Keywords

Virtually all of the numerous publications on the new possibilities of computer-aided data analysis refer to it as a revolution. A revolution that, following the spread of the PC and the internet, has new and far-reaching consequences for the digitalisation of everyday life. The exponential growth in data sets that can no longer be analysed using conventional methods is referred to as big data (Hilbert and López 2011). Currently, it takes less than two years for the global data volume to double. This unrestrained growth is likely to continue in the future, with the development of the Internet of Things, data from social networks, the growing number of sensors in smartphones, cameras in all places, GPS-based motion profiles or radio waves that contribute to contactless identification (RFID), intelligent clothing and many other data sources.

Yet the revolution that is talked about in relation to big data does not refer to a supposedly uncontrollable flood of data, but rather to the technological possibility of using these data as a valuable resource. The goal is therefore to collect even more of them. Thanks to increasing computing power, innovative techniques of data acquisition and ever more intelligent algorithms in the evaluation software, they promise companies, organisations, politicians and private users enormous gains in knowledge in many different fields of activity. It is becoming increasingly possible to identify relationships that were previously too complex for human capacities, using them to develop models that deliver valid predictions for future behaviour. Scholars have been talking about the end of theory, as the abundance of data allows almost any correlation to be calculated without making assumptions, and the data-based competition between companies for customers it will lead to is referred to as “digital Darwinism” (Kreutzer and Land 2015). In another context, the data revolution creates the possibility of total surveillance of one's own body, turning sick people into medical professionals and hospitals into globally oriented databases (Herland et al. 2014). Urban planners are envisioning future “smart cities” where energy and transportation flows, utilities and social communications are linked into a comprehensive control system (Dameri and Rosenthal-Sabroux 2014). In the economic realm, the consequences of increasingly autonomous, data-driven decision-making processes (Power 2015) as well as the prospects of a healthcare system permanently interlinked with humans (Khoury and Ioannidis, 2014) are discussed.

This revolution marks a fundamental expansion of cyberspace, allowing for networked miniature computers to intrude into the real-space everyday environment. In cities in particular, this is happening so naturally that users of smart homes or customers facing the advertising on their display that reacts autonomously are less and less aware of the space in which they are acting in any given moment. Cyberspace and physical real space are fused in ways that are slowly transforming physical space into digitalised space.

In relation to the debate on social inequality (and considering the limitations that have been discussed), the ubiquity of data mediated by computers could improve opportunities for advancement, as they imply a further spread of digital modes. Resources relevant to advancement, such as information and contacts, are becoming even easier to obtain, and with them opportunities for utilisation and qualification.

But releasing data is not all the digitalised economy is doing. It is also recording them, utilising personal data with maximum efficiency. Personal data generally refers to information about an identified or identifiable natural person. The possibility of obtaining personal data is firmly embedded in ambient intelligence, both online and offline, and must therefore be considered as a general condition for digitally enhanced modes of action.

As will be shown, private-sector data utilisation in particular can be accompanied by highly specific influences with a stratifying effect both in real space and in virtual space. In this context, too, we are dealing with a revolution that is taking place slowly and quietly, but which will massively restrict the opportunities for advancement of disadvantaged segments of the population in the long term.

This section will look at the utilisation contexts in the data-related value creation process, based on the sources that are used for generating personal data. Looking at actors such as data traders and companies, who will not be able to compete without customer data in the future, the techniques of influencing individuals in cyberspace and digitalised real space will be presented, followed by an analysis of the socially stratifying effects of big data.

4.1 Personalised Data and Their New Sources

In all of our daily activities, we leave data traces. Starting with the name on the birth certificate, the accumulation of personal data continues with the application for official documents and the registration of a place of residence with a specific address. We shop at certain places, possibly paying with a debit card or perhaps even using a personalised loyalty card. We sign credit and insurance contracts, participate in surveys and contests or book a vacation. In the virtual sphere, we communicate on social networks, take an interest in certain news, use search engines and shop online according to our individual preferences. In each of these situations, personal data are generated.

While the storage of basic data by public authorities beyond a certain scope has repeatedly been the subject of critical debate (census, data retention), public awareness of the private-sector's interest in personal data has only developed more recently. Meanwhile, for many private-sector businesses, capturing the everyday routines of individuals via their real-space and digital traces has long been standard practice: Information from offline and online sources is collected, categorised and analysed. Big data enables companies to track each consumer individually according to their preferences and habits across all basic functions of existence by analysing consumption, search behaviour, sales, page views, geographic location, demographics, social situation or contacts with others. The methods and technologies used for this purpose can be summarised under the term data mining; a process where raw data are transformed into information that can be put into value in various ways.

Data mining is particularly important in the context of internet use. It is true that the possibilities of obtaining consumer data in physical space have increased notably in recent years, as postal addresses, vouchers or customer cards are evaluated via modern database management systems. But for all parties with a commercial interest, the steady online influx of data holds infinitely more potential, as IP addresses, personalised logins, the setting of cookies and other tracking methods have made it easy to follow individual traces on the internet. With the help of search queries, ratings, web browser information or biometric data, it has become possible to precisely identify individuals and systematically evaluate their online activities (cf. Bujlow et al. 2017).

The predominance of smartphones on the mobile device market provides particularly favourable conditions for personalised data analysis. Since a smartphone is usually only used by its owner and accompanies them permanently, the device contains many user specifics. Addresses and photos stored in it, active contacts, calendar entries, visited pages as well as extensive location data provide a wealth of information about its owner.

In addition, data that are collected from the digital networking of numerous everyday objects can measure individual actions and activities. Microcontrollers, i.e., small, powerful computers on chips, enable elements of the physical environment to communicate with each other and to process information in a personalised manner. Such connectivity goes far beyond the portable data storage devices (wearables) already mentioned and is increasingly permeating our everyday lives (Jin et al. 2018). For example, households can be equipped with remote-controlled heating, cooling or water supply systems, intelligent household appliances can be adjusted to the preferences of their owners, sensors can autonomously alert the police and fire department in the event of faults and medical care can be geared to regularly used measuring devices. Thanks to its cost and environmental efficiency, smart metering is becoming increasingly common in many households. To optimise energy consumption, data on the power consumption of each household is being recorded in real time. However, the data not only reveal the consumption curve over the course of the day, but also allow conclusions to be drawn about which devices are used and when. In this way, each household´s preferences and habits become transparent. Power consumption can be used to determine when people get up, and screen brightness can be used to determine which films are being watched (Cuijpers and Koops 2013).

The merging of objects from the physical world with their representation in the world of data to form hybrid systems is often seen as the new revolution of the internet. The Internet of Things implies a memorisation capability of objects to record, store and retrieve information. Furthermore, this information can also be processed in a context-dependent manner. It is significant that the availability of data for a particular object does not depend exclusively on external sources, such as information from smartphones or from external providers. In addition, many objects are able to use sensory capabilities to gather information on their own and interact with the environment. An extension of the Internet of Things to the residential and living areas of the urban population results in the concept of the smart city, that is, a data-driven urban development aimed at social, economic and technical innovations. While there are various ideas about what this concept looks like in concrete terms (Cocchia 2014 for an overview), a smart city involves finding solutions to current problems through the intelligent networking of urban space. Across the globe, the focus lies on infrastructure improvement, environmental protection, community participation, economic productivity and public utilities, with data-driven intelligence promising increased efficiency, performance and competitiveness. Another central component of the concept of the smart city is the digital measurement of social routines in urban space.

This digital penetration adds further complexity to the conceptualisation of space as a relational arrangement. In order to detect different structural features with regard to capital acquisition, physical real space and cyberspace were first compared with each other, taking into account reciprocal references and the option of simultaneous utilisation. Something as simple as a tweet in a pedestrian zone involves a double action in space: First, depending on its importance, it leads to changes in cyberspace that other Twitter users refer to. A transfer back to physical space can happen as soon as the tweet influences actions there (A more extreme example of this could be the crash of a stock price as a result of a tweet, which in turn would lead to capital withdrawal and altered conditions for action in numerous places around the world). At the same time, the tweeting user directly influences the constitution of the real space they are in through their presence.

With the digital networking of social goods and people, yet another structurally defining form of the digital now comes into play, resulting in an external predetermination of the conditions for action. With reference to the global interlinking of actions (the tweet causing repercussions across borders could again be mentioned here), the duality of space has been generalised to the extent that action results and action conditions can diverge spatially and agents can no longer be identified in a global constitution process. With digitalisation, however, the structure is not only (partly) shaped by external factors, but the external becomes an actual part of the structure. Social assets, as well as people themselves, can be linked to technology, data, messages and instructions. With smart implants, they can even merge into one unit. The power to constitute space no longer lies solely with those actors who act in real space under the rules that apply there, but to a significant extent it lies with those actors who control the code of the technologies used (Kitchin and Dodge 2011). The limitations to bring about a structural change in cyberspace described above therefore also characterise digitalised real space to an increasing extent. As has been explained with reference to cyberspace, the external predetermination of the digitally expanded framework of action does not seem to be accompanied by a limitation of individual possibilities at first. On the contrary, the countless offers of cyberspace and digitalised real spaces provide many additional options, their attractiveness constantly reaffirming the digital lifestyle. The surrender of specific rights, including personal data, is the only price to be paid.

From the point of view of data utilisation, digitalised real space, compared to virtual space, is characterised by a direct recording of personal characteristics. Routines and preferences are being transmitted directly and authentically, inviting a response in real space on the part of the public, but also on the part of commercial actors. The penetration of digital recording and processing capacities into the private sphere must be seen as a key prerequisite for corresponding data use, as has recently become possible and common through the acquisition of networked consumer products (such as cars, televisions, wearables or kitchen appliances).

The so-called sentient city—an urban space interspersed with sensors, cameras, actuators and data clouds and endowed with autonomous response and control capabilities—is already being examined as a social and political sphere for the future (Shepard 2011; Thrift 2014). The new conditions discussed with respect to this urban context are expressed in several ways (cf. Crang and Graham 2007): On the one hand, the sentient city implies an extension of real space by overlaying physical objects with virtual objects. The conventional topography of the city gives way to a personalised environment that can affect the individual in real time. Signs, ads and physical access points in urban spaces change depending on the individual identified. On the other hand, humans actively transfer functions to the environment via various technical aids surrounding them, creating an environment that is an expression of delegated functions and interests (Amoore and Piotukh 2016). Expanding this perspective, we could envision algorithms that automatically bring together people, places and objects (Safransky 2020). With the growing ability of computers to learn autonomously, i.e., the development of artificial intelligence, correlations between cause and effect, action and reaction are no longer trackable.

Such spacing (here more in the sense of “positioning” as opposed to “erecting”) takes place largely unnoticed, as spaces are marked and occupied mostly without physical change. The corresponding technical conditions are visually inconspicuous, the data and working algorithms as invisible as their creators (Greenspan 2021). In contrast, when the positioning of people changes, the constitution of space does happen noticeably. Its composition and the atmosphere that changes with it then shape perception and the operation of synthesis in a way that clearly offers the capital-poor segments of the population fewer opportunities for connection (Eubanks 2018).

However the sentient city will look in the coming decades in technical and social terms, at present the diversity of data sources is already evident, as is their potential for selective exploitation. This is also underscored by a thriving data market that is nurtured by professional data traders and absorbed by enterprises.

4.2 An Efficient Form of Address: On the Utilisation of Personal Data

The private-sector exploitation of data is shared by companies that have their own online and offline data collection capacities or generate data themselves by offering internet services, by specialised service providers that extract data from the internet and process it for third parties and by data traders who specialise in selling data. All three groups also act as buyers in data markets. A clear distinction between them is not practicable, since the service packages of many providers include both the collection and the processing and trading of data. With the help of various analysis tools, even small companies outside the field can become players in the international data business.

Every form of new, digitally-based gathering of information builds on a technical infrastructure that makes the systematic accumulation of data possible in the first place. Even if the actors involved in data processing and data trading have not created this technical infrastructure themselves, they are links in the complex chain of a data-based value creation process that ultimately helps to define the programming of codes as well as the setting of rules in cyberspace and in digitalised real space. The opportunity to monetise what can be captured digitally is having an impact on the technical orientation of many offerings. In business, personal data are paving the way for various applications, especially in the product areas of marketing, people search and risk protection.

Database marketing is generally understood as the development and use of customer databases to increase marketing productivity through more effective acquisition, retention and development of customers (Blattberg et al. 2008, p. 4). Closely related to database marketing are the concepts of direct marketing and dialog marketing. The term direct marketing describes all marketing activities that aim to address a target person and achieve a measurable reaction (response), with the goal of establishing long-term direct contact and maintain it as permanently as possible. Personal sales letters, e-mails, SMS, calls by call centres or even personal interaction with a representative ensure that direct contact with the customer or prospect can be established and that their reactions can be recorded and systematically evaluated in databases. The term dialog marketing, which has been used more in recent years than direct marketing, focuses on longer-term interaction with the target person. While direct and dialog marketing emphasise interaction with individual customers and potential prospects, which can be realised both via virtual and real-space channels, online marketing (also referred to as web marketing or internet marketing) focuses exclusively on communication channels on the internet. In addition to the provider's website, this includes various forms of online advertising (such as classic banner advertising). In addition, there are advertising messages contained in media sharing platforms (e.g., YouTube or Flickr), online stores and downstream e-mail advertising, as well as all measures that initiate visits to a specific website via search engines (search engine marketing) or via social networks (social media marketing). The strategic goal of providers here is to influence blogs in order to bring goods and services into the conversation. They closely follow online conversations (text mining), try to gain a foothold in forums and communities, influence word-of-mouth or post targeted messages on social media sites (Droste 2014; Dwyer 2009; Yadav et al. 2015).

Mobile marketing is another rapidly growing segment of online marketing. With the help of location-based services, messages and offers can be tailored to the user's particular location. This structuring of information around the user in real space will facilitate a highly personalised approach in the future. On smartphones, advertising can be displayed near the product, and with the help of augmented reality commercial offers can be presented based on previous consumption habits (O'Mahony 2015). For example, there are apps that help a consumer navigate through a supermarket based on their preferences (e.g., Millonig and Gartner, 2011). Expanding on this development from an economic perspective, environments will likely one day be designed in an individualised manner, focusing on marketability.

Many companies are attempting to verify the identity of individuals or validate the classification of people, accounts or products to minimise default risks and assess the basis of business interaction. The insurance and healthcare industries rely on the data traders’ portfolio in the fight against manipulated billing and falsified applications, as do the financial services sector and internet mail order businesses in their fight against fraud attempts. A key measure in combating fraud is to quickly identify individuals and existing connections between individuals. Thanks to personalised data, many customers can be assigned a specific risk value that provides information about their liquidity and their reliability. In the area of creditworthiness, sophisticated scoring models have become established that convert a large number of different life data into the probability value of ongoing liquidity.

It is also becoming more common for employers to take an interest in applicants’ traces on the internet, increasing the pressure of presenting a flawless resume beyond the CV.

The business areas using personal data are widely scattered, they keep growing and only a selection of them can be outlined here. In each case, the goal lies in predicting the customer, whose needs and business risks can not only be identified but increasingly also predicted. As part of analytical customer relationship management, many service providers offer to analyse a customer's (buying) behaviour in order to draw new information from it and identify future patterns (predictive analytics). This can then be used to gain further insights into the customer lifecycle, derive new ways of addressing or identify susceptibility to competitor products. It has even become possible to predict the future liquidity of customers. Furthermore, personal analytics can assess customer value, give a termination forecast or provide information on typical investment behaviour.

The subject (as potential customer, business partner or employee) in all its complexity is replaced by a compilation of data whose commodifiable characteristics can now be selectively addressed by economic actors, which will be examined in more detail below.

4.3 The Personalised Interaction Between Physical Space and Cyberspace

As knowledge about customers increases, it is becoming easy for informed service providers to target them as data subjects. As described, given the diversity of data-based utilisation contexts, this address necessarily takes on different forms. It can take place in physical real space or in virtual space, but also as a mixed form in digitalised real space. It can take place through an immediate translation of collected data (sentient city) or it can refer to data compiled over a larger period in advance. Mixed forms are also common here, with direct customer targeting (e.g., personalised display advertising in a store) taking place based on collected data (for example, via the customer card). What the various ways of approaching customers with digitally stored knowledge share in common is that they convert individuals into potentially commodifiable characteristics in advance. It is not about capturing the complex personality of a target person, but primarily those characteristics that are relevant for a potential business. Instead of assessing the borrower´s creditworthiness based on their personal appearance, the lender will direct his offers toward the exact items of available data sets. Online retailers primarily present products in line with the customer's identified preferences, with no interest in other personal attributes, and advertisers target their offers more closely to the user's specific preferences than to general trends.

For a deeper analysis, we will first look at the virtual translation of data into cyberspace offerings, where targeting based on user characteristics can take place immediately and in real time.

Addresses taking place via ad servers assign websites with different page content to different target groups in order to display the appropriate banner. What is appropriate is decided based on one or more characteristics sifted from the pool of all persons, or their IP address. Accordingly, the advertisement for a golf vacation for the target group of men over 50 will appear on a more sophisticated news site, while the ad for a soft drink for children is integrated into an online game. The time of the display or the regional reference can also be adjusted individually. More recently, semantic targeting, which addresses the user only in relevant editorial contexts, has been used for further individualisation. It has also become possible to evaluate developments within the pages, such as comments from readers (Aaltonen and Tempini 2014, p. 103ff).

While the focus here is on the recorded individual features or feature bundles and types that lead to a specified form of address, in other virtual spaces the derivation from the observed behaviour of many other users plays a central role for individualised targeting: If it can be demonstrated for many consumers that demand of X is followed by demand of Y, predictions can also be made for other users with corresponding behaviour patterns. This collaborative filtering allows conclusions to be drawn for sales items regarding the likelihood of a positive response. The “You might also be interested in this” notice is not only a central component of the business models of Amazon, Netflix or Google News, but is ubiquitous on the web.

Social networks like Facebook sort the wealth of user information they collect into databases that advertisers around the globe can access. Target groups can be defined by various criteria (e.g., location, age, gender, language, interests and behaviour) and addressed accordingly on their Facebook page. For companies, tailored target groups can be sifted out using customer lists, website traffic or app activity. The ads can then be embedded in different ways (Desktop News Feed, Mobile News Feed, Right Column and Audience Network) and timed as desired. In this way, users are confronted at specific points in time on the desktop computer or mobile with messages that coincide with the range of topics relevant to them and with their specific preferences.

In all three cases, cyberspace is customised to the individual user, with the advertising company determining the relevant characteristics for addressing the respective customer. The advantages of cyberspace that seem to lie in its wider range of offers in comparison with the more segmented and restricted real space are partly lost as a result of the ubiquity of personalised, commercially exploitable content. Although this is by no means true for all content, this personalisation is present in various forms on most pages and is not always recognisable as such. This is particularly the case with native advertising, where messages are unobtrusively integrated into texts, videos, blogs and feeds. Depending on the channel, the layout of the message can be precisely adapted to the environment. The boundaries between the commercial and the public, the individually directed and the generally relevant are blurred by creating a feeling of familiarity and credibility (Schauster et al. 2016).

Turning from cyberspace to real space, the latter can also be personalised in the interplay with virtual data storages. The presentation of the various recording methods and the sketching of the networked city made it clear already that a strict separation of the two spheres was not sensible from the point of view of data generation. Such a dualism would ignore the numerous links between digital and analog sources, allowing for mutual enrichment precisely in their interplay. At the same time, it is often through the linking of the two spaces that the data collected actually become valuable. For example, with location-based services, the user's physical locations and routes are a valuable data add-on to the provider's database. The additional possibilities for collecting knowledge about customers even in sensitive areas or for predicting individual behaviour patterns can be related to this dataset (see also the examples of Cheung 2014; Michael and Michael 2011; Michael and Clarke 2013). Conversely, are taken out of virtual space and its processing capacities and put into value in real space. Many areas of marketing and personal services are becoming much more efficient with the influx of location-based data. Thanks to mobile phone data or wearables, the virtual user profile derived from clicks and entries is enhanced through real-space tracking, making it much more precise. Someone who frequently visits the local home improvement stores appears in the database as a potential loan borrower or reader of online articles related to home improvement. People who live in expensive neighbourhoods or frequently travel there on business might be interested in high-priced products.

The items about each individual, whether obtained virtually or derived from real space, create a corridor of personal life content and preferences, and addressing them within this corridor will lead to a successful transaction. From the supplier's perspective, preferences and opportunities have the highest chance of being directly converted into products and services when there is a correspondence between the two. When patterns of consumption are known, providers will direct their attention at those individuals who have the greatest potential interest in the product, message or service. Consequently, a highly filtered selection with a specific content reaches the respective target person, while unfamiliar things are withheld.

Here, beyond the generation of data, location-based services also serve to optimise the business environment. As locations and location-based activities are known and transmitted to data collectors in real time, on-site targeting can be realised based on the dispositions of the target person. Geofencing describes the automated triggering of an action in a predefined spatial section: The right products and services appear where the person is at a given moment, and they are offered in a way that appeals to them.

If real space can be restructured based on commercial criteria analogous to the further expansion of the internet, it will also end up offering a reproduction of the personality traits of an individual that are already known. Augmented reality applications have already taken such a step. They create hybrid, personalised worlds by presenting individually relevant content (Gong et al. 2017).

The further modern network technologies penetrate the everyday life of individuals; the more likely it becomes that the latter will be confronted with set pieces of and information about themselves. What is crucial is that this confrontation happens imperceptibly and that the time cannot be assessed. Furthermore, it tends to take place everywhere, as personalised data have long since left the virtual realm and can have an effect in both spheres. As is true for the workings of algorithms of large internet corporations, it is equally impossible to reconstruct why and based on what internet activities certain brochures end up in one´s mailbox, why an ad appears on the smartphone at a certain point in the city, and to what extent an insurance policy taken out online is more or less expensive depending on the residential address.

4.4 Mirrored Spaces

“A world constructed from the familiar is a world in which there's nothing to learn. If personalization is too acute, it could prevent us from coming into contact with the mind-blowing, preconception-shattering experiences and ideas that change how we think about the world and ourselves”. With this statement, Eli Pariser (2011, p. 9) outlined early on the constricting effect of personalised data utilisation that constantly provides internet users with a restricted and preselected range of information. Pariser popularised the term “filter bubble”, which describes a personalised cyberspace that translates algorithmically collected data about each individual into information and offers that are as close and as intuitive to the user as possible. Based on identified preferences, people can be connected to information they are likely to be interested in, lured with incentives that have worked before and confronted with products that they will likely fancy. As these virtual offerings become more accessible than others, the result is a personalised stream of content offering users in the filter bubble fewer and fewer alternatives and choices. The filter bubble or echo chamber is like a mirrored room that emphasises the familiar and blocks out foreign elements. The mirror analogy expresses the self-referentiality of the viewer that is enforced in cyberspace via the externally controlled translation of the personally relevant. The environment becomes an image of one's own. Since only the one available version of the internet is being used, we do not notice what we are missing. In fact, however, in algorithmically curated information environments, the hit lists for search queries differ just as much as the content compilation on the entry page of the social network or the advertising offered next to an article.

Against this backdrop, Pariser and many others have repeatedly raised the question what such narrowed, fragmented world views mean for social cohesion and the political system when there are no collectively relevant topics left and the struggle for solutions in democratic discourse takes place without shared basic information (Bozdag and van den Hoven 2015; Spohr 2017). The public sphere is being replaced by closed communities whose members systematically isolate themselves from deviating realities, mutually reassuring each other within their filter bubbles. In this “balkanisation of cyberspace” then lies the danger of a multiple division of society, which in the long term will no longer be able to agree on what is relevant or to communicate at all about the same things. With respect to this challenge, the empirical findings on the filter bubble are mixed so far: While the study by Zuiderveen Borgesius et al. (2016) highlights the growing relevance of the debate about filter bubbles, it concludes that at this point there is no empirical evidence to justify a strong concern about filter bubbles (Ibid., p. 10). The large-scale study of around ten million U.S. Facebook users conducted by Bakshy et al. (2015) also indicates that it was not so much algorithmic ranking as personal choices that led to ideological demarcation. By contrast, other scholars have been able to identify ideological and partisan bubbles in Twitter discussions (Barberá et al. 2015; Boutyline and Willer 2017; Colleoni et al. 2014) and Facebook groups (Jacobson et al. 2016). Dylko et al. (2017) have demonstrated that the algorithmic preselection taking place on social network services like Facebook does favour the interaction with contents that confirm existing opinions and that the level of cognitive dissonance can be reduced effectively with algorithms operating in the background. Schmidt et al. (2017) also clearly confirm the thesis that news is consumed selectively. They analysed how often, how long and with whom 376 million Facebook users shared stories from 920 media outlets around the world between 2010 and 2015 and found out that most Facebook users only interact with a few news sources and prefer to share news from these portals with friends. The content and orientation of these portals are often the community that participates. In spite of the extreme diversity of offerings that exists in cyberspace, users move in self-selected clusters: “Despite the wide availability of content and heterogeneous narratives, there is major segregation and growing polarisation in online news consumption” (Schmidt et al. 2017, p. 3038).

Psychologically, bubbles can be explained as the result of our avoidance of cognitive dissonance, i.e., the desire of humans to reconcile their different cognitions, such as perceptions, thoughts or attitudes, without contradictions. But it is only cyberspace and its algorithmically tailored offerings that allow people to banish other realities in a consistent and systematic way. Defining oneself via the clicks made leads into a virtual environment which in turn encourages recursive use. The tendency to selectively choose news that reflects one's own opinion and attitude (“selective exposure”) takes on a double direction in the internet age: It is not just the user picking out the information, but the information is also precisely tailored to each individual user. As outlined above, this has become possible and commonplace on the basis of comprehensive tracking and targeting.

While it is becoming increasingly obvious that information bubbles are having an impact in cyberspace, and issues associated with it (such as fake news, election manipulation, and the loss of social cohesion) are the topic of numerous feature articles, confirming Eli Pariser, little has been said on the socio-economic implications of the filter bubble. For an evaluation of the new options for action in cyberspace, however, its inherent self-referentiality is extremely relevant: The spaces provided by the private sector convert the user into personalised offers and information. The structures in cyberspace, ostensibly free of charge, are paid for click by click with the user´s own transparency, bringing that which can be commercialised ever closer to the individual. In the mirrored spaces of cyberspace, users primarily see themselves. Increasingly, everybody is confronted with offers, information and contacts that resemble their known preferences.

There are various reasons why the known and familiar reach the user more easily and are embedded as a fixed rule of assignment in the algorithmically organised space: In connection with the digital habitus, it has already been emphasised how difficult it is to appropriate digital content outside one´s own milieu. Entering environments on the internet that are unfamiliar and hard to predict requires more effort than staying in one´s personal comfort zone. As outlined, however, putting too much emphasis on the tendency to stick to the familiar would mean ignoring the opportunities for capital acquisition. Access to resources in cyberspace is, in principle, open—for those who are willing, motivated and receive guidance, for example, from an educational institution. With increasing transparency of the customer, however, this access is obscured in that cyberspace, as a sequence of virtual spaces, does not keep its rigid structure, but constantly offers itself to the user anew in a changed form. Because of cyberspace´s ability to turn the socially familiar into its content, it becomes more likely that users will persist in this space. In the mirrored room, everyone is surrounded by the known and familiar.

Numerous studies demonstrate a higher effectiveness of those messages that are aligned with the personality traits of the target person (for an overview, see Hirsh et al. 2012). Since these messages in cyberspace are aligned to the user's activities, the requested pages can be equipped with specific stimuli, which in turn make the next step in virtual space more likely. Priming via image, sound or speech leads to the activation of implicit memory content, which specifically influences feelings and subsequent behaviour. In the same way, the offerings of cyberspace can be enhanced with framing effects, where variations of the same message can be used to influence the user's behaviour in different ways. In this way, the virtual context that reflects the user's specific interests can be selectively furnished with reinforcing content (Wu and Cheng 2011).

