Microsoft word - ambient_intelligence_v3.0a_final.doc
Social, Economic, and Ethical Implications of Ambient Intelligence and Ubiquitous Computing°
Jürgen Bohn, Vlad Coroamă, Marc Langheinrich, Friedemann Mattern, Michael Rohs
Institute for Pervasive Computing, ETH Zurich, Switzerland
Summary. Visions of ambient intelligence and ubiquitous computing involve integrating tiny microelectronic processors and sensors into everyday objects in order to make them “smart.” Smart things can explore their environment, communicate with other smart things, and interact with humans, therefore helping users to cope with their tasks in new, intuitive ways. Although many concepts have already been tested out as prototypes in field trials, the repercussions of such extensive integration of computer technology into our everyday lives are difficult to predict. This article is a first attempt to classify the social, economic, and ethical implications of this development. Introduction
The increasing miniaturization of computer technology will, in the foreseeable fu-ture, result in processors and tiny sensors being integrated into more and more everyday objects, leading to the disappearance of traditional PC input and output media such as keyboards, mice, and screens. Instead, we will communicate di-rectly with our clothes, watches, pens, and furniture – and these objects will com-municate with each other and with other people’s objects.
More than 10 years ago, Mark Weiser foresaw this development and described
it in his influential article “The Computer for the 21st Century” [Wei91]. Weiser coined the term “ubiquitous computing,” referring to omnipresent computers that serve people in their everyday lives at home and at work, functioning invisibly and unobtrusively in the background and freeing people to a large extent from tedious routine tasks. In its 1999 vision statement, the European Union’s Information So-ciety Technologies Program Advisory Group (ISTAG) used the term “ambient in-telligence” in a similar fashion to describe a vision where “people will be sur-rounded by intelligent and intuitive interfaces embedded in everyday objects around us and an environment recognizing and responding to the presence of indi-viduals in an invisible way” [Aho01].
This paper is based on an earlier conference contribution: “Disappearing Computers Eve-
rywhere – Living in a World of Smart Everyday Objects”. Proc. of New Media, Technol-ogy and Everyday Life in Europe Conference, London, UK, April 2003.
2 Social, Economic, and Ethical Implications
The vision of a future filled with smart and interacting everyday objects offers a
whole range of fascinating possibilities. For example, parents will no longer lose track of their children, even in the busiest of crowds, when location sensors and communications modules are sewn into their clothes. Similar devices attached to timetables and signposts could guide blind people in unknown environments by “talking” to them via a wireless headset [CoR03]. Tiny communicating computers could also play a valuable role in protecting the environment, for example as sen-sors the size of dust particles that detect the dispersion of oil spills or forest fires. Another interesting possibility is that of linking any sort of information to every-day objects, allowing for example future washing machines to query our dirty clothes for washing instructions.
While developments in information technology never had the explicit goal of
changing society, but rather did so as a side effect, the visions of ubiquitous com-puting and ambient intelligence expressly proposes to transform society by fully computerizing it. It is therefore very likely that this will have long-term conse-quences for our everyday lives and ethical values that are much more far-reaching than the Internet with all its discussions about spam e-mails, cyber crime, and child pornography. With its orientation towards the public as well as the private, the personal as well as the commercial, the vision of an ambient-intelligence land-scape aspires to create technology that will accompany us throughout our entire lives, day in and day out. And if Mark Weiser’s vision of “invisible computing” actually materializes, we won’t even notice any of it.
It seems to be clear that with these technical developments – pushed through
largely unnoticed by the general public and extending quite rapidly into our eve-ryday lives – unanticipated (if not unacceptable) standards could soon be set for the rest of our lives. In the following, we examine the driving factors behind the visions of ubiquitous computing and ambient intelligence – from a technical as well as an economic perspective – and we try to illustrate the social and ethical implications of a “smart world” that connects everything to everything else, where anywhere can potentially be contacted from anywhere else, and where everybody could conceivably interact with anybody (and anything) else.
The driving force behind continuing technological progress in the field of ambient intelligence and ubiquitous computing is the long-term trend in microelectronics: Moore’s Law [Moo65], drawn up in the late 1960s by Gordon Moore and roughly stating that the power of microprocessors doubles about every 18 months, has held true with astonishing accuracy and consistency. A similarly high increase in cost-efficiency can be observed for some other technological parameters such as stor-age capacity and communications bandwidth. To put it another way, prices for mi-croelectronic functionality with an equivalent amount of computing power are fal-ling radically over time. This trend, which is expected to continue for at least another 15 years, means that computer processors and storage components will
Social, Economic, and Ethical Implications 3
become much more powerful, smaller, and cheaper in the future, so that there will be an almost unlimited supply of them.
Even more important are the results of microsystem technology and nanotech-
nology. These could lead, for example, to flexible displays or electronic paper. Another interesting development is radio sensors that can report their readings within a few meters distance without an explicit energy supply – such sensors ob-tain the necessary energy from the environment or directly from the measuring process itself.
Electronic labels (so-called “smart labels” or RFID tags) also operate without
their own energy supply. Depending on their construction, these are less than a square millimeter in area and thinner than a piece of paper. In some ways, this is a further development of the well-known anti-theft technology involving security gates in department stores. However, this is not just about the binary information “paid / stolen”; within milliseconds, several hundred characters could be read and written “wirelessly” up to a distance of about two meters [Fin03].
What is interesting about such remote-inquiry electronic markers is that they
enable objects to be clearly identified and recognized, and therefore linked in real time to an associated data record held on the Internet or in a remote database. This ultimately means that specific data can be associated with any kind of object. If everyday objects can be uniquely identified from a distance and furnished with in-formation, this opens up application possibilities that go far beyond the original purpose of automated warehousing or supermarkets without cashiers.
Significant advances have also been made in the field of wireless communica-
tions. Especially interesting are recent short-range communications technologies that require very little energy, making it possible to produce designs that are much smaller and cheaper than today’s mobile phones. Intensive research is also being carried out into improved options for indicating the position of mobile objects. As well as increased accuracy (currently around ten meters for the GPS system), the aim is also to make the devices much smaller.
If you summarize these technology trends and developments – tiny, cheap
processors with integrated sensors and wireless communications capability, at-taching information to everyday objects, the remote identification of objects, the precise localization of objects, flexible displays based on polymers, and electronic paper – it becomes clear that the technological basis for a strange new world has been created: everyday objects that are in some respects “smart,” and with which we can even communicate under certain circumstances.
There are various ways of implementing such communication with things. As
one example, imagine everyday objects such as furniture, packaged food, medica-tion, clothing, or toys being equipped with an electronic label containing a specific Internet address as digital information. If you can then read this Internet address with a portable device just by pointing it at the object, this device can, independ-ently and with no further assistance from the object in question, access and display the associated information from the Internet via the mobile phone network. The user has the impression that the object itself has “transmitted” the information, al-though in fact it has been supplied to the display device via the Internet. The in-formation could be, for example, operating instructions, or cooking instructions
4 Social, Economic, and Ethical Implications
for a ready-to-serve meal, or the information leaflet for medication. The details of what is displayed may depend on the “context” – for example, whether the user is a good customer and paid a lot of money for the product, whether he is over 18 years of age, what language he speaks, or his current location, – but also maybe whether he has paid his taxes on time…
The foreseeable technological developments will therefore add an additional
new quality to everyday objects – these might be able not only to communicate with people and other “smart” objects, but also to discover where they are, which other objects are in their vicinity, and what has happened to them in the past, for example. Objects and devices could thus behave in a context-sensitive manner and appear to be “smart,” without actually being “intelligent.”
While technological advances such as miniaturization, increasing computing
power, and wireless connectivity open up the possibility of new applications, crit-ics [Ada00, Ara95, Luc99] argue that it is not yet clear how these possibilities are actually going to be put into practice: we are “brilliant on means, but pretty hope-less when it comes to ends” [Tha01]. However, this innovation dilemma – we may know how we can create incredible things, but we don’t know what needs they are supposed to meet – is only superficial. The following section describes potential economic benefits that ubiquitous-computing technology offers when it comes to industrial processes – benefits that will be a prominent driver for the proliferation of ambient intelligence, perhaps even more than the above-mentioned technologi-cal progress itself.
The Ambient Economy
“The most profound revolutions are not the ones trumpeted by pundits, but those that sneak in when we are not looking” [Wei93]. What Mark Weiser formulated over ten years ago accurately describes the current atmosphere surrounding the fields of ambient intelligence and ubiquitous computing. While personal gadgetry in the form of smartphones and Internet fridges continues to bedazzle the press, industry has quietly begun setting its sights on the enormous business potential that technologies such as wireless sensors, RFID tags, and positioning systems have to offer. Analysts call it the real-time economy or now-economy [Sie02a], where more and more entities in the economic process, such as goods, factories, and vehicles, are being enhanced with comprehensive methods of monitoring and information extraction. Ultimately, their whole lifecycle, beginning with the “birth” of their components and ending with their complete consumption (or recy-cling), can be witnessed (and, to some extent, even controlled) in real time.
Two important technologies form the core of these new economic processes
and applications: the ability to track real-world entities, and the introspection ca-pabilities of smart objects. Tracking objects in real-time allows for more efficient business processes, while objects that can monitor their own status via embedded sensors allow for a range of innovative business models.
Social, Economic, and Ethical Implications 5
Staying on Track: Smart Products
Inventory management obviously benefits from accurate, real-time information on the location and condition of goods, equipment, and manpower. If a company does not know the location and condition of its stock, and how long it has been in the warehouse, significant costs are incurred. Missed profits, oversized inventories, and the devaluation of goods depreciating in the warehouse are possible conse-quences of a lack of information. The stocktaking required for business or legal reasons also typically requires a considerable amount of effort. Stocktaking is not only expensive; it is inherently error-prone as well. A factory floor or warehouse equipped with technologies such as indoor localization and automatic identifica-tion can largely automate the stocktaking process, thereby reducing costs.
If several companies along a supply chain simultaneously use such precise in-
ventory data in addition to real-time order information, they can achieve additional savings by significantly attenuating the so-called “bullwhip effect” [Lee97]. This effect, often noticed in practice, describes the following phenomenon: although consumer demand for a product remains almost constant over time, small changes in this demand amplify along the supply chain and ultimately result in either ex-cess production (and associated storage costs) or sudden interruptions to supply (and associated missed sales). However, the more information transparency there is along the supply chain, the more these undesirable effects are attenuated [Jos00]. By making comprehensive information available along the supply chain, a significant reduction in the bullwhip effect can be achieved.
A further step towards the now economy is the constant monitoring of critical
product parameters (e.g., of temperature-sensitive goods such as chemicals or gro-ceries) by tiny wireless sensors. Equipped with communications capabilities, such “introspective” goods are not only able to monitor themselves, but can also com-municate relevant parameters to the outside world [Fle02]. Smart goods could ob-serve their condition while in transit and trigger an alarm in the event of excessive temperatures, which could – if appropriate – lead to an automatic reordering of damaged goods. Alternatively, the goods could also attempt to take corrective ac-tion, for example by controlling the temperature of their container: “As sensors improve and always-on connectivity becomes a reality, products will be able to do something about their condition” [Fer02]. In this way, “self-conscious” products (i.e., products that perceive their condition, analyze it, and attempt to change their situation if they are dissatisfied with it) would lower the total transaction costs by reducing the time necessary to procure replacements for damaged goods.
Anything, Anywhere: Innovative Business Models
The benefits of a world full of smart objects do not stop at the factory floor. Once comprehensive infrastructures for tracking goods and facilitating communication among “self-conscious” objects are in place, ambient intelligence throughout our environment – in private homes, cars, trains, and public places – would facilitate a range of new applications and business models. The prospect of their revenue
6 Social, Economic, and Ethical Implications
streams might become another substantial driver for the deployment of ambient-intelligence technologies in the near future.
“See a great sweater on someone walking by? Find out the brand and price, and place an order. Or maybe you'll be wearing the sweater and earning a commission every time someone near you sees and buys.” This vision [Fer02] describes a fu-ture, maybe not that far off, in which the boundary between the real world and the world of information has become blurred. Invisible tags embedded in most prod-ucts would allow consumer devices to read out the unique identification number of an item and use this to access the object’s virtual representation in the information world, which in turn could provide the user with a wealth of background informa-tion (e.g., ingredients, product reviews, etc.) as well as direct links to online (or even real-world) shops selling this item. People could shop on the move – on the streets, in buses, or whilst chilling out in their favorite bar at night – and every item sold could in turn become a new sales channel.
In a more proactive fashion, smart products could begin to subtly advertise
themselves, or even use cross-marketing to advertise their “friends.” A smart re-frigerator could, for example, recommend recipes based on both the groceries it contