25-4-2019
Smart cities more secure
but less transparent?
How two Dutch smart security initiatives account
for the public value of transparency
Master’s programme: Crisis & Security Management
Student: Laurens van der Aa
Student number: s1615742
Date of admission: 25-04-2020
Subject: Smart cities more secure but less transparent?
Word count: 17995
Thesis Supervisor: Dr.ir. Vlad Niculescu-Dincă
2
Acknowledgement
First off all, I would like to express my special thanks of gratitude to my supervisor Dr.ir. Vlad Niculescu-Dincă for his guidance during the process.
I want to extend my gratidude to the second reader dr. Tommy van Steen for his feedback at the start and final stage of the process.
The completion of this undertaking would not have been possible without the participation of the five respondents. Their contributions are very much appreciated.
3
Abstract
At this moment, not much research has been done into Dutch smart city initiatives since this is a somewhat new phenomenon. The research that exists mostly covers the effects on privacy but other public values are still underexposed. The purpose of this study is, therefore, to get a better insight into the way Dutch smart security initiatives in public space impact transparency. The smart security initiatives studied in this thesis are projects that focus on improving security in the public space using algorithms that analyse data which is produced by sensors and cameras. Smart security can be seen as part of the larger domain of smart cities.
The theoretical framework used in this study is based on Actor-Network-Theory as this provides the symmetrical role interpretation of human and non-human nodes necessary for smart city research. For this study, the city of Eindhoven and the city of The Hague have been selected, as they are currently developing the most security-related use cases relative to other cities.
The results suggest that although the two studied initiatives are still in their infancy, considerable work is being done to ensure transparency. In both cities, methods to create awareness about the presence of sensors are devised. Eindhoven has taken the first steps to develop a policy concerning data transparency with their open data guidelines but this has created some new challenges. Lastly though, while the importance of algorithmic transparency was acknowledged in both cities, no exact procedures or regulation were discovered and especially concerning machine learning much is still unclear.
4
Table of contents
1 Introduction ... 6 2 Theoretical framework ... 9 2.1 Smart cities ... 9 2.1.1 Big data ... 9 2.1.2 Algorithms ... 10 2.1.3 Network construction ... 112.2 Different kind of security networks ... 12
2.3 Actor-Network Theory ... 13
2.4 Transparency ... 15
2.5 Transparency in smart cities ... 16
2.5.1 Sensor awareness ... 17
2.5.2 In- & output... 18
2.5.3 Back-end ... 19 3 Methodology ... 23 3.1 Research design ... 23 3.2 Case selection ... 24 3.2.1 Case description ... 24 3.2.2 Justification ... 25 3.3 Data collection... 27 3.4 Operationalisation ... 28 3.5 Limitations ... 32 3.5.1 Validity ... 32 3.5.2 Reliability ... 33 4 Analysis... 34 4.1 Sensor awareness... 34
4.2 In- & output ... 37
4.2.1 Ownership ... 37
4.2.2 Openness ... 38
4.3 Back-end... 42
4.4 Organisational openness ... 49
5
4.4.2 Private sector involvement ... 50
4.4.3 Purpose formulation ... 52
5 Conclusion and discussion ... 55
6
1 Introduction
Digitalisation is reaching the very capillaries of our society. As we speak, the public space is changing and more and more ordinary analogue objects such as lampposts, roads and waste containers are being equipped with sensors to make them ‘smart’. In these so-called ‘smart cities’, innovative information technologies are thus built in the very fabric of urban environments. These ‘pervasive and ubiquitous digitally instrumented devices (…) are used to monitor, manage and regulate city flows and processes, often in real-time ‘(Kitchin, 2014b, p. 2).
The monitoring devices produce vast amounts of data, also known as “big data”. Advocates believe these systems and big data will help municipalities to address issues of city security and other domains such as mobility and sustainability. A promising and hopeful thought in a time of urbanisation and population growth that can cause issues for municipalities in exactly those domains (Kitchin & Dodge, 2019). No wonder that city managers all over the world are eager to implement smart city pilots and initiatives. Take, for instance, the well-known case of the city of Barcelona. In 2013 the City Council released a strategy to promote economic growth and the well-being of citizens by using new technologies to digitise the urban system and the governance of city (Veselitskaya et al., 2019).
Adversaries, however, warn to beware of promises that cannot be kept and say that the feasibility of these projects is still unclear or that the potential should not be exaggerated. The impact on public values, on the other hand, should not be underestimated, they say. Because these ‘sensing’ devices are monitoring the public space – a place that is open and freely accessible to everyone and might, therefore, even belong to everyone – the impact of the use of these devices is not limited to a small group. An interesting point was raised by Zedner (2003), talking about the pursuit of security in public space. It is often stated, she said, that there is no freedom without security, but the pursuit of security, on the other hand, can lead to erosion of civil liberties as well. We are accustomed to the suspension of normal democratic freedoms when we enter zones of high security such as airports. The pursuit of security is prioritised over other liberties and if we don’t want to accede to the intrusions upon these liberties, we can only choose to leave the premises (Zedner, 2003). Leaving the premises, however, is no option when whole cities are turning into zones of high security. Then acceding to these intrusions upon civil liberties becomes an obligation instead of free choice. Zedner was talking about the effect of the pursuit of security on civil liberties but the very same is true for public values. The pervasiveness of these new technologies in the
7 public sphere is, therefore, something to watch critically. Due to the novelty of the academic field, research into smart cities is still scarce. The research that is around though mostly covers conceptualisation difficulties (see, e.g., Hollands, 2008) and a considerable part of the literature is concerned with the potentials and opportunities (as discussed by Clark, 2020). Even if the impact on public values are covered, the focus is predominantly on the issue of privacy; the impact on other values is still underexposed (see, e.g., Kitchin, 2014b; Kitchin & Dodge, 2019; Van Zoonen, 2016).
Plans for smart initiatives in the public space of the city of Toronto raised concerns for multiple public values. The involvement of Google sister company’s Sidewalk Lab in the collection of data was the cause for concerns about privacy and fears for an overall lack of transparency (Cecco, 2019). Because several cities in the Netherlands started comparable smart city initiatives, it raises similar public value related questions about Dutch projects. Already several scholars addressed the impact smart cities on privacy. The literature on the effects on the value of transparency, however, is less extensive but not less important. Transparency means that people have information to question whether they are treated fairly (Van Barneveld et al., 2018) and because of the growing use of these devices in public space, increasingly important.