Users´ tendency to view content that is familiar can also be explained psychologically by the mere exposure effect, which means that people will develop a preference for something as a result of repeated exposure. This effect can be used to explain the demand for certain content by the conditioning of the user, whose choices are influenced by past experiences. Messages can then be formulated accordingly. For instance, repeating an advertisement can help ensure that the consumer will perceive it positively. The more an individual likes a specific stimulus and the more they associate it with pleasant feelings, the better this works (Felser 2015, p. 81ff).

In terms of opportunities for social advancement, the personalised structure of cyberspace implies significant limitations. The opportunity to acquire cultural and social capital has been related to overcoming site effects and being able to participate. In a hierarchically constituted real space, there are obstacles such as physical inaccessibility and distance, the lack of visibility and the social distance to the target group, which cyberspace seems to eliminate in part. But on closer inspection, the available offerings of the digitally mediated supply structure are now transformed into personalised offerings that reflexively use that which exists already: Accessing a song on Spotify or a video on YouTube will result in suggestions that are very similar to what has been requested before. Ordering clothes on Amazon will lead to new offers of the same style. Once interest in a product has been expressed, it is used to shape the virtual environment in different forms, and the display's automatic word recognition system automatically reformulates past entries. The opportunity to obtain resources relevant for upward mobility is therefore impeded structurally. In the seemingly hierarchy-free environment of cyberspace, what has to be overcome is no longer just the filter of relevance inherent in the habitus. Rather, it is the ability of broad areas of cyberspace to exactly adapt to this habitus, to mirror it, displacing alternative as well as instructive contexts of experience and education. In the process of digital socialisation, serendipity, the accidental discovery of something not sought after, something enriching, becomes less likely. Whether information on cultural and leisure events, political discourse, relevant news or the latest consumer trends: All of them are ordered in cyberspace according to the preferences stored in data sets. They refer to and revisit the contextual and compositional factors in the socialisation process.

This recursivity characterises the acquisition of social capital as well: Right after logging on to Facebook for the first time, the user is assigned to numerous people they have met in the past at school, at their place of residence or at their job training. In the mirrored space of cyberspace, the opportunity to cultivate social capital emphasised in Chapter 3.2 (Ellison et al. 2011; Steinfield et al. 2008, among others) is impeded by an algorithmically driven selection that largely refers to the same social context that already shaped the user in the past. The fields of interest to be assigned to each individual and the locations communicated algorithmically in the network, which could serve to initiate further contacts, also frequently reflect user properties. Without the need to log in, the IP address already reveals the current location of the user and provides specific information. The map service settings are location-based, selected shopping suggestions and events refer in particular to the known real space, and travel recommendations are in line with interests expressed in the past. For the segments of the population lacking in capital, their personalised cyberspace will not contain ads for the theatre, or the public square mentioned earlier or references to exclusive vacation spots for the rich, despite the fact that there are no actual physical distances or social distinction processes at work. Virtual spaces of opportunity simply become less visible for the individual. Unfortunately, the isolated use of cyberspace always ensures that actions in the virtual cannot be perceived in alignment with the outside world. The danger is that cyberspace is then perceived not as a variant of reality, but as absolute reality.

However, the decreasing probability of being able to perceive resources that are foreign to one's milieu or habitus in mirrored spaces is only one—albeit quite central—consequence of the economic exploitation of personal data. Connecting some of the business areas outlined above with the individual's options for action reveals further limitations: Depending on the utilisation context and the corresponding coding of the framework of action, the externally controlled regimentation must present itself differently based on collected data. While online ads or suggested news content merely provide an effective incentive that the user can decide to skip, the data-based assignment of a specific risk score means that “unsuitable” offers—for example, in online insurance, loans or health insurance premiums—are systematically withheld from the customer and “suitable” offers are suggested instead. The decision on what is appropriate for the user and what will reach them is based on the relevance of the available data as defined by the respective business field. Such forms of “social sorting” were compiled early on by David Lyon in his Surveillance Studies (Lyon 2003). Lyon focuses on the public and private-sector forms of group and personal data processing, describing the enabling or disabling of actions depending on the selection characteristic (gender, ethnicity, occupation, social status, etc.). Again, the code functions here as an “invisible door” that determines who has access to experiences, events and information and which parts of the population can interact and participate (Ibid., p. 13, p. 23ff). The opportunity to explore cyberspace independent of the available capital is clearly revealed as an illusion. All markers of inequality such as income, education or social status can be systematically used to constitute cyberspace, resulting in spaces that differ in how they are equipped and in how they can be accessed. The above-mentioned inaccessibility in real space due to barriers or distances corresponds to a virtual space that regulates access to resources on a user-specific basis. In addition, contents of cyberspace can also be completely faded out, eliminating the ability to perceive other options.

