European Search? How to counter-imagine and counteract hegemonic search with European search engine projects

Abstract

This article investigates how developers of alternative search engines challenge increasingly corporate imaginaries of digital futures by building out counter-imaginaries of search engines devoted to social values instead of mere profit maximization. Drawing on three in-depth case studies of European search engines, it analyzes how search engine developers counter-imagine hegemonic search, what social values support their imaginaries, and how they are intertwined with their sociotechnical practices. This analysis shows that notions like privacy, independence, and openness appear to be fluid, context-dependent, and changing over time, leading to a certain “value pragmatics” that allows the projects to scale beyond their own communities of practice. It further shows how European values, and broader notions of Europe as “unified or pluralistic,” are constructed and co-produced with developers’ attempts to counter-imagine and counteract hegemonic search. To conclude, I suggest three points of intervention that may help alternative search engine projects, and digital technologies more generally, to not only make their counter-imaginaries more powerful, but also acquire the necessary resources to build their technologies and infrastructures accordingly. I finally discuss how “European values,” in all their richness and diversity, can contribute to this undertaking.

This article is a part of special theme on The State of Google Critique and Intervention. To see a full list of all articles in this special theme, please click here: https://journals.sagepub.com/page/bds/collections/stateofgooglecritiqueandintervention

Introduction

The European search engine market is strongly dominated by Google (Alphabet), with a stable market share of more than 90%. In Europe, Google is followed by Microsoft's search engine, Bing (3.63%), Russia's search engine, Yandex (1.96%), and Yahoo! (0.97%), which uses Bing's search results. All of these search engines are provided by for-profit companies and none of them have more than a 5% market share. The market for alternative search engines that support a social cause is dominated by DuckDuckGo (0.53%), a US-American, privacy-friendly search engine, and Ecosia (0.29%), a German “green” search engine trying to protect the environment by using parts of its advertising revenue to support tree planting projects. ¹ Even though the numbers differ among European countries, with Germany having a larger share of privacy-friendly search engines, for example, and the Czech Republic having its own local search engine, Seznam, with a relatively high number of users (10.8%) due to late investments by Google and better results in the Czech language ² , the overall picture is pretty clear: Google is the undisputed number one on the European search engine market. Its hegemonic position has triggered criticism from early on, much of which has focused on search engine bias and lack of algorithmic transparency (Introna and Nissenbaum, 2000; Mager, 2012a; Noble, 2018), data-driven business models that contribute to “surveillance capitalism” (Fuchs, 2011; Zuboff, 2019), as well as the company's exploitation of its quasi-monopolist position to gain a competitive advantage (Lewandowski et al., 2018). It has further led to the idea of creating a European competitor that would allow Europe to escape its dependence on US-American, and increasingly Chinese, digital technologies, platforms, and infrastructures.

Already in 2005, the French president at the time, Jacques Chirac, announced project Quaero, meant to create a search engine “to rival Google and Yahoo,” which he interpreted as a “threat of Anglo-Saxon cultural imperialism.” ³ The project was presented as a joint French/German project and received a significant amount of money from the European Union. Shortly after the announcement, the joint project was split up into a French project focusing on multi-media search and a German project, Theseus, which focused on semantic technologies. While smaller technologies and search tools grew out of this publicly funded project, it failed in its formulated aim of creating a European competitor to Google, partly due to “misguided and unnecessary nationalism,” as critics bluntly put it. ⁴ There are a number of things to learn from this brief European search engine history: First, there has been a long-standing desire to build a European search engine that would push back against hegemonic search. Second, the European context appears to be multi-cultural, heterogeneous, and highly diverse, and this needs to be accounted for in future European technology development projects. How to achieve all of this, however, is an open question that will be discussed in this article. Over the past five years (2017−2022), I studied three alternative search engine projects aiming at social change, all based in Europe: the privacy-friendly search engine Startpage, the peer-to-peer search engine YaCy, and the Open Web Index initiative. Albeit small in scale, these projects may be seen as powerful in terms of challenging increasingly corporate imaginaries of digital futures by collectively building out counter-imaginaries of hegemonic search emphasizing social values rather than profit gain. Drawing on these in-depth case studies, I will address the following research questions: How do developers of alternative search engines counter-imagine and counteract hegemonic search with their projects? What are the social values supporting their imaginaries and how are they co-produced with their sociotechnical practices? And how are European values, and notions of Europe as “unified or pluralistic” (Mahfoud, 2021: 324), intertwined with their imaginaries and practices?

In the following sections, I draw on literature from Science and Technology Studies (STS), critical new media studies, and the politics of scaling. Within this large body of research, I will particularly discuss the increasing commodification of sociotechnical imaginaries (Mager and Katzenbach, 2021), the notion of “counter-imaginaries” (Kazansky and Milan, 2021), the politics of scaling (Pfotenhauer et al., 2022) and “nonscalability” (Tsing, 2012), as well as the co-production of digital technologies/infrastructures and a European identity (Mager, 2017, 2018; Mahfoud, 2021; Mobach and Felt, 2022). After describing my study and methods, I will discuss the empirical analysis in two parts: First, I will investigate search engine developers’ counter-imaginaries, what social causes drive their imaginaries, and how they are intertwined with their sociotechnical practices. Second, I will analyze what challenges and constraints developers experience in their scaling strategies, what trade-offs they have to contend with, and what “value pragmatics” this implies. In both sections, a particular focus will be put on the way European values are constructed in developers’ attempts to counter-imagine and counteract hegemonic search with their projects. To conclude, I suggest three points of intervention that may help alternative tech projects to strengthen their counter-imaginaries and acquire the necessary resources to build their technologies and infrastructures; especially in the European context with its long-standing desire to counteract big tech companies like Google.