The use of smart artefacts such as sensors and cameras has only recently found its way to the urban management domain as well as advanced software, processing data in the back-end. The effects of new methods of data analysis such as the use of machine learning, are still unclear but could impact the traceability of decisions and in turn democratic accountability (Kroll et al., 2017). It requires re-establishing of responsibilities where new public interests arise and existing interests may no longer be properly respected (Geonovum, 2018). Concerning the latter, due to the novelty of all these artefacts, it is good to look into the question whether these new developments impact or still properly respect the existing interest of transparency. Because the public space belongs to everyone and they have the right to know what happens there, argues Geonovum (2018).
Thus far, these developments in Dutch cities have not been studied extensively and the goal of this research is, therefore, twofold. On the one hand, this research wants to extend the academic focus by discussing the effects on the public value of transparency. On the other hand, it wants to add to the commencing body of literature on Dutch smart initiatives. Following this line of thought, the main research question of this thesis is:
8
RQ: How do Dutch smart city security projects in public space account for the value of transparency?
In order to answer the main research question, it is necessary to answer the following sub-questions first:
SQ1: What is meant when talking about smart cities?
SQ2: What is meant with the term transparency?
SQ3: What does transparency mean in relation to smart city projects?
SQ4: How is transparency affected by the implementation of these Dutch initiatives?
The rest of this paper is divided into five further sections. The second section gives an overview of the current academic field and presents the theoretical framework on which this research is based. This chapter elaborates on the involved concepts and the theory behind their relationship. The following section describes the methodology, by discussing the design of this research, the way of data collection and how reliability and validity issues are being addressed. The fourth section is reserved for the analysis and presents the research findings. The conclusions are drawn in the fifth section, which is followed by a discussion of the results in the sixth and final chapter.
9
2 Theoretical framework
Prior to this research, it was essential to get a better picture of the literature on transparency and answer the first sub-question What is meant when talking about smart cities?. The first step was to find out what the public administration literature tells us.
2.1 Smart cities
Although as well the academic field of smart city research as smart city developments in practice is still in its infancy, a multitude of conceptual understandings of the term exists. In general, we can distinguish two main perceptions in the literature, says Kitchin (2014b). The first notion of smart cities is used to refer more broadly to a knowledge economy on a city-region level. Smart cities, from this point of view, are cities where the governance and economy are driven by innovation, creativity and entrepreneurship.
The second interpretation, he says, refers to the increasing use of digital instruments in the urban space, which he denotes as ‘everyware’. Everyware can be understood as the technological devices built in the fabric of the urban environment, mentioned earlier, used for regulating a city from a technological perspective. Artefacts such as camera, sensor networks and so on monitor and regulate city flows and processes and help to get a better understanding of the city to enhance efficiency, liveability and sustainability (Kitchin, 2014b).
Although both visions are fairly different, they agree upon the importance of information and communication technologies and the data these technologies produce. According to Kitchin (2014b), data is oftentimes seen to provide neutral and objective measures and, thus, speaking the inherent truth. Analysis of this data is then a way to achieve evidence-informed decisions and reach technocratic governance. Whether data really is neutral and objective will be discussed later in this thesis, for now, the recurring link in the literature between data and smart cities is relevant. Many scholars describe the importance of data in smart city decision-making (see f.e. Batty, 2013; Hashem et al., 2016; Kitchin, 2014b), but what is meant with this term data and what is information?
2.1.1 Big data
The term ‘data’ has been widely used in a variety of domains and is variously understood, which caused the term to lose some of its clarity. Data and information are often used interchangeably,
10 however, according to the literature data are not the same as information. As defined by Rob Kitchin (2014a), “data can be understood as the raw material produced by abstracting the world into categories, measures and other representational forms (…) that constitute the building blocks from which information and knowledge are created” (p. 1). So data (for clarity reasons oftentimes called raw data) precedes information and is the resource out of which information can be created through a process of distillation. This process entails processing, analysing and interpreting and adds value and meaning by revealing relationships.
Within the literature, the difference between information and data is somewhat contested. Some see information for instance as the accumulation of data, others see it as data plus meaning after processing and analysis (Kitchin, 2014a). Because the terms are used interchangeably, it is not always clear what is meant when either one is used. Where possible, in this thesis a clear distinction will be made between the two; using data for the raw material after measuring and information for the processed and analysed data with added value and meaning.
The idea is that the right processing and analysis of data from sensors in the city creates information, that helps decision-makers such as the city counsellors or police agents make right and well-founded decisions. Data are, thus, playing an increasingly important role in municipal decision-making and this prompts the question of whether transparency is needed to check on legitimacy. Furthermore, the fact that the data originate from the public space and citizens the objects of the analysis prompts questions concerning data ownership and openness.
Although definitions of big data vary in the literature, most scholars agree that big data is high in volume, variety and velocity (referring to the rate of generation). Due to the pervasiveness of the everyware that is monitoring the city and collecting data, the amount of data is increasing rapidly and the concept of big data is certainly applicable. It is, therefore, becoming increasingly difficult for human operators to process, analyse and create useful information out of the raw data using traditional data management technologies (Gandomi & Haider, 2015). Therefore, computer codes, known as algorithms, automating decision-making are increasingly used.
2.1.2 Algorithms
Nowadays, software algorithms are increasingly automating the processing and analysis of data and therefore they play an essential role in the creation of information. Cambridge dictionary defines an algorithm as ‘a set of mathematical instructions or rules that, especially if given to a
11 computer, will help to calculate an answer to a problem’ (Cambridge English Dictionary, n.d.). According to Diakopoulos (2014), we can distinguish four types of decisions that algorithms make on the basis of these rules regarding data, which include prioritization, classification, association and filtering. Chaining these decisions leads to higher-level decisions and complicates the inner working. The rules are written down by engineers or based on found patterns in large amounts of data. The latter form whereby humans orchestrate a computerized rule-creation process rather than construct the rules themselves is growing in popularity and is known as machine learning. The discovering patterns in the vast amounts of data are used as the basis for the algorithmic rules. Hence, the rules are quite literally learned and induced by historical examples.
2.1.3 Network construction
Smart city networks usually also consist of other than technological nodes. Roughly three different parties involved in the smart city monitoring can be distinguished. (I) there are the public stakeholders such as municipalities that monitor the city for the performing of public duties or to make available to the public, (II) private parties or individual people that mine data for their own business operation or research and (III) commercial service providers that mine data for the benefit of their customers (Van Barneveld et al., 2018).
These three different stakeholders can form the following five archetypical partnerships. The first one is the type in which municipalities collect data independently of other organisations. Cities start their own data collecting initiative with sensors that are mounted on municipally-owned objects (or in some cases on privately owned objects). Examples of this type, are sensors mounted next to traffic lights and detection loops in the road surface (Van Barneveld et al., 2018).
Second, a municipality makes use of a service provided by a commercial service provider. The commercial party uses sensors which are mounted on municipally-owned objects (or in some cases owned by other private parties). Characteristic of this archetype is the role of the municipality as the client and its relationship with the contractor. It means that all ideals, interests and values such as transparency should be laid down contractually. An example of this partnership is sensors mounted on privately owned shops that measure WIFI signals and report to the municipality the number of passers-by (Van Barneveld et al., 2018).
12 The third archetype is the public-private partnership. This partnership exists in several forms but recently one specific type of this kind of partnership started to gain traction. This collaboration does no longer solely involve the public and private sector but also academia plays a vital role as the third actor. In this model which is
called Triple Helix, the innovation of new products is done by academia, the role of government is to formulate policies and (financially) support development and the private sector is present to develop and market these products as well as doing process innovations (Arnkil et al., 2010). Due to the novelty of the idea of smart cities, technological artefacts and software code are still in development as
well as their sociological implications. In this innovative domain, a huge role is reserved for academia. For this reason, the triple helix partnership is often seen in this field. Some scholars argue that the public should be included in these partnerships since they are the ones confronted with the new technologies (Arnkil et al., 2010). For this reason, an iteration on the triple helix collaboration exists, that adds a fourth pillar. Notable in these collaborations is decision-making on topics with conflicting interests (Van Barneveld et al., 2018).
The fourth type is that of private stakeholders that make use of the service of a commercial party. In this case sensors of the commercial party are often mounted on privately owned objects of the client. Take for example the WIFI sensors monitors costumers and reports to the shop owner the duration of their visit to the store (Van Barneveld et al., 2018).
The last archetype that could be distinguished is private parties independently collecting data with sensors mounted on their own objects. An example is digital advertising columns with cameras that count passers-by and monitor walkways (Van Barneveld et al., 2018).
2.2 Different kind of security networks
Smart city projects concerned with security can be seen as security networks. Following the typology of Whelan & Dupont (2017), we find that these networks can be qualified as subnational security networks. Typically these networks are goal-oriented and involve both public and private
13 security actors. Communication between the actors is usually of an informal nature in structured meetings or virtual systems.
Although ties via virtual systems are mentioned the typology of Whelan and Dupont (2017) falls short in expressing the importance of technology in smart city projects. Technological components such as everyware as well as algorithms processing big data are vital elements in smart cities. After doing a review of the literature on smart cities, Dameri (2013) came to the same conclusion. She found four frequently mentioned components, of which technology was the most considered one. Each technology could be used to improve urban quality and those technological solutions work alongside human operators such as for example police agents.
In The Surveillant Assemblage, Haggerty & Ericson (2000) look for a suitable concept to describe such a network with human and technological nodes. According to them, popular literary works in the field such as George Orwell’s 1984 and Michel Foucault’s Discipline and Punish, don’t provide a sufficient term. The former because contemporary smart projects surpass even his dystopian view; the latter because it totally omits technological developments. Since the term ‘assemblage’, used by Deleuze and Guattari, introduces the notion of multiplicity and heterogeneity, this concept proves to be better suited for this research. ‘Assemblages’ consist of a ‘multiplicity of heterogeneous objects, whose unity comes solely from the fact that these items function together, that they “work” together as a functional entity’ (in Haggerty & Ericson, 2000, p. 608).
The beforementioned importance of technology in smart city assemblages calls for a theory that assumes symmetrical treatment of human and non-human nodes. The academic literature of scholars such as Whelan and Dupont did not sufficiently account for this symmetrical treatment so a different kind of theory needed to be found.
2.3 Actor-Network Theory
One influential theory in dealing with this kind of networks is the Actor-Network Theory (ANT). The theory focusses on the relationship and negotiations between the nodes in the network. Because the word actors is usually used for human stakeholders, in ANT the word actant is chosen to denote a human or non-human node – which can be technological artefacts but also animals or bacteria. ANT argues that prejudging the relative power or influence on the basis of the nature of
14 the actant is undesirable and it, therefore, proposes symmetrical treatment of humans and non-human actants (Niculescu Dinca, 2016).
In ANT, technology and society are closely related to one another on a symmetrical basis. Where the theory of technological determinism assumes that technology determines society’s structures and values, ANT argues that this a reciprocal process. The introduction of new technologies changes society but existing norms and values in society change the way we use technologies and embrace or just reject them (Law, 1992). In addition, ANT also differs from technological determinism in that it doesn’t reduce technologies to mere neutral tools. Neutrality in technologies would stem from the idea that they are value-free objects and interaction with them would not change the perception of the surroundings. ANT, however, argues that this is not true and that they actively mediate perceptions, even beyond their intended function (Niculescu Dinca, 2016).
For assessing the way technologies relate to human behaviour, Latour and Akrich proposed the notion of scripts. Technologies direct the behaviour of the user via its form or instructions, in the same way, theatre scripts influence the behaviour of an actor. They steer in a certain direction but do not determine the exact performance. These instruction manuals together with certain values are already built in the artefact during the design stage with every choice the designer makes. Later, these choices make an artefact prescriptive causing the artefact to guide and alter the behaviour of its user. It is even possible for an artefact to be prescriptive in an unforeseen way with negative and unintended consequences (Fallan, 2008; Niculescu Dinca, 2016).
In Science in Action, Latour describes the process of black-boxing in science. Before hypotheses become facts, a transparent process of research has to take place. After new ideas become less controversial, they slowly become accepted facts, that are not questioned any further. In other words, they turn into opaque black boxes. According to Latour, this happens with technological development as well, when the inner workings of the technology are no longer questioned. When technologies become accepted, the sole focus is on the in- and output and the inner workings become more and more opaque. We forget the historical background of the technology or just don’t understand the way it was built anymore. According to Latour, by making the inner workings of certain technologies transparent, the public is again able to question the moral prescriptions incorporated in the technology. It allows them to choose whether they want to
15 obey the moral prescriptions incorporated in the technology or to test whether these are in accordance with the law (Van Den Eede, 2011).
In the foregoing paragraph, the importance of transparency in technology is described. More about this in chapter 2.5, first it is good to delve into sub-question 2: What is meant with the
term transparency?.
2.4 Transparency
The public administration literature on the topic of transparency has a long history. The idea of transparency comes from Bentham’s panopticon idea and says that people will behave correctly when there is a possibility that they are being watched. Well-known is the panopticon prison whereby the inmates could be watched by inspectors in a central tower. Because the inmates were not able to see whether they were watched by the inspectors, misbehaviour could always lead to punishment. Later Bentham added the constitutional-panopticon to his theory whereby the citizens are monitoring the governors. This is a somewhat reversed version of the prison panopticons as in this case the many are watching the few. Nowadays in the Western world, transparency is a basic requirement of democratic governance to prevent the abuse of power (Galič et al., 2017).
Although the idea of transparency has been around for centuries, the concept is often kept ambiguous and even in the academic literature, the concept is often ill-defined (Meijer, 2014). Definitions that are found all have three things in common; (I) they include the availability of information which (II) enables the public or other actors to (III) monitor performance (Meijer, 2014). The importance of performance monitoring is partly due to the close relationship between transparency and accountability. Hale (2008) describes accountability by splitting the concept into two parts; enforcement and answerability. Enforcement can be seen as the power of the one accountable to exert pressure and to influence the behaviour of another actor. But to be able to influence an actor in the right direction, it is necessary to have information about its behaviour. This is where the term answerability comes into play, which is defined as the right of the one accountable to receive information and the obligation of the agents to share information about the agent’s actions. Knowing this a fine working definition of transparency is found in the following: ‘the availability of information about an actor allowing other actors to monitor the workings or performance of this actor’ (Meijer, 2014).
16 So transparency is about information necessary to check whether an actor or network of actors is functioning as agreed upon via contractual agreements or to check whether decisions are in conflict with the law. Both are not only of importance in relation to private organisations, but the public legislative power also has to be tested against constitutional principles such as equality and fairness. Because this can be seen as the bedrock of democratic legitimacy, transparency about the process of decision-making is a necessity. Although, the executive power should only implement law prescribed by the legislative power, due to the increasing scope of regulation there is a growing chance of misuse or even abuse of the executive power. Again, transparency about the functioning of public actors helps to check these powers on fairness and prevent decision-making on a discriminatory basis (Bugari, 2004).
As Van Staveren et al. (2014) describe, it can sometimes be hard for people that device and implement policy, to look critically at their work. Being supported by parliamentary decisions or the thought of a higher goal like security, it can be tempting to see their choices and used means as normatively desirable. But the legitimacy of a goal like security does not say anything about the legitimacy of the chosen means. This, of course, is not only true for policymakers but everyone working in these networks.
There are two forms of transparency. Internal transparency can be seen as the availability of information about the performance of other actors in the network. External transparency, on the other hand, is then logically the extent to which this is information is visible to the outside world and whether it allows inferring conclusions from it (Reynaers & Grimmelikhuijsen, 2015). Current smart city initiatives are more and more present in the public space and as a result, more and more people will come in contact with its components and its implications. This paper will, therefore, focus on the extent to which the public receives information because this is vital to question the chosen means on the legitimacy and on unlawful inference.
After this brief overview of the public administration literature on transparency, it is time to answer the third sub-question: What does transparency mean in relation to smart city projects?.
2.5 Transparency in smart cities
Full transparency in smart cities is only achieved when all of the earlier described components of the assemblage are made transparent. First of all, this would mean openness about the involved stakeholders in an assemblage. As described in chapter 2.1.3, these assemblages consist of
17 knowledge institutes, private or public stakeholders or a partnership between them. Only when citizens know who is collecting, they can do further research into the intended use of the sensors, produced data and used back-end algorithms. First, let’s look into how the public can be made aware of present sensors and the kind that is used.
2.5.1 Sensor awareness
As we have seen, besides public and private human actants, non-human actants in the form of technological artefacts are equally important. Awareness about sensors in public space is nowadays ever more important. Due to contemporary advances in the downsizing of technological artefacts, it is now possible to install sensors and cameras that fade into the background. In fact, oftentimes these technologies are specifically designed to be invisible to a wider audience and to withhold whether they are monitoring and for what purpose. Because of this, the general public is often unaware of the presence of these technologies, which is undesirable say Mikusz, Houben, Davies, Moessner, & Langheinrich (2018). They argue that a lack of awareness can lead to frustration and mistrust of the technology. Griffiths (2016) adds that the invisibility of these objects means that the public has no free choice to be monitored or not.
To date, in the Netherlands no real legislation exists, dictating whether and how the public should be informed about the presence of monitoring devices. Although the General Data Protection Regulation (GDPR) of the European Union is not specifically focussed on data mining by sensors and cameras in public space, still some articles apply to this domain since new obligations related to transparency are implemented with the arrival of the GDPR. Articles 13 and 14 are especially interesting in this regard because they entail new notification duties for data “controllers”. Where personal data is obtained, the controller has to inform the subjects about, for instance, the intended purposes, the period of storage and the identity of the controller (Wachter, 2018).
In a whitepaper about smart city sensors and the role of municipalities, the umbrella organisation of the Dutch municipalities (VNG) states a desirable end result concerning sensors transparency: “everybody should be able to know what kind of sensors are used, the location these
are at and which purpose they have” (Van Barneveld et al., 2018, p. 7). The way this should be
communicated, however, is still unclear because national legislation does not exist yet. Due to the lack of legislation, it is even allowed to have no indication of sensors at all. When cameras are
18 used, they only have to be clearly visible and the public doesn’t have to be explicitly informed. However, oftentimes signs are used when cameras monitor public space. Article 12 of the GDPR mentions standardised icons on signs accompanied by short texts as a possible way to inform people of sensors (Rathenau Instituut, 2018). According to Wachter (2018) however, their power to inform the public is very limited. Some complicated processes are just impossible to describe in symbols, for instance, what kind of symbols can describe what will happen to collected data and the kind of algorithms that are used. Standardised icons may in those cases prove insufficient.
Another suggested option is a register for sensors and cameras in public space, accessible for the public. An obligation for private and public actors to report the use of sensors helps to map all current and future sensors but this kind of legislation does not exist yet. Another option could be to oblige a permit for the use of sensors. This too helps to map all the sensors and could possibly prevent the proliferation of monitoring devices in public space (Geonovum, 2017).
2.5.2 In- & output
Another section of the transparency goal formulation of the VNG is concerned with the kind of
data that is collected (Van Barneveld et al., 2018, p. 7). To make no distinction between data or
information, these are hereinafter also referred to as in- & output. Earlier in chapter 2.1, the neutrality and objectivity of data were briefly discussed. Some might say that data speak the inherent truth and by analysis of data evidence-informed decisions and technocratic governance would be reached (Kitchin, 2014b). The idea of neutrality and objectivity oftentimes stems from the ideas that (I) data and especially big data can capture a whole domain without omitting a thing, (II) no a priori theories or models are needed to capture it, (III) data speaks for itself and is therefore free of human bias or framing and (IV) meaning transcends domain-specific knowledge so everyone is able to understand it (Kitchin, 2014c).
First of all, although big data might try to describe and capture everything, it’ll always be a representation created by using particular tools, a sample that will never reach the level of an all-seeing God’s eye. Second, building technological systems without theoretical models is impossible. Sensors and cameras used for mining the data are designed based on theories which are built in the aforementioned scripts of ANT and are tested and refined scientifically. Third, making sense of data always needs some sort of conceptual framework or some sort of lens. Even if data is processed to information automatically by algorithms certain values and contextualization
19 are added. This can happen (un)intentionally by the software engineers or passed on by historical data which could never be fully neutral as well, but more about this in the following chapter. Fourth and finally, the idea that everyone can understand data would cause an outright reductionist form of quantitative analysis which ignores the effects on other domains such as for instance culture and policy (Kitchin, 2014c). Transparency would, therefore, mean that the beforementioned scripts, chosen conceptual frameworks and added values are uncovered and made visible which in turn enables us to test whether these sensors are used as intended and legitimately.
Data or in- & output transparency first prompts the questions about ownership and for this, it is important to differentiate between public and private actors because different rules apply to each of these sectors. Since transparency is an appreciated democratic value, public actors are bound by certain rules, dictating what kind of data and information should be open to the public. In the Netherlands, the Wet Openbaarheid van Bestuur (WOB) is the most important regulation in this regard. It stipulates that administrative information of the government and its bodies should be open unless determined differently by other laws. Information is deemed confidential when it is among other things, a threat to the Crown, national security or in violation of the law for the protection of personal information. The term Open Data is used to denote the kind of data or information that is open and free to use (Geonovum, 2017). From a public perspective, data can be Open Data or confidential due to certain legislation.
Private actors, however, are not bound by a philosophy of openness, in fact, some companies even depend their business model on opaqueness by using data a commodity for sale. This adds a third form of data and information, besides confidential and publicly available there is also the commodity that is for sale. This third form is not only used in the private sector though. There are a few examples where it is used by the government as well; think of extracts of the chamber of commerce or population register.
2.5.3 Back-end
Just as with data which was discussed in the previous chapter, ANT tells us that technologies are not neutral tools. Insight in the way technologies are built would, therefore, help us when analysing its incorporated scripts to check whether the artefacts are indeed used as stated in the purpose statement but also help us analyse the technology’s incorporated values on biases and fairness.
20 Nowadays, software algorithms are increasingly automating the processing (prioritization, classification, association and filtering) and analysis of data and, therefore, play an ever more important role in the creation of information. When these technologies become accepted, the sole focus is on the in- and output and the inner workings become more and more opaque. We forget the historical background of the technology or just don’t understand the way it was built anymore. When technologies become opaque because human operators don’t understand their inner workings anymore, there is a risk that the incorporated values bias these operators (Diakopoulos, 2014). Of course, biases and wrong decisions existed way before the invention of the computer, but human errors occur oftentimes locally and randomly, whereas errors in software could reappear systematically everywhere and as long as it is used (The Lancet Respiratory Medicine, 2018).
As mentioned in the previous chapter, these incorporated values could be the result of particular choices made (un)intentionally by software engineers or by the software itself when machine learning techniques are used. Choices include among other things prioritisation of data above other data or combination of certain sources. If the software has been coded manually, transparency is important because it is not for software engineers to define an ex-ante understanding of democratic values and because people should never a prior accept and trust all the choices they made (Kroll et al., 2017).
If, however, machine learning techniques have been used, the risk of biased systems containing rules that are discriminatory is even bigger, says Kroll et al. (2017). First of all, algorithms can lead to discriminatory results when the data used for training contains past prejudice, biases or even only show a distorted picture of the overall population. Secondly, choices in training data or feature selection can lead to a distorted sample and distorted picture of the population with a change of discriminatory results. This can happen for instance when gender is used as a selection factor in job applications. In this example membership of so-called protected classes – these groups such as gender, race and age enjoy special protection by law against discrimination – is used directly as an input of a model. Another example is a model that relies on factors that serve as proxies for protected classes; think of a model that uses a person’s zip code to determine their race. The third cause for discriminatory machine learning software is the possibility of masking intentional discrimination by using the aforementioned techniques such as skewing the training data or picking proxies for the group of interest.
21 Leaving final decision-making to human operators doesn’t solve this problem, as they can still be influenced by the technology by directing their attention or by recommending certain steps. Moreover, opaque inner workings, complicate the answerability of the decision-making of an operator. When it’s unclear on what basis prioritization or classification has taken place to turn data into usable information, it is impossible to say whether decisions on the basis of that information were fair and just (Diakopoulos, 2014). It raises the question of how can people be accountable for a decision they didn’t make themselves and can’t explain?
The foregoing shows why it is good to know on what basis information is created. Opening up these systems could reveal what criteria, limits and uncertainties exist in them, however, there are a number of issues to take into account. First of all, the involvement of private corporations could complicate achieving the goal of transparency, says Diakopoulos (2014). Private corporations often limit their overall transparency for multiple reasons. Firstly, full transparency might reveal precious product advantages over competitors. It could, furthermore, hurt their ability to do business when trade secrets are revealed. The third reason Diakopoulos (2014) mentions, is the possibility of manipulation when the inner workings of technologies are revealed. This issue does not only apply to private corporations though but also public actors could very well be reluctant to share information to prevent manipulation and strategic gaming of the system. Take, for instance, the algorithms used by tax authorities that look for signs of tax evasion. When the public knows exactly how these systems come to a conclusion, it becomes much easier to cheat and the predictive abilities of the system might be lost (Kroll et al., 2017).