Finally, algorithmic decision procedures also influence the third obstacle to capital acquisition described above, which is social distance. To begin with, it was stressed how segments of the population lacking in capital could strategically conceal the habitus, temporarily relieving themselves from having to know and apply the rules of distinction. However, as has been outlined, a consistent valorisation of personal data takes effect at an earlier stage, as the virtual environment is already pre-structured by the registered dispositions of the user. The habitus is already known and virtual content refers to its different aspects to varying degrees. The opportunity to take advantage of resources that facilitate social advancement is predefined in an unequal manner.

Moreover, it is evident that the available information about a person moves back and forth between cyberspace and physical real space, and in both spaces this information can be used either to create an incentive or to impose rules. Things like individual reliability scores, a person´s health or special consumer preferences not only result in a personalised address in cyberspace, but also have a direct influence on the everyday life of each individual in real space and their options for action and perception.

4.5 Interim Conclusion: Digital Self-Confrontation

The use of cyberspace, the internet and other networked objects is accompanied by the accumulation of personalised data. While the internet is praised as a medium that can transform the boundaries of individual experience, turning spatial mobility into social mobility, making actual use of this potential requires the individual to engage in constant self-measuring. New products and technologies such as the Internet of Things, cloud services, sensors and countless smartphone applications represent the foundations of such measuring. While it can be countered with knowledge and technical effort, avoiding it completely is almost impossible.

While the permanent provision of data within these structural specifications has been analysed from the perspective of surveillance, interpreting the consequences of a “culture of self-exposure” as a voluntary disciplinary process (Bauman and Lyon 2013; Lyon 2018; Michael and Michael 2010; Zuboff 2019, among others), the recursive aspects of a digitally influenced socialisation call for a different focus: There are strong indications that the self-referential use of digital technologies is closely related to the social question. Across all basic functions of existence, many new options for action can be identified in cyberspace and in networked real space, all primarily addressing the familiar. What clearly stands out in such a mirrored space is that customers are increasingly addressed with products and services which they are known to like and be familiar with and which fit into their living environment and budget. It is equally obvious that each individual is left in the place where they are with their specific habitus.

Cyberspace, which could initially be described as an arrangement that can be accessed easily and experienced anonymously, hardly has any isolated niches left that can be explored without being tracked. Based on the evaluation of personal data, users enter each area with specific indexes that assess them and that can be exchanged within the virtual sphere. Such an exchange, which is able to personalise spaces in advance, leaves users with limited options, which means that the resources available in cyberspace are also structured by accessibility and perception.

What each individual obtains from cyberspace or through interconnected devices ultimately impacts everyday experiences in the real world, whether it be internet acquaintances, vacation recommendations, shopping addresses or leisure tips. In addition, transmitter–receiver systems are active in smart real space, which can selectively display, hide or regulate, again resulting in a recursive acquisition, as has been illustrated by numerous examples. The field of social contacts, preferred places or shops then provide an opportunity structure with numerous feedback or place effects all standing in the way of cross-milieu acquisition. Beyond the filtered use of services in cyberspace (contacts, advertising, financing offers), it is also filtered offerings that have an impact on the options for action and perception in real space—and which inevitably influence the possibilities for the constitution of space. The process of “mirroring” takes place across spaces.

Up to this point, the demonstrated relativisation of opportunities of capital acquisition can serve as a new starting point to reflect on digital offerings in relation to education. Beyond the inequalities of digital access and use, the main point is that the opportunity to obtain resources digitally can be predefined by all the user's characteristics. It is therefore not only about having the skills to use a computer for education and resource acquisition, but also about knowing under which conditions these resources can reach a user at all.

For the study´s overarching objective, which is the examination of current stratification processes in the context of digitalisation, the focus on cyberspace and the processes in real space influenced by it is not yet sufficient. So far, cyberspace has been conceptualised as a strategic option of choice, and this includes the possibility of consistently choosing to turn away from the effects of datasets. According to this assumption, the many links created by the use of smartphones or certain gadgets can be severed by digital abstinence. This would leave mainly automated capture systems such as cameras, RFID technologies or intelligent networked devices and their increasing prevalence as mediators of possibilities or restrictions.

As a matter of fact, the data-driven economy is encroaching on everyday life regardless of an individual's affinity for computer use, ownership of a smartphone or the presence of sender-receiver systems. As shown by the multiple forms of data acquisition and exchange, personal information from numerous channels flows together without immediately being commodified in a specific context. Extensive personal data about each individual are stockpiled in databases that are generally invisible and sold by global traders. Regardless of the context in which they are collected, these data facilitate personalised offerings for a wide variety of utilisation contexts. Information on individuals has thus become completely disconnected from the data collection infrastructure, and for a fee it can be monetised offline at any time.

This billion-dollar data trading market should also be highly relevant in the context of the opportunity of capital acquisition. If virtual offerings can potentially define individual scopes of possibility, then this needs to be considered also for traded, decontextualised data. How does their use affect the question of situational (dis-)advantage in the urban space with its unequal opportunities? To find an answer, we must take a further step and transfer the data merchants’ products back to individual cities.