Hegemonic rhetoric and counter-imaginaries

Visions, narratives, and imaginaries are powerful vehicles for shaping digital futures in certain ways. Jasanoff and Kim (2009) have coined the notion “sociotechnical imaginaries” to capture the constitutive role imaginaries play in the shaping of social and political orders in the context of technology politics. They compare sociotechnical imaginaries to discourses, metaphors, and cultural meanings out of which actors build their policy preferences. In comparison to policy agendas, however, sociotechnical imaginaries are characterized as less explicit, less goal-directed, and less politically accountable (Jasanoff and Kim, 2009: 123). While the original concept of sociotechnical imaginaries strongly focused on state actors and nationhood, recent research has shown that powerful imaginaries are also articulated and enacted by corporate actors, civil society, research communities, and other organized groups in processes much more complex and non-linear than those envisaged in the initial concept (Felt and Öchsner, 2019; Mager, 2017, 2018; Olbrich and Witjes, 2016). In the context of digital technologies, the growing importance of corporate actors in envisioning, and thereby constructing, digital futures has been particularly highlighted (Mager and Katzenbach, 2021). In regard to Facebook (Meta), for example, Haupt (2021: 237) has shown how Mark Zuckerberg rhetorically constructs a “corporate vision of a better world, but also blend[s] it with the digital technologies and practices involved in making this vision reality.” In a series of public engagement events, Markham invited participants to reach into the black boxes of digital technologies through critical interventions. While the interventions helped participants to critically engage with digital platforms and datafication attempts, present and possible future visions appeared to be strongly determined by contemporary technologies conceptualized as not really giving people a choice, e.g., either we connect to social media or “we do not exist,” as Markham (2021: 397) describes the theme of “inevitability” brought forward by the participants. She concludes that “ideologies embedded in everyday discourses, materialities, and infrastructures function to self-regenerate. Power becomes hegemonic because both the control mechanisms and the ideologies are invisible, naturalized, and then neutralized” (Markham, 2021: 397). In the context of search engines, Mager (2012b, 2014) has defined the concept “algorithmic ideology” to elaborate how capitalist ideology gets embedded in and solidified through corporate technologies. Ideologies, and larger sociotechnical imaginaries, can thus be seen as tightly intertwined with the technologies they co-produce. Given the hegemonic position of big tech companies in imagining and shaping digital technologies, sociotechnical imaginaries have been described as increasingly commodified, but also as multiple and contested at the same time (Mager and Katzenbach, 2021).

Accordingly, a growing body of research has started to investigate the role imaginaries play in citizen engagement with datafication and data infrastructures (Mansell, 2012; Milan and ten Oever, 2016; Lehtiniemi and Ruckenstein, 2019). Lehtiniemi and Ruckenstein (2019: 3) have used the concept of “alternative social imaginaries” to investigate a data activism initiative aiming to shape a more sustainable citizen-centric data economy. Kazansky and Milan (2021) have introduced the notion “counter-imaginaries” to capture counter-cultural voices and practices of technology development that aim at social change. “These counter-imaginaries make apparent how civil society seeks to respond to the ever-complex technological change and the risks it conceals” (Kazansky and Milan, 2021: 366). Like dominant imaginaries, they not only enable us to understand how civil society counter-imagines digital futures, but also to observe practitioners in action as they try to shape their technological present and future (Kazansky and Milan, 2021: 366). The notion of counter-imaginaries is thus well suited to investigating not only how search engine developers counter-imagine hegemonic search, but also how they try to build their search technologies and infrastructures accordingly. In the words of Hilgartner (2015), alternative search engine developers may be seen as an “avant-garde” that aims to drive a wave of change. In his research on “sociotechnical vanguards,” the author defines them as “relatively small collectives that formulate and act intentionally to realize particular sociotechnical visions of the future that have yet to be accepted by wider collectives, such as the nation” (Hilgartner, 2015: 36). In this article, I will discuss what strategies developers of alternative search engines follow to scale and grow their projects beyond their own “communities of practice” (Wenger, 1998) and how counter-imaginaries can be anchored in larger European imaginaries.

Politics of scaling and European values

Research on the politics of scaling conceptualizes figures like Mark Zuckerberg, PayPal founder and venture capitalist Peter Thiel, and Tesla CEO Elon Musk as “obsessed” with scaling, while framing it as an indispensable part of contemporary innovation discourses and social, political, and economic life at large (Pfotenhauer et al., 2022: 4). In their analysis of economies of scale, Pfotenhauer et al. (2022) argue that the impetus of scaling is closely related to “corporate America” and disruptive practices of Silicon Valley companies like Uber, and their venture capitalists. These have been characterized as “Blitzscaling,” a “shock-and-awe tactic” aimed at social disruption that strives to “achieve massive scale at incredible speed” (Hoffman and Yeh, 2018; cited in Pfotenhauer et al., 2022: 4). Cohen (2013) has made a similar argument from a legal perspective. She argues for critically engaging with the rhetoric of privacy as “antiprogressive” and “overly costly,” as framed by Silicon Valley companies, or the larger “surveillance-innovation complex,” as Cohen (2016) called it. Against this background, Tsing argues for a nonscalability theory that pays attention to the “mounting pile of ruins that scalability leaves behind” (Tsing, 2012: 506). Not because nonscalability is necessarily better, but because it opens up the view on “diversity-in-the-making.” Nonscalability hence enables us to analyze how diversity, local specificities, and moral values—the “situatedness” of my case studies—contribute to developer practices. The term “situatedness,” which has a long tradition in science and technology studies (Haraway, 1988; see also Butler, 1990; Thompson, 2001), allows for considering differences in social, cultural, political, economic, and institutional positionality, but also for a “normative critique of hegemonic power structures and colonial tendencies that threaten to erase epistemic and political diversity” (Pfotenhauer et al., 2022: 6).

The three search engines involved in my research are differently situated for various reasons, and not only geographical ones. A central reason lies in their ownership structure being closely related to their ideological underpinnings, which I will describe as market-oriented, civil-society driven, and state-funded in the next section. Despite these differences, however, all three case studies situate themselves in the larger European context whereby constructing different notions of Europe tightly intertwined with their practices and experiences. This is in line with research having argued for developing European digital technologies, platforms, and infrastructures. Van Dijck (2021a, 2021b), most notably, made a plea for working towards “European platform societies” to counteract the overly dominant American and Chinese “platform ecosystems”. She uses the metaphor of the “platformization tree” to describe how big tech companies exert and extend their hegemonic power on all levels of digital infrastructures (Van Dijck 2021a: 2805–2807). The roots of the tree consist of computer hardware, cables, Internet protocols and the like, the trunk includes internet services and software comprising web browsers, search engines, social networking platforms, online advertising, and, finally, the branches of the tree encompass sectoral applications that are built on top of it (see also Rieder's (2022) analysis of the political economy of technical systems in this regard). US-American tech companies (Google, Apple, Facebook, and Amazon (GAFA)), but increasingly also Chinese companies (including Baidu, Alibaba, and Tencent (BAT)) successfully managed to occupy and shape large parts of the trunk, which makes them indispensable parts of connected ecosystems. The question thus is how to diversify “the tree” from the bottom up or, to put it in van Dijck's (2021a: 2815) geopolitical terms: How to develop a “European platformization tree” that

does not have a trunk that grows taller and thicker fed by proprietary data flows, but has a ‘federated’, decentralized shape. […] Such a tree may help grow a different kind of ecosystem – one that allows for more variety, openness, and interoperability at all levels.

The notion of Europe as “federated,” associated with diversity and openness, recurs in the imaginaries of alternative search engine developers, as we will see later. However, other notions of Europe are also constructed in my case studies.

This corresponds to research having shown how European values are differently constructed and co-produced with data practices, governance of digital technology, and large-scale research infrastructures. Having analyzed the data practices of statisticians, Ruppert and Scheel have shown how a “European people” is enacted in and through data politics and practices, whereby data contribute to enacting the realities that they refer to (Ruppert and Scheel, 2021: 16). In the context of the European data protection reform, Mager (2017) has analyzed how search technology and a European identity are both made and unmade in heated negotiations around this legislative act. While the rhetoric of “European values”—the fundamental right of data protection, most importantly—was strongly pushed in EU policy discourses, practical negotiations of EU-wide data protection standards pictured Europe as a “multiply imagined community” (Jasanoff, 2005) due to its political, cultural, and economic diversity: “Fundamentally different visions and values rooted in different historical experiences, socio-political traditions, economic cultures and ideological foundations all participate in the co-production of search technology and Europe” (Mager, 2017: 255). In the context of their research on the technoscientific infrastructure of the European Organization for Nuclear Research (CERN), Mobach and Felt have shown how the “bringing to life of a technoscientific infrastructure has been performing Europeanness in multiple ways” (Mobach and Felt 2022). In her research on the European Human Brain Project, Mahfoud (2021) has coined the notion of Europe as “unified or pluralized” to discuss the tight entanglements between large-scale science and technology and narratives about Europe as such: “through these narratives Europe itself is posited as a problem – the tension between unification and pluralism is serving as both metaphor for and backdrop to contestations over how scientific communities should be bringing data together in European ‘big science’ projects” (Mahfoud, 2021: 338). All this research complicates clear-cut notions of Europe by showing how different notions of Europe are co-produced with practices of shaping digital technologies and infrastructures. How developers of alternative search engines construct different notions of Europe in the context of their sociotechnical practices will be analyzed by drawing on three European search engine projects.

Studying alternative search engines based in Europe

From 2017 until 2022, I studied three European search engine projects of very different kinds: the privacy-friendly search engine Startpage ⁵ , the peer-to-peer search engine YaCy ⁶ , and the Open Web Index initiative. ⁷ The long duration of my fieldwork enabled me to deeply engage with all three developer teams and follow their developments over time. The three case studies differ significantly in terms of the technologies and infrastructures they develop, their social, cultural, and economic embeddings, as well as their “communities of practice” (Wenger, 1998). Moreover, they have different ownership structures and ideological underpinnings that I roughly categorize as market-oriented, civil society-driven, and state-funded. This categorization corresponds to European societies with a “long tradition of organizing their democracies based on balanced cooperation between market, state, and civil society actors (Mager, 2018)” (van Dijck, 2021a: 2814, italics in original). It further corresponds to the different ways the three case studies try to counter-imagine and counteract hegemonic search, starting with different approaches to the web index. Search engines do not search the web “live”, but rather search a database of websites that have been crawled and indexed before—an archive of the web, so to speak. The size, freshness, and maintenance of such a web index are hence crucial for search engines and can be achieved in different ways.

Following the logics of the market, Startpage (SP) aims to run a profitable business with privacy-friendly search technology. SP is headquartered in The Hague and cooperates with Google to benefit from Google's web index and search engine results. This enables SP to focus on data protection as its unique selling point. The roots of SP go back to 1998, when its predecessor Ixquick was developed as a meta-search engine, which was turned into a privacy-friendly search engine in 2005. SP offers Google search results without storing, using, or transmitting personal user data to Google, except from the search terms. SP users are thus provided with non-personalized Google search results and advertisements related to their search terms, but not to their personal data or “profiles.” More recently, Startpage introduced additional privacy features such as “anonymous view” for websites or a browser extension that detects and blocks trackers and cookies on websites. The company also runs StartMail, offering ad-free, encrypted email without user surveillance.

The civil society-driven search engine YaCy was created in 2003 by a German free software developer. YaCy tries to build its own web index using principles of peer-to-peer (P2P) networks. This index is supposed to be a de-centralized, open, and independent alternative to corporate, centralized web indexes like the Google index. Surfing the web through the YaCy proxy enables users, or “peers,” to build up their own web indexes, which could then be shared with other YaCy users. A crawler further enlarges this jointly created web index. Moreover, YaCy can be used for intranet solutions, search boxes on websites, or just as a search tool on one's own machine, independent from other peers. Since 2016, the YaCy maintainer, together with open source developers from all over the world, has also worked on an open source virtual assistant, SUSI.AI, ⁸ which aims to become an open source alternative to Alexa or Google home.