In reality, only a small percentage of people is able to understand software code and could potentially game the system or check the output on any unwanted biases. This, however, is another issue that complicates achieving full transparency in algorithms. Even if some form of openness about the inner workings is given, few people are able to analyse it (Diakopoulos, 2014). Demanding a step-by-step explanation of the algorithm would counteract this. But third and finally, software is oftentimes a dynamic product that changes over time. In many cases, engineers tweak their code continuously to make it better suited to the changing world. Because of this, there is a risk that published, explained and analysed source code quickly becomes outdated. This is even more true for systems that make use of machine learning or predictive analytics. These systems are observation-based and can update themselves after each decision; incorporating new
22 examples as training data. This adds to the risk that rule disclosure has become outdated, the moment it can be analysed (Kroll et al., 2017).
23
3 Methodology
The following chapter discusses the methodological model, that has been chosen to conduct this research. This includes the design of the research and the operationalisation of the concepts. In addition, this chapter also contains a description of the selected cases and a brief discussion of the method of data collection and analysis. Finally, a discussion of the validity and reliability is given.
3.1 Research design
The purpose of this research was to get an in-depth insight into the phenomenon of smart cities. The research was, therefore, conducted using a case study design, since this method allows to focus on certain characteristics and attributes of a certain phenomenon.
A case study is a type of qualitative research in which in-depth data is gathered about a specific poorly understood situation for the purpose of learning. Yin (1989) captures the need and use of case study research in a clear way: "the distinctive need for case studies arises out of the desire to understand complex social phenomena (…) real-life interventions that are too complex for the survey or experimental strategies (as cited in Maassen, 1999, p. 26). The purpose of this thesis was to learn more about the impact of Dutch smart city initiatives on transparency, which up until then was poorly understood. Definitely a complex social phenomenon and thus, multiple case study was a suitable method.
Njie & Asimiran (2014) describe three types of case study research, with each different characteristics and purposes. The intrinsic case study is all about getting a better understanding of and a zealous focus on a particular case when ignoring other curiosities. An instrumental case study, on the other hand, uses a case to provide insight into an issue but the case itself is of secondary interest. Because, at the time of writing, all smart city projects in the Netherlands were still in their infancy, it might have been possible that not all factors were fully developed yet. Therefore, the multiple case study method was chosen that jointly studies multiple cases on a certain phenomenon.
24
3.2 Case selection
3.2.1 Case description
At the moment of writing, multiple cities in the Netherlands were researching the possibilities and opportunities of smart city initiatives. Only a hand full, though, actually started a pilot project. This meant that there were not many examples to study and the examples that were around, were often still in the development phase and covered a relatively small area.
As we see later, the Living Lab framework is often used in Dutch smart city projects to research innovative technologies. Although no common definition of the term exists, four main characteristics can be discovered. First of all, the research and developing of new technologies take place in a real-world setting, where stakeholders of multiple organisations and with different backgrounds interact with each other. Secondly, end-users play an important role in all stages of the developing process. Third, the multidisciplinary nature of the stakeholders is a key aspect in achieving the goal. Fourth and finally, collaboration in as well the physical as the virtual world is off great importance (Van Bueren & Steen, 2017).
3.2.1.1 Ams Institute
The Ams Institute is a collaboration of public, private and educational actors founded in 2014. Their mission, as they say, is to get to know or sense the city of Amsterdam and design solutions for its challenges. They hope to create an innovative, sustainable and just city by applying technological solutions and focus specifically on themes such as water, energy, waste, food, data and mobility (Van Bueren & Steen, 2017).
3.2.1.2 Field lab Rotterdam
Fieldlab Rotterdam is a partnership between the municipality of Rotterdam, the Ministry of Justice and Security, the Dutch police force, educational institutes and private actors which started in 2018 (Fieldlab Rotterdam, 2019b). In the Rotterdam district of Lombardijen, the consortium wants to install street lighting, that is enhanced with a variety of sensors. The idea is that those sensors and cameras will recognise deviant behaviour and help to prevent domestic burglaries. When the smart lamppost are fully tested in their testing environment, they will be placed on the public street. At that moment, the public will have the chance to get involved more (Fieldlab Rotterdam, 2019b).
25 Currently, however, no artefacts are placed in a real-world setting yet as the lab is still in the developmental stage.
3.2.1.3 Living Lab Stratumseind
In this lab, the municipality collaborates with knowledge institutes and private partners to provide a more safe environment in the popular restaurant and bar street Stratumseind in Eindhoven. In an interview with the Rathenau Instituut, project manager Tinus Kanters explains that by gathering large amounts of data on a variety of domains such as the number of visitors, the weather, sound levels, etcetera, this lab wants, for instance, to predict fights and other security issues. Furthermore, the lab wants to test and develop mechanisms to nudge people into desirable behaviour. Examples of these nudging techniques are the spreading of calming scents or the changing of the colour of the lighting (Rathenau Instituut, 2019). Currently, several technological artefacts are tested and used in the real-world setting of the Stratumseind bar street.
3.2.1.4 Living Lab Scheveningen
The Living Lab Scheveningen is a test environment near a popular beach in the city of The Hague. According to IT policy advisor of the municipality of The Hague Van Meerten (2019), the site is designed for the municipality, private parties and knowledge institutes to develop innovative ideas and solutions. The increasing growth and densification of the city create challenges on the domains of, among other things accessibility, sustainability and safety. The use of new technologies, Max van Meerten says, could offer possible solutions to these problems and the municipality of The Hague has, therefore, started the program Smart City Den Haag to test these technologies and to contribute to an attractive living environment.
Just as with the Stratumesind Living Lab and Field lab Rotterdam, the roadmap of The Hague mentions the use of sensors and cameras and integrate those in public space to prevent nuisance. In both cases ‘smart’ lamppost are used as hubs for attaching sensors and cameras (Rathenau Instituut, 2019; Van Meerten, 2019).
3.2.2 Justification
At the time of writing, the focus of the Ams Institute and affiliated Living Labs in the city of Amsterdam was more on themes such as sustainability and mobility and less on issues related to
26 security. In this thesis, however, smart security assemblages were studied and the projects of the city of Amsterdam were therefore not suitable for this research.
The remaining three projects in the cities of Rotterdam, Eindhoven and The Hague all focus to some extent on smart security. Since a real-world setting is an important aspect of Living Lab projects, further selection has been made on the basis of similarities between test environments of the labs. The fact that the Stratumseind Living Lab in Eindhoven was already experimenting in a real-world setting for about six years now, gave the lab a frontrunner status in the Netherlands. Most other labs were, after all, still developing theoretical ideas and had not yet started real-life experiments. Due to this head start, the Stratumseind Living Lab in Eindhoven was in an important and interesting case to study. The Living Lab research area, located in a popular street with bars and restaurants, resembled the idea of The Hague. Here the Living Lab Scheveningen covered an area located near the beach at a boulevard close to bars and restaurants. In Rotterdam, however, they were, at the time, still planning to locate the testing area in a residential area, plagued by burglaries. Of course, the Rotterdam case would still be concerned with security in a public space, but the number of visitors and the type of security issues would certainly differ. For this reason, the Stratumseind Living lab, as well as the Living Lab Scheveningen, have been chosen for this research. In table 1, an overview is given of the current smart city projects in the Netherlands.
TABLE 1: SMART INITIATIVES IN THE NETHERLANDS
Name Stage of
development
Other remarks Source
Ams Institute Focusses on other themes than security
(Van Bueren & Steen, 2017)
Living lab Scheveningen
Starting up Focus on security a popular site near the beach
(Van Meerten, 2019)
Startumseind Living lab
Implemented initiatives
Focus on securing a popular bar street
(Stratumseind, 2019)
Field lab Rotterdam Starting up Focus on security a neighbourhood
27
3.3 Data collection
Data gathering was performed via in-depth semi-structured interviews. With respect to the selection of the interviewees, it was important that they were or had been involved in one of the two chosen smart city projects. The interviews were conducted with representatives from the public sector such as project or program managers of municipalities, as well as people working for research institutes and finally a product manager of a private company was included. The interviewees were selected on the basis of snowball sampling so participants referred to or recommended others that might be of interest to the research. This strategy was chosen because it was initially impossible to know who has been involved in the two projects, especially when dealing with organisations as large municipalities of The Hague and Eindhoven. Using snowball sampling, the knowledge and the network is used of people better known within the specific smart city projects.
TABLE 2:RESPONDENT CODING
Code Respondent Description
R1 A civil servant of the municipality of Eindhoven, involved in the Eindhoven Living
Lab smart city initiative.
R2 As an employee of the Dutch Institute for Technology Safety and Security (DITSS) – a knowledge institute concerned with Dutch smart city projects – this respondent has been involved in and part of multiple Dutch projects. One of which is the Eindhoven case.
R3 A civil servant of the municipality of The Hague, involved in the Scheveningen
Living Lab smart city initiative.
R4 An employee of The Hage Security Delta – a cluster of businesses, governments and
knowledge institutes and this position enables a unique view. They have been involved in several smart city projects among which the Scheveningen Living Lab and The Hague international zone project.
R5 A product-manager of a private sector company involved in the Stratumseind Living Lab.
28
3.4 Operationalisation
Transparency can be understood as the availability of information about the functioning of the assemblage to the outside world. In a whitepaper about sensors and the role of municipalities, the umbrella organisation of the Dutch municipalities (VNG) made the following statement about the desired end result concerning transparency: ‘everyone should be able to know who is collecting what kind of data, with what kind of sensors, located where in public space, and for what purpose’ (Van Barneveld et al., 2018). It states a transparency goal formulation on all aforementioned parts of smart city, which makes it a useful guideline for the rest of the thesis. After the interviews were done, they were all transcribed and coded using a thematic approach. In this thesis the effects of smart city implementation on the public value of transparency were studied. The following table 3, presents the codebook that will be used to code the interview transcripts on quotes of interest. The concepts have been divided into several indicators that show how transparency is affected by Dutch smart city project.
29
TABLE 3:CODEBOOK
Concept Indicator Description
Network clarification
_________ _________
Stakeholders Parts that describe the different stakeholders that are involved in the smart city initiative. Excluded are parts that describe informing the public about the stakeholders present in the network.
Openness contractual agreements
Sections describing the degree to which contractual agreements about for instance data usage and storage can be consulted by the public.
Goal formulation Fragments that describe the importance of a clear goal formulation or describe examples of clear goal formulations.
Business model Parts labelled business model distinguish different earning opportunities in relation to smart cities or delineate issues concerning these models.
Awareness creating
_________ _________
Methods All methods used and available to make the public aware of the presence of sensors in public space. Examples are among other things: Information sign, Media usage, (group) data walks, website/ register/ database.
30 Information supplied This is understood as the kind of information that is provided to the public concerning the use of everyware. This includes information about the location of the sensors, the type of sensors, the kind of data collected the possibility of nudging. Also included are fragments that are describing citizen involvement in decision-making
Excluded are fragments specifying the in- & output and stakeholders in the network that are not related to informing the public or fragments about software working in the back-end.
Public law _________ _________
Sensor install regulation Parts about legislation regulating the way sensors and cameras are installed in public space. This includes an obligation to report new everyware to the municipality, a permit requirement when installing a new
everyware and fines in the case people did not obey the aforementioned legislation.
Other legislation or guidelines
Other legislation concerning smart city initiatives.
Request for open government
Parts that describe the influence of laws about open government
In- & Output _________
Specification Fragments specifying the kind of input that is collected and the output that is generated.
31
_________ These include among other things: personal data, anonymised data etc.
Ownership Fragments that describe the ideal or real owner of data or information collected in public space
Openness Fragments describing the degree and manner the in- & output is made public and available. This includes, for example, confidential data and information, pay per use and open data/information.
Data/ information discussion
Parts about the data and information spectrum and its vague dividing lines.
Back-end _________ _________
Coding method Fragments about the coding method describe whether some form of machine learning was used or that the rules in the software were manually coded or a mixture of both methods.
Decision-making These parts describe how the system comes to a certain decision and how actants further down the network are controlled. The two possibilities of interest here are (I) human operator or (II) fully automated decision-making.
32 Transparency level This describes the openness of the
functioning of technological parts of the smart city network to other nodes in the network as well as the public at large is described. It can be seen a spectrum with on the one side fully confidential source code and on the other open-source code.
In the literature system outcome explanation is mentioned as a way to boost transparency without publishing the source code.
Other Parts about the software back-end that do are not covered by the earlier labels
3.5 Limitations
3.5.1 Validity
The validity in academic research is concerned with the question of whether the method of research could provide an answer that is an adequate reflection of the truth. The validity can be divided into two dimensions, which are the internal and external validity. The former dictates whether there is enough basis for concluding that the correlations found in the research area are, in fact, causal and not the result of some other non-considered factor. (Roe & Just, 2009). Because transparency is highly valued in a democratic society, there was some risk that the answers given by the interviews were biased due to social desirability. This was especially risky in the Scheveningen case because, due to its early development, not much had been published about it. Triangulation of the data was therefore complicated.
Because of the exploratory nature of this research, it was better to be flexible during the interviews and have the possibility to elaborate on unexpected topics. Although structured interviews have a higher internal validity or causality, semi-structured interviews lend themselves better for this purpose. Testing the questionnaire with an authority in the field helped to increase internal validity. This helped to make sure the questions yielded results that answer the research questions.
33 External validity, on the other hand, focusses on whether the results of the research are applicable to the general population. Research can be internal valid with results that are replicable for that specific case but that are not applicable to the whole population. This happens when the chosen case(s) or experiment(s) don’t represent the entire population of interest (Schram, 2005). To ensure that the results are externally valid, two cases have been studied, allowing to check whether the results match or whether it is case-specific. As said before, smart city projects in the Netherlands were still in their infancy, at the time of writing, and the number of projects was low. Finding two cases of smart security in the Netherlands that were in the same stage of development was not possible and is a limitation. The two cases that have been picked were the ones that are most similar.
3.5.2 Reliability
The reliability or repeatability of research is about the accuracy and consistency of results. In most quantitative and even in some qualitative research it is possible to use the test-retest method. Hereby the same test is done multiple times and when the consistency of results is above a certain degree, the test is seen as reliable. When using interviews as a data collection method, you can’t expect people to participate over and over again in the same interview. A different method has to be found to check if results are repeatable. By triangulating different data sources, answers of respondents can be checked on their reliability (Golafshani, 2003). Available municipal documents, newspapers and multiple academic sources have been checked.
When coding the transcripts, the intra-coder reliability was checked by coding the same transcript two times and look how much of these overlap. Averaged over the five transcripts the intra-coder reliability was 88 per cent, which is thought to be sufficient.
34
4 Analysis
This chapter answers the fourth and finally sub-question How is transparency affected by the
implementation of these Dutch initiatives? and follows the same structure that was predominantly
used during the interviews. Since sensors can be seen as the front-end of the technology, the questionnaire started with sensor awareness. This was often logically followed by questions related to data and information and then the algorithmic back-end of the technology was discussed. Finally, questions about the organisational aspects of the network were asked.
4.1 Sensor awareness
The interviewees were asked whether the organisation they work for believes that people should be informed about the presence of sensors in the public space and what information should be provided. All of the interviewees agreed that people should be informed to some degree but they mentioned different types of information; which were among other things information about the location of a sensor, the sensor type and the owner or a contact person of some sort for further questions.
So it didn’t turn out not to be a question of ‘if’ as all participants of Eindhoven and The Hague agreed that informing people was important, it turned out to be rather a question of ‘how. Talking about Stratumseind, (R2) – an employee of DITSS a knowledge institute involved in the Stratumseind case – mentioned several tactics when asked whether residents were informed prior to the project:
“Yes bar owners and residents’ associations were involved (…) and we organised, of course, several information evenings but yeah.. not all residents show up there, and I’m sure that a note has been written to all residents in the lines of ‘this is what we are planning to do’.. but not all of them read those. Furthermore, articles have been published in Eindhoven Dagblad en Groot Eindhoven (ed. local newspapers). Enough attention has been paid to it, so you could have picked it up.. but we didn’t put an information letter in every mailbox. However, I’m talking about residents and entrepreneurs now, I don’t talk about visitors to the Stratumseind.” (Appendix II,
35 The quote of (R2) shows the extensive effort they put in to inform residents and local entrepreneurs. Close contact with bar owners and residents’ associations and tactics such as special information evenings and items in local media show that they wanted residents to be informed prior to the project. The quote ends with a remark though, visitors from outside the region could have missed all these awareness creating methods as they were mainly focussed on informing locals. Another method which could be used to notify people that missed all earlier mentioned ones, came up in conversation with a civil servant of the municipality of Eindhoven (R1). When asked whether the municipality of Eindhoven thinks informing the public is important, (R1) responded:
“We laid down in our principles that it must happen. Reality is more stubborn though. We made a design for a warning sign that we think could inform people. However, we believe this is a national discussion.” (Appendix I, par 16).
“We are currently in conversation with the Ministry because we believe a sign should be designed for the Netherlands as a whole, a warning sign with a text along the lines of ‘You are entering an area where sensors are present’. But if that takes too long then we, as the municipality of Eindhoven, will do it ourselves.” (Appendix I, par 18).
The two quotes show that the municipality of Eindhoven, actually would have liked to inform people about everyware in the public space using warning signs already but that this is not the case yet. The signs with a text along the lines of ‘You are entering an area where sensors are present’ and a website and QR-code where people could get more information were never put in production and never installed in the Stratumseind. According to Tinus Kanter, the reason for this is that citizens are not helped with different signs in every other city, so they are currently in conversation with the Ministry to get a single sign approved for the Netherlands as a whole (Appendix I, par 18 & par 20). Earlier we saw that Wachter (2018) had some reservation about the usefulness of warnings sign as they do not provide enough space for informing texts about complicated processes as the used algorithms. The plan to use QR codes linked to a website with more information is a solution for this.
36 The end of the second quote shows that the city of Eindhoven believes that the public must be informed as a matter of urgency. During an interview with (R2), a different kind of urgency relating to the use of warning signs came up:
“Yes.. well... look if they (ed. sensors) are present in public space then people should be informed. But I am convinced that within ten to twenty years the signs can be removed because the entire public space is filled up with sensors.” (Appendix II, par 70).
So (R2) shares the opinion that right now warning sings could indeed be used to inform people about the presence of sensors. His statement is followed, however, by an interesting comment on the use of warning signs. In the near future the signs would lose their usefulness, due to the potentially high number of sensors in the public space. Interestingly, a civil servant of the municipality of The Hague (R3) shares this opinion but goes even further and wonders whether informing the public should happen via visible objects in the public space at all:
“The question is whether you want to fill the entire city with warning signs, just like in America when you are on the escalator that you continuously hear 'mind your step, mind your step', the question is whether that is the way we would like to go. I would… but that is a personal opinion, I would like to arrange it differently. Open, making sure that it is accessible.. we are also building a digital city a 3d model of the city of The Hague, in which you can see all those things… that will become public as well. (…) But to do that with signs on the street, I think that is old-fashioned thinking.” (Appendix III, par 62).
The vision of the city of The Hague is an attractive outdoor space with sensors integrated into existing objects, warning signs on every corner would only ruin this. They would prefer to inform the public via an online platform that maps out all existing sensors. The idea is similar to that of Eindhoven, both want to provide information online but in Eindhoven, the idea is to use signs that refer to the website.
