs to lead to relevant results. It is thus not surprising that relevance is a central concept for developing and understanding information retrieval systems, which search engines after all are. “Historically”, Tefko Saracevic (2016, p. 11) writes “relevance actually crept in unannounced”, but it has since developed into one of the most important in information science. “The concept of relevance is the key issue to the functioning and evaluation of IR systems”, writes, for instance, Pia Borlund (2003, p. 923), pinpointing what is at stake. To date, most search engines return a list of links to web pages with information related to the search request. At the most rudimentary level, they look for whether the word searched for appears on a website or not or whether the pattern of an image corresponds to images online. Applying semantics to web data, search engines increasingly also return what is more properly described as answers to specific questions directly on the search results page. Either way, the results have to be displayed in an order that makes sense to the person who carried out the search in the first place. This order is established according to certain relevance criteria. Quite simply, what comes first in the list should be more relevant than what follows. It is primarily due to their capacity to deliver relevant results and show them high up in the results list that a search engine becomes useful and popular in the first place. Hence, getting the order in which search results are presented right is crucial, not least for the commercial success of a search engine. The notion of relevance however, while it sounds straightforward is an old and returning problem in information science. This is not surprising given that one of the discipline’s oldest concerns is about developing tools for locating information and information resources. In 1975, Tefko Saracevic (1975, p. 324) wrote, in the first paper of a series of literature reviews on relevance research, that “[t]he success of information science, whatever there is, is due to the fact that it did address itself to relevance”. As Saracevic reminds us, relevance has been tweaked and researched on since the 1930s in the writings of Samuel C. Bradford and in the tradition of information retrieval since the 1940s (Saracevic 1975).