The third project, the Open Web Index (OWI), relies on state funding to develop a comprehensive index of the web that is open to the public, which would enable a diverse range of different search engines to emerge. More specifically, it counts on the European Union (EU) to fund such a large-scale infrastructure project. The OWI initiative was formed in 2014 and was closely related to the activities of the German nonprofit organization “SUMA-EV–Association for Free Access to Knowledge,” which runs the meta-search engine MetaGer. The OWI group was initiated by search engine researchers, computer scientists, journalists, and other interested stakeholders, mainly from Germany. In 2018, a bottom-up approach toward building an open web index started to take shape, trying to interconnect data centers and universities in order to build the index step by step. Just recently, this newly launched “Open Search Foundation” (OSF), ⁹ together with 14 European research institutions and computer centers, received funding from the EU to “create an open European infrastructure for internet search, based on European values and jurisdiction,” according to its website. ¹⁰

To analyze the three case studies, I used a qualitative mixed-methods approach combining interviews, participatory observations, website analyses, and joint workshops with the developer teams. I initially planned to use the method of “mind scripting” for all three case studies, but I had to refine my methodological toolkit throughout the empirical work to be able to grasp the case studies’ different “situatedness” (Haraway, 1988) in terms of technology, work practices, and ownership. Mind scripting has been developed to make software developers reflect upon the value systems and normative ideas that guide their work practices, both explicitly and implicitly (Allhutter and Hofmann, 2010; Allhutter, 2012). Rooted in a culture of reflection and debate, this method corresponded well to the work practices of the OWI/OSF group, which was largely made up of researchers and scientists open to participating in such a workshop. Accordingly, I conducted a two-day mind scripting workshop with OWI/OSF advocates in Berlin (2018). In contrast, the hands-on developer culture of YaCy/SUSI.AI and the corporate structures of SP complicated the realization of mind scripting workshops with these developer teams. I therefore decided to engage with YaCy/SUSI.AI developers in their own coding environments by joining them at open tech summits and community events, which ultimately resulted in two joint hands-on workshops at the FOSSASIA open tech summit (Singapore, 2019) and at the Ars Electronica Festival (Linz, Austria, 2020). Finally, I visited the SP headquarters in The Hague for one week, which allowed me to grasp the company spirit, attend meetings, work flows, joint after-work activities, and conduct a wide range of interviews with SP employees from management, engineering, usability design, PR, and support.

Altogether, 40 semi-structured interviews were conducted with SP employees, OWI and OSF contributors, as well as YaCy and SUSI.AI developers. Twenty-seven interviews were conducted face-to-face at the SP headquarters in The Hague (2019), open tech summits and meet-ups in Berlin (2018), the Chaos Computer Congress in Leipzig (2019), the FOSSASIA open tech summit in Singapore (2019), the Ars Electronica Festival in Linz (2020), as well as in Berlin, Hamburg, and Munich, where many of the OWI/OSF members are based. Thirteen interviews were conducted online as a preparation for and a follow-up to the respective fieldwork on-site. The qualitative interviews followed a rough interview guideline, asking the developers about the early days of their projects, how the projects developed over time, what challenges and breakthroughs they experienced, if/how the European context mattered in their practices, and what they wished for in the future. The mind scripting workshop with the OWI/OSI team and the hands-on workshops with SUSI.AI developers enabled me to deepen insights I gained from the interviews, to participate in their collective work practices and reflections, and to come up with new analytical categories such as the importance of scaling. All interviews and workshops were transcribed, coded with the help of the qualitative text analysis software MAXQDA, and analyzed according to the research questions following a Grounded Theory approach. This approach enabled me to cyclically collect data, analyze it, and go back to my fieldwork, thereby grounding my theories and concepts in empirical materials (Glaser and Strauss, 1968). It further allowed me to constantly refine my coding scheme that was made up of both top-down categories resulting from the research questions (such as social values, e.g., privacy or hacker ethics constituting the respective counter-imaginaries) and bottom-up categories, including “in-vivo” codes (terms that were used by the interviewees themselves such as Europe as federated or cooperative).

Empirical analysis: counter-imagining and counteracting hegemonic search

Counter-imaginaries, social causes, and European values

All three developer teams conceptualized their projects in opposition to hegemonic search, and to Google in particular. Especially when talking about the beginning of their projects, the motivation to dethrone Google and build an alternative was strongly articulated in the interviews. While the YaCy creator described his motivation as wanting to build a search engine to “fight against the big and powerful; that you can provide your own alternative,” an OWI/OSF advocate referred to Google by saying “it is astonishing and obvious that we need our own search infrastructure in Europe to escape from this digital colonialism,” which resembles Chirac's initial announcement of Quaero. How the developers counter-imagined hegemonic search and how they constructed European values in the context of their practices, however, was tightly intertwined with their sociotechnical practices and ideological underpinnings, which I categorized as market-oriented, civil society-driven, and state-funded.

