Keywords

1 Introduction

In today’s digital economy we pay for technological services with our data (Elvy 2017). Our data is used to build profiles, for advertising and marketing. In the loT context, the collection of data does not end after the consumer purchases a device online or in a store, but instead increases once the consumer begins to use the device, as well as accessible websites and mobile applications (Elvy 2018). In a review article on AmI (ambient intelligence), Augusto et al. (2010) asks: “how can you defend your privacy if wearing networked devices makes you trackable everywhere you go?” This question is becoming more and more relevant to everyday life. According to Streitz et al. (2019) the big challenges for ambient intelligence concern two trade-offs: “Keeping the human in the loop and in control … (and) Ensuring privacy by being in control of making decisions over the use of personal data vs. intrusion of often unwanted, unsupervised and importunate data collection methods as a prerequisite of providing smartness, for example, in terms of smart services.”

In this paper we are primarily concerned with privacy and not with how data can be used to offer customized services. The current study builds on findings from two national consent studies (Tjostheim and Waterworth 2020) in which we tested willingness to share personal data and attitudes towards the use of data to build personal profiles and assess susceptibility to phishing. Our analysis indicated that a willingness to share data makes users vulnerable to phishing. This is only one aspect that illustrates that the citizen is the weaker party in the data and services economy. In these two national studies, in response to the statement: “it should be made harder to collect data and build digital profiles”, 75% and 69%, respectively, indicated full or partial agreement. The studies revealed particularly strong negative attitudes towards the sharing of data with third parties and the analysis of data “to learn as much as possible about their users,” with 89% negative and 86%, respectively, responding negatively.

There is an asymmetry of knowledge or power if we are monitored without being aware of the monitoring. Some authors use the metaphor a one-way mirror to explain how it works (Cyphers and Gebhart 2019). Even more well-known is panopticism, a term that first was used for the architectural design of a prison by Jeremy Bentham. Usually, we think of a panopticon as embodying an oppressive, top-down system of observation and control, with Big Brother watching us, as in Orwell’s 1984. But Bentham distinguished between four different types of panopticon. In this paper, we focus on his constitutional panopticon, as described in Constitutional Code (Bentham 1830). We use this also referred to as the inverse panopticon, or reversed panopticon; (Galič e al. 2017) as a theoretical concept within an explorative study of users of apps and smart environments.

For most readers the word panopticon has a negative connotation. But with an inverse panopticon, there is envisaged to be a bottom-up surveillance through which governing functionaries are monitored through panoptic methods (Brunon-Ernst 2013) rather than the other way around. Mann (2004) calls this sousveillance or watching from below through procedures of using technology to mirror and confront bureaucratic organizations (Mann et al. 2003).

In the Constitutional Code the focus is on the governing functionaries of public officials, ministers in a government. In this architectural arrangement the prime minister is in the center, in an oval shaped building but without the central inspection tower. The key to good behavior, good government, according to Bentham is transparency and publicity – the many should watch the few rather than the other way round.

The literature on the inverse panopticon is sparse, but research on privacy and the sharing on personal data gives us an indication on what we might expect. Given the widespread concern about how data is used and shared with third parties (Barth and de Jong 2017; Preibusch 2013), we expected that the inverse panopticon concept will be perceived as a meaningful and useful concept, but we have not identified any studies with large samples on this subject. Against this backdrop, our research addresses citizens’ perspectives on surveillance and the inverse panopticon within ambient environments. The most well-known devices are perhaps smart speakers, but in a modern house smart-home technologies to control light, temperature and door-locks are also common.

In the present study we focus on attitudes towards devices that the citizen could use to protect privacy, which can be termed inverse panopticon applications.

2 Method

We carried out the study in cooperation with a market research companies PollStat AS, to achieve our aim of national studies on an issue affecting a broad section of the population. The market research company uses quota sampling to provide a sample that is nationally representative in terms of age, gender, education and geographical region. In our survey, we targeted citizens 16–69 years old.

From the national panel of online users, we recruited 1289 participants. In Norway, nearly all citizens own a smartphone and therefore a high percentage are categorized as online users. There are citizens that are hard to recruit in surveys, these are primarily citizens with very low socio-demographic characteristics, but for this study they are not a targeted group. In a longer perspective it is likely that IoT devices will be even more common among all groups of citizens.

The national bureau of statistics in Norway, Statistics Norway, has the demographic profile of Norwegians in the population register. In the first table we compare the profile of our respondents with a similar survey but not from the national bureau of statistics. The market research company cannot recruit directly from the population registry, but it is a goal to recruit a sample with a similar profile. Table 1 shows that our sample is similar to the sample of the ICT-survey named the use of ICT in Norwegian households by Statistics Norway, 2019.

Table 1. Educational profile of the participants in the survey.
Full size table

After asking the respondents about IoT-devices, we introduced two devices. The first was an application for smart speakers is an ultrasonic jammer that prevents smart speakers from recording human speech (Chen et al. 2019). A similar device was also featured in a later New York Times article (New York Times 2020). The second device was similar to the SpecEye application (Li et al. 2019), a screen exposure detection system that can be used to identify the presence of digital cameras. Another solution is AudioSentry, an application that hides reflections off the user from the sensing signal by covering them with the reflections from an obfuscation signal (Gao et al. 2019).

We presented the illustrated devices to respondents after the questions about digital profiles and privacy (see Fig. 1). Studies show that respondents can be influenced by the question order. According to Strack (1992), when respondents perceive survey questions as belonging to the same conversational context, the responses tend to adhere to Grice’s (1975) principle of conversation, that is the content of previous conversation. “When respondents answer questions that pertain to the ongoing conversation, they make their responses as relevant to the conversation as possible.” (DeMoranville and Bienstock 2003: 2018)

Fig. 1.
figure 1

The Bracelet (the Norwegian text is translated under the pictures)

Full size image

We did not assume that the respondents had knowledge about devices such as the bracelet or a device that can detect cameras. Our assumption was that online users have limited knowledge about the idea of an inverse panopticon. The text above the four pictures was: “PCs, tables, smartphones and smart speakers have cameras and microphones. They can be on or off. In some occasions you decide when they are on or off. In other occasions, the recording takes place without the possibility for you to check that this is happening. This is an example of a bracelet that is sending ultrasound to jam the microphones – as a consequence the audio-recording is damaged.”

We decided to illustrate a bracelet similar to the one by the Chen et al. (2019). It is a prototype, but as an illustration it served our purpose. Since the survey started in January 2020, we could not use photos from the New York Times article (which was published February 14, 2020).

3 Results

3.1 Concerning Privacy Protecting Devices

After showing the illustrations we asked the respondents to choose an alternative from low interest to high interest. Table 2 shows that across gender, age and education 30% had low interest, 30% no opinion, and 40% were interested in a device jamming microphones. These numbers are for all that answered the survey regardless of use of IoT-devices.

Table 2. Demographics and attitudes to a device jamming microphones
Full size table

Many devices such as smartphones and PCs have cameras. Inside buildings and outdoors, there are often surveillance cameras. After the illustration about the bracelet for jamming microphones, we asked about a device that can detect cameras in the vicinity of the person. We did not use an illustration for his device, only text.

Table 3. Demographics and attitudes to a device for detecting cameras
Full size table

In the survey, the illustration about the bracelet and the question about surveillance cameras followed a number of questions about privacy, how big tech companies use personal data to build profiles, phishing and misuse of data.

Table 3 shows that across gender, age and education 25% had low interest, 25% no opinion, and 50% were interested in a device for detecting cameras in the vicinity of the person. Figures 2, 3 and 4 present these findings graphically.

The survey-data do not reveal demographic differences– 40% had an interest in using a jamming device. The patterns in Figs. 2, 3 and 4 are similar.

Fig. 2.
figure 2

Interest in the bracelet for different genders

Full size image
Fig. 3.
figure 3

Interest in the bracelet for different age groups

Full size image
Fig. 4.
figure 4

Interest in the bracelet with different educational levels

Full size image

Figures 5, 6 and 7 present the results for a device that can detect cameras. Figure 5 shows that approx. 50% of both male and female answered that were interested in using the device about twice as many that answered low interest. For all three age-groups, slightly more than 50% answer high interest (Fig. 6). Also, for all four educational groups, approximately 50% answered high interest twice as many as those who answered low interested (Fig. 7).

Fig. 5.
figure 5

Interest in a device for detecting cameras for different genders

Full size image
Fig. 6.
figure 6

Interest in a device for detecting cameras for different ages

Full size image
Fig. 7.
figure 7

Interest in a device for detecting cameras for different educational levels

Full size image

Figures 8 and 9 show that those who reported that they have taken steps to protect their privacy are more interested in the device that can tell them about cameras in their vicinity. For all four questions 1-in-4 answer very interested and this 10% higher than for those that reported that they have not taken steps to protect their privacy.

Fig. 8.
figure 8

The citizens who have taken steps to protect privacy

Full size image
Fig. 9.
figure 9

The citizens who have not taken steps to protect privacy

Full size image

3.2 The Users of Smartwatches and Smart Speakers

In this section we focus on groups are those that either have a smart speaker or a smartwatch, two IoT devices that have become quite common the recent years (Malkin et al. 2019). It should be noted that many of the citizens do not own or have any experience with IoT-devices such as smart speakers. Of in total 1289 respondents, 451 (32%) answered that they use a smart speaker daily or occasionally (Fig. 12).

Fig. 10.
figure 10

Responses to a camera detecting device for those with or without an IoT device

Full size image

The survey indicated that those with an IoT device responded more positively to the jamming and camera detection devices (see Figs. 11 and 12).

Fig. 11.
figure 11

Responses to a bracelet for those with or without an IoT device

Full size image
Fig. 12.
figure 12

Responses of citizens with/without an IoT smartwatch and a smart speaker

Full size image

There might be a difference between owning device, but not using it often and using the IoT device on a daily basis. We asked those with a smartwatch connected to the Internet and those with a smart speaker about daily use (see Fig. 12). In a survey, it is easier to answer a specific question than a more general one and to interpret the answers.

4 Discussion and Conclusions

How do we support users’ decision-making regarding the collection and distribution of their data within ambient environments? For a possible approach to this difficult issue, we go back almost two centuries to the concept of the (constitutional) panopticon, as envisaged by moral philosopher Jeremy Bentham.

We presented two devices that are not consumer products today and we assume that the devices were new to the citizens. It is therefore not a surprise that many, approximately 30%, did not have an opinion. We illustrated the bracelet device with pictures. Then we asked about the opinion about this device and a similar device. Before that we had asked a number of questions about privacy. With reference to Starck (1992), we assume that many of the respondents already think about privacy when answering the question about the two devices. We do not think that the answers, their stated preferences, can necessarily be used as a good indication of revealed preferences (Ben-Akiva et al. 1994), their likely behavior. However, in an exploratory study such as this, the illustrations with questions serve our purpose. We have identified an interested in privacy protecting devices – approximately 40% answered that they were interested in using the devices.

In a newspaper article February 14 (New York Times 2020), the law and computer science professor Woodrow Hartzog advocated regulations because, in his view, this kind of surveillance by the big tech companies, with smart speakers and cameras, etc., will not stop. Policymakers need to pass laws that more effectively guard our privacy.

While we do not see the inverse panopticon as the single answer to concerns and dissatisfactions about the asymmetries of power in favor of the big tech companies in the digital economy, it may be a potentially promising perspective for new studies and design work concerning if, what and how information is collected in smart homes from equipment such as cameras, screens and speakers.