In line with its market orientation, Startpage (SP) aims to provide users with a “viable alternative” (SP CEO) to big search and its data-driven business model based on heavy user tracking. It counter-imagines hegemonic search as a privacy-friendly endeavour to be achieved with privacy features resembling notions of privacy by design. Contrary to Silicon Valley's rhetoric of privacy as “anti-progressive” and “overly costly” (Cohen 2013), SP tries to turn privacy into a competitive advantage. Being in alliance with Google in order to benefit from its comprehensive web index and search engine results creates certain tensions and ambivalences, as will be discussed later. But SP developers are adamant that privacy cannot be compromised: “Yeah, definitely the privacy we will never ever compromise. That is fixed and they (Google) know that” (SP business-to-business relations). Contrary to Startpage, the civil society-driven search engine YaCy, counter-imagines hegemonic search as an independent, de-centralized technology following peer-to-peer principles. Their counter-imaginary to hegemonic search can be described as more radical in the sense of trying to build their own infrastructure deeply rooted in classical hacker ethics: “The philosophy behind it includes freedom of information, self-determination, the uncensored”. Or, as the YaCy developer essentially characterized it in an informal conversation at the Chaos Communication Congress: “Hacking is not about technology, it's about self-empowerment” (CCC 2019, fieldnotes). These quotes resonate with the “hacker ethics” put forward by the German Chaos Computer Club, ¹¹ which partly draws on Levy's (1984) initial hacker ethics, including values like free access to computers and free information, decentralization and mistrust of authorities, making public data available while protecting private data, as well as “the hands-on imperative” (Levy, 1984: 22). The third project, the Open Web Index (OWI), counts on state funding to create a comprehensive, competitive web index that is open to the public. Counter-imagining hegemonic search as basic infrastructure, the OWI initiator straightforwardly said, “it's an absurd situation, as if a private vendor would own the streets.” This is in line with research arguing that public service media and internet technologies are increasingly needed these days (Iosifidis, 2011; 2011; Fuchs and Unterberger, 2021), a plea that grew significantly stronger after Elon Musk acquired Twitter just recently. ¹² Given the “sheer size” (OWI initiator) of a comprehensive open web index, “a pan-European initiative” was imagined to fund such an index (Lewandowski, 2014).

Despite their distinct counter-imaginaries, all three projects situated themselves in the European context and thereby constructed European values tightly intertwined with their sociotechnical practices and experiences. Startpage developers constructed privacy as a European value to be achieved in practice. Their European company location in The Hague was described as an advantage, for example, because “the trend is more privacy friendly here” and “we fall under European regulations, which gives people […] more trust,” according to the SP management. Trust was framed as a very important element for privacy. Running European servers was also described as a way of evoking trust on the user side since they are “inside European jurisdiction”, at least for those users who are “educated” and look for the “right properties,” as one of the SP usability designers said. External quality marks such as the EU's privacy seal, “EuroPriSe,” ¹³ which SP acquired through a tough technical and legal audit, were further framed as raising trust in the search engine. Finally, alliances with European institutions, policy actors, and civil society were seen as ways of raising awareness about data protection, and, in turn, Startpage as a privacy-friendly search engine. In the context of their practices, SP developers constructed privacy as a European value by contrasting Europe to other cultural contexts, most importantly the US with its big tech companies and practices of user-surveillance. This resembles EU policy discourses that frame data protection as a “core European value” by crafting a European identity in opposition to “the other,” especially the US (Mager, 2017).

Developers of the civil society-driven search engine YaCy, and the open source virtual assistant SUSI.AI, also situated themselves in the European context, though in a very different way. In the context of their work practices and experiences, they shaped Europe as highly bureaucratic. Initially, the YaCy developer mainly counted on public funding, research and development calls, and consulting for public institutions to “keep the project alive.” Recruiting public money in Europe was described as very bureaucratic and distant from the “hands-on imperative” (Levy, 1984) of the developer community. Referring to his specific experiences with German funding bodies, one of the main SUSI.AI developers complained about juries that lacked both technical and entrepreneurial skills and therefore mainly looked at formal criteria such as CVs, university degrees, and academic papers. Contrasting European funding bureaucracy with Google, the YaCy developer described Google's funding strategy as “cherry picking” projects with innovative ideas and “the capacity to take initiative” rather than formal criteria. This way, Google was described as much closer to developer communities with their hands-on coding cultures than to European funding, which the developer interpreted as a “totally distinct world.” A SUSI.AI developer finally referred to the bureaucracy involved in starting a European business: “Innovation definitely happens in Europe, it's just difficult to have innovation through small businesses because it's difficult to start a small business in Europe.” In this context, European bureaucracy was also opposed to “the other”, the US and partly China in this case, which have been described as having more dynamic funding and start-up cultures rooted in different entrepreneurial spirits. The developer concluded by saying: “In the US, it's like, if you fail 10 times we love you; because you’ve made all those mistakes we don’t want you to make any more. In Europe you need to build slowly and show you’re profitable.”

The Open Web Index (OWI) advocates imagined a joint European effort in terms of funding and supporting the building of an open web index. The notion of search engines as “basic infrastructure” was strongly associated with Europe's long tradition of public funding and public service media. Besides public funding, however, diversity was staged as a central characteristic of the open web index, both in terms of diversity of knowledge and diversity of search engines. Having been asked what the advantage of an open web index would be, compared to hegemonic search engines, the initiator said, “Well, the unique selling point is that access to an index would be provided in the first place and that thousands of different services could flower on this index, which is currently impossible.” Contrary to one-size-fits-all search engines like Google, an open web index would “lay the groundwork” (OSF initiator) to enable different actors and institutions to build their own individual search tools, ranking instruments, applications, and services on top of it and would therefore better meet the needs of specific user groups, including public institutions, civil society actors, but also industrial actors. Rather than “downscaling” (Breiger, 2015) big tech platforms and networks to meet users’ diverse needs and practices, an open web index was framed as better corresponding to different cultural values, sociopolitical contexts, user needs, and localized demands right from the start. In this context, the notion of Europe as federated, multicultural, and diverse was shaped, evoking van Dijck's (2021a) work on the “European platformization tree.” This notion of Europe gained further strength in the context of the bottom-up approach towards web crawling that has taken shape in recent years to lift the OWI from the ground, as will be discussed in the next section.

Scaling strategies, value pragmatics, and Europe as “unified or pluralistic”

All search engine projects encountered challenges and constraints in their attempts to counteract hegemonic search with their sociotechnical practices, especially in regard to their scaling strategies to (be)come sustainable in the longer run. In these practices, notions of privacy, independence, and openness appeared to be constantly negotiated and renegotiated, resulting in a certain “value pragmatics” enabling them to grow their projects. Moreover, different notions of Europe, such as Europe as “unified and pluralistic” (Mahfoud, 2021), were constructed and co-produced with sociotechnical developer practices and infrastructures rather than being fixed or stable.

Counter-imagining hegemonic search as a privacy-friendly endeavor, SP strongly focuses on data minimization in its sociotechnical developer and scaling practices. Data minimization was described as creating a number of challenges and constraints—deep down in SP's hardware, on its software level, as well as on the surface level of its social media and marketing strategies. On the hardware level, running their own (European) servers was seen as much more complicated than using cloud computing, for example. On the software level, data minimization was framed as adding “algorithmic complexity” (SP developer) to their coding practices, while also having wider implications for users. If the service gets uncomfortable for users, they may start to “mistrust” SP altogether (SP usability designer). This implies a trade-off between privacy and usability, partly also between “full transparency and still making a good marketing case,” as the SP CEO put it. This trade-off was heavily debated in regard to social media and marketing strategies SP wanted to intensify after their relaunch in 2018; particularly in German-speaking countries, which SP developers described as their “natural market” given the strong emphasis on privacy in this cultural context. Talking about Facebook marketing, the SP CEO put it like this:

We had some ethical discussions in the company: Is that right to do? No, it's not right because you use information, very personal information, and you exploit that. Yes, it is right, because we want to have an impact and we want to get those people from the bad situation they’re in to a better situation. Not always easy.

In this context, their alliance with hegemonic search was critically reflected on, too, “because eventually the data that Google gathers is part of why they’re so successful, not just financially, but also with their product” (SP management). The same applies to venture capital as a means of growing their company. All these examples show that SP's notion of privacy was constantly negotiated, renegotiated, and co-produced with hardware solutions, software practices, usability design, and marketing strategies. Moreover, it changed over time due to lessons learned along the way, such as the realization that putting a simple counter on the website does not necessarily count as user tracking. The SP CEO concluded: “So, our whole knowledge and ideas about privacy developed and refined a lot.” This indicates that SP developed a certain “value pragmatics” as an outcome of a complex interplay of engineering practices, infrastructural requirements, data practices, and scaling strategies that are all at play when designing privacy-friendly features. Nissenbaum's work on “privacy in context” comes to mind here, which highlights that privacy should not be seen as fixed or stable, but rather as co-produced with the contexts around them, e.g., the context in which personal information is generated, processed, and distributed through digital technologies (Nissenbaum, 2010: 2). It further shows that specific cultural contexts, “German-speaking countries” most particularly, were raised when talking about ways of growing their user base with marketing attempts. This indicates that “European values” like privacy, which are strongly emphasized in abstract terms, were partly conflated with more situated cultural specificities evoking the notion of Europe as pluralized rather than unified.

Counter-imagining hegemonic search as an independent, de-centralized technology, YaCy and SUSI.AI developers counted on community-driven scaling strategies. Open tech summits and other meetups were described as central locations to “reach out to the community” (SUSI.AI contributor), resembling Wenger's (1998: 214) notion of “communities of practice” as a “privileged locus” for both the acquisition and the creation of knowledge. Community-driven scaling attempts involved a central challenge, which the YaCy maintainer described as a “chicken-and-egg-problem”: “Especially with search engines like YaCy, which can only work out if many people use them and which are only used by many people if they are good, right?” To solve such issues, the developer added, certain trade-offs between developers’ ideology and end users’ needs are required. Lack of money and resources, however, increasingly became an issue too, especially for SUSI.AI developers who wanted to scale the project. Having been disappointed by European funding agencies, as elaborated earlier, SUSI.AI developers decided to experiment with the Google Summer of Code funding program (GSOC), a funding scheme for open source projects benefitting from paid computer science students who contribute to the project over the summer. While the peer-to-peer search engine YaCy was initially rooted in the rather strict German free software movement, more pragmatic values entered the developer community, with people now contributing to SUSI.AI from all over the world through GSOC, most importantly from India. This multi-culturalism entailed a diversification of values and ideologies within their own developer community that ranged from “market liberalism” to “communism,” according to an Asian SUSI.AI contributor I interviewed at the FOSSASIA summit in Singapore. Moreover, the strong emphasis on data protection was renegotiated in more pragmatic approaches towards developing open source virtual assistants that rely on open data. In the context of machine learning used for speech-to-text translations, a SUSI.AI contributor concluded that “open is not trivial”: “What does open data mean? With which data was the model trained? Can we use them? Can we train the model anew?” Accordingly, values like independence and openness appeared to be renegotiated and co-produced with new technologies, machine learning models, and funding opportunities, also leading to certain “value pragmatics.” This resonates with Coleman's (2013) research on the heterogeneity of “hacker ethics.” It further corresponds to Birkinbine's (2020: 8−9) distinction between the free software movement, tightly intertwined with its radical founder Richard Stallman, and open source communities described as more open, flexible, and less anti-capitalist. In the interviews, German developers tended to relate themselves to data protection and anti-commercial ideology more than Asian developers, who inscribed themselves in more pragmatic, partly liberal, open source communities. This indicates that just like privacy, other supposedly “European values” like openness and independence were partly conflated with more specific, culturally situated contexts. Moreover, the imaginary of bureaucratic Europe was localized when talking about frustrating experiences the developers had with particular German funding bodies, as argued earlier, hinting at multiculturalism and diversity within European countries instead of notions of Europe as a coherent whole.

Counter-imagining hegemonic search as basic infrastructure to be built from scratch, funding was discussed as a central issue by my interviewees. The OWI initiator put it like this: “It's not a research project and it's not a project that would fit into any funding program because of its sheer size.” Since recruiting major EU investments from the top down (hundreds of millions of Euros were hoped for) appeared to be challenging and to take a long time in practice, the imaginary of a bottom-up approach toward web indexing took shape (they finally received an EU grant of 8.5 million Euros to start the project, at least ¹⁴ ). The main initiator of this newly founded Open Search Foundation (OSF) envisioned libraries, data centers, universities, and other European entities building a crawling and indexing algorithm that would accumulate a web index step by step. Later in the interview, he described the project as “a kind of computational movement in Europe” evoking the notion of Europe as being culturally diverse, federated, and cooperative: “This is why we need a special spirit here to benefit from the federated, rather cooperative structures in Europe.” In contrast to the US, which has a “huge lever in terms of a big market” that would allow for strategies of “Blitzscaling” (Hoffman and Yeh, 2018; Pfotenhauer et al., 2022), Europe would need to slowly coordinate its federated structures to build a comprehensive web index from the bottom up, the OSF initiator concluded. This resonates with Tsing's (2012) notion of “nonscalability,” enabling us to grasp “diversity-in-the-making.” Moreover, the long duration that is needed to build such a large-scale search infrastructure proved to be challenging according to the mind scripting workshop participants. This is in line with infrastructure studies which argue that it is not only lack of time and resources, but also temporality that matters in building large-scale infrastructures. Karasti et al. (2010) introduced the notions of “project time” and “infrastructure time” to better understand the multiple temporalities that are at stake. Ribes and Finholt (2009) use the concept of “the long now” to capture tensions between demands of the present and a desired future that infrastructure developers have to constantly balance. In particular, “taken-for-granted short-term temporalities” (Karasti et al., 2010: 380) hamper more long-term funding, but also conceal the long durations needed for building up large-scale infrastructures.

The bottom-up approach towards web crawling and indexing therefore also resulted in a certain “value pragmatics” that were described in terms of breaking the big task of indexing the web into small “projects”, which could be made “manageable”. Then “people specialize only on those small pieces and work together in a very open community way” according to an OSF developer. This would enable different projects to “run in the same infrastructure” and also attract “low-level potential end users”, the OSF contributor added. This, however, also points to the “infrastructural complexity” (Star and Ruhleder, 1996; Karasti et al., 2010) involved in building such an infrastructure and the governance questions that come along with it. One question raised by my interviewees is the question of what data and documents to exclude from the index and on what infrastructural level. Drawing a comparison to Google, an OSF developer explained: “then you have this, let's say, decision which is made very low down in Google. […] and the end users have to live with that”. Contrary to meta-search engines that cannot challenge “the American rules about indexing, what is being found and so on”, an open web index would allow for “totally reorganizing the way you collect the data and how you build indexes and how you make competition on all levels”, as he explained. All these examples resemble the entrepreneurial practice of breaking big tasks into small projects, but also the notion of a pluralized Europe associated with federalism, multiculturalism, and diversity. They show that different approaches to building an open web index are intertwined with different notions of Europe that resonate with Mahfoud's (2021) research on the Human Brain Project. In the light of this research, the OWI and OSF approaches may not only be seen as enacting different visions of the open web index, but also as co-producing different imaginaries of Europe as “unified or pluralistic” (Mahfoud, 2021: 324). How to coordinate different projects, crawling attempts, and ordering mechanisms – or “how to unify while retaining diversity?” (Mahfoud, 2021: 338) – remained an open question for my interviewees. What their project underlines, however, is that just like European values, broader notions of Europe appeared to be highly context-dependent due to the cultural heterogeneity of Europe, but also due to the “infrastructural complexities” (Star and Ruhleder, 1996) involved in building a comprehensive web index from scratch.

Discussion: Three points of intervention

In the empirical analysis, I have shown how alternative search engine projects try to challenge increasingly corporate imaginaries of digital futures by collectively building out counter-imaginaries to hegemonic search that are devoted to privacy, independence, and openness. I have shown that trade-offs between ideology and feasibility, partly also between ideology and usability, are needed to enable them to scale and (be)come more sustainable in the longer run. In these trade-offs, social values appeared to be negotiated and renegotiated, as well as to change over time, resulting in a certain “value pragmatics” that allowed the projects to grow beyond their own “communities of practice” (Wenger 1998). Moreover, European values, and broader notions of Europe, turned out to be context-dependent and co-produced with sociotechnical developer practices and search infrastructures. In the following, I suggest three points of intervention that can help counter-imaginaries to grow and alternative technologies and infrastructures to flower . How “European values” can contribute to this undertaking will be finally discussed.

Funding and slow scalability: Considering the “infrastructural complexity” (Star and Ruhleder, 1996) involved in building and running a comprehensive web index, major funding would be required, first of all. Not only is the amount of money relevant in this respect, the temporality also needs to be considered. In addition to funding short-term “cutting-edge technology projects,” “innovation that emerges in the long-term” is critically important, according to Karasti et al. (2010: 407−408). Moreover, bureaucratic processes of public funding described by my interviewees may be reconsidered given that certain “communities of practice” (Wenger, 1998) are organized around the “hands-on imperative” (Levy, 1984) rather than formal CVs or academic practices of writing grant applications. Especially in Europe, where well-funded public institutions and media have a long tradition, bolder ways of funding and fostering digital technologies and platforms need to be found. Funding programs that better correspond to the hands-on spirit of developer communities would bring public institutions closer to the social values, technical skills, and knowledge that are needed to build more open, democratic, and sustainable technologies. Such programs would be able to better correspond to the “situatedness” (Haraway, 1988) of localized projects associated with notions of Europe as diverse, federated, and cooperative. Rather than “downscaling” (Breiger, 2015) big tech platforms and infrastructures to meet specific user needs, such projects would be embedded in localized contexts right from the beginning. This would facilitate more heterogeneous search engine landscapes, thus diversifying the access to and ordering of knowledge. To reach this goal, Europe may be advised to count on “slow scalability” and long-term funding, which would be needed to build search engines and infrastructures devoted to the public good instead of using quick venture capital and strategies of “Blitzscaling” (Hoffman and Yeh, 2018; Pfotenhauer et al., 2022) promoted by big tech companies.

Continuous auditing and advice: Second, the fluidity of notions like privacy, independence, and openness needs to be considered in the development and governance of digital technologies. All three case studies have shown that both the technologies and their value systems are constantly changing and transforming along with the introduction of new features, practices, infrastructures, marketing strategies, and funding opportunities. This confirms research by Gürses and Hoboken (2018: 598), who have argued that the “agile turn” in software development needs to be considered in governance practices since “the way in which digital functionality comes into the world” affects “privacy and the conditions for its governance”. Rieder and Hofmann (2020) suggest the notion of “platform observability,” in contrast to transparency, to highlight the fluid and transformative nature of digital platforms, practices, and infrastructures. This would require new institutions like a “European Platform Observatory” that would be provided with highly specialized technical expertise, a public interest mandate, adequate funding, and strong regulatory support so as to practically hold platforms accountable (Rieder and Hofmann, 2020: 23). Given the results of my study, however, continuous auditing and advice are not only needed after the implementation of digital technologies, platforms, and algorithmic systems, but even more so before that. Especially in the phase of developing digital tools and infrastructures, constant advice and public scrutiny are needed with regard to legal requirements, ethical and governance issues, as well as social implications. This supports research showing that ex-post auditing of profiling systems in public sectors (Allhutter et al., 2020) or of human rights protocols (ten Oever, 2021; Ermoshina and Musiani, 2022) is often too late and that new institutions would be needed with enough resources and interdisciplinary expertise, including those from civil society actors, to accompany sociotechnical development processes right from the start.

Opening up data?: Finally, data was mentioned as a necessary ingredient for growing European infrastructures. The lack of user data, open data infrastructures, and training data for AI developments have all been raised as possible constraints in the context of developing digital technologies devoted to the public good. The central question is thus what framework conditions would be needed to grow an alternative digital ecosystem that “does not have a trunk that grows taller and thicker fed by proprietary data flows, but has a ‘federated’, decentralized shape” (van Dijck, 2021a: 2815). Given that “open is not trivial,” as one of my interviewees put it in the context of machine learning, the question is how to open up proprietary data. Data sharing mandates have been discussed in this context as a way to legally force big tech companies to open up their data and share it with societal stakeholders (Grafenstein et al., 2019). Van Dijck (2021a: 2815) refers to the principle of “data sovereignty” in this context, which would give

users the ability to control the storage, accessibility, and processing of their own (meta)data. When switching between different platforms, users could be allowed to choose a specific data regime: they can keep their self-generated data private, donate it anonymously to a “data commons”, or put their data at the disposal of particular platform operators.

More recently, there have been demands for collective ways of owning and protecting user data that would make it possible to go beyond individual control of and responsibilities towards personal data. In the health context, Prainsack et al. have proposed the notion of “data solidarity” to strengthen collective control and ownership of data (Prainsack et al., 2022; see also Prainsack 2019). Alternative modes of data governance would also be a necessary requirement for the “European Platform Observatory,” raising the complicated question of “how data and analytical capacities should be made available to whom, and for what purpose. This clearly goes beyond data access” (Rieder and Hofmann, 2020: 21).

Conclusion

At the beginning of this article, I raised the question of how to consider multi-cultural, heterogeneous, and highly diverse European contexts in digital technology developments. Some answers can be drawn from my research. First, a diverse set of “European values” can be used to strategically position alternative search engines in opposition to hegemonic search, Google in particular. They can be used to make their counter-imaginaries more powerful by anchoring them in larger European imaginaries revolving around data protection (Mager, 2017), but also around broader notions such as openness, fairness, and sustainability—all stated values in recent EU policy documents outlining “the European way for the Digital Decade” (European Commission, 2021). Future research is invited to investigate more deeply how European policy imaginaries relate to, overlap with, and contradict counter-imaginaries of technology projects from below. Second, notions like “bureaucratic Europe” highlight the challenges and constraints technology projects experience in Europe and how to intervene – also in more localized, “situated” (Haraway, 1988) contexts that are partly conflated with more abstract framings of Europe. The notion of Europe as unified lends itself particularly well to these purposes since it makes it possible to draw a bigger picture by distinguishing Europe from both the US and China with their corporate digital technologies and surveillance practices. Finally, the notion of Europe as federated, multicultural, and diverse can be strengthened to promote digital technologies and infrastructures devoted to values like decentralization, collectivity, and cooperation. Diversifying technology developments can contribute to a range of different search engines, social media platforms, and infrastructures, which would enable users to bypass hegemonic gatekeepers and their commercial bias and discriminatory content that have long been criticized (Introna and Nissenbaum, 2000; Mager, 2012a; Noble, 2018). It could lead to “fundamentally different projects that challenge power at their source,” as data justice scholars have called for (D’Ignazio and Klein, 2020: 65).

The importance of maintaining diversity in technology design and creating alternatives to corporate, centralized platforms can also be seen in light of Elon Musk's recent acquisition of Twitter and the worrying developments that followed. The non-centralized social network Mastodon, created by a German free software developer, is quickly gaining traction now that many users are fleeing Twitter. Mastodon counts on federalism not only in terms of its technical shape, but also in terms of its mode of governance, which has been coined “convenantal” instead of “contractual” (Gehl and Zulli, 2022). However, once the social network grows significantly, many of the challenges and constraints discussed in this article will need to be urgently dealt with– and not only those of a technical nature, but especially those of a cultural, social, and political nature, including questions of governance, anti-discrimination, and sustainability. To properly address these questions major resources will be needed in terms of long-term funding, but also in terms of interdisciplinary expertise and oversight abilities. Given its long-standing desire to build European infrastructures, the European Union (but also other actors like public service media) may well be advised to not miss the momentum and start thinking about ways of supporting and engaging with digital technologies driven by distinct European values, in all their richness and diversity, rather than trying to mimic big tech companies evoked in notions of “rivaling” Google & co. The possible first steps that could be taken in this direction have been discussed in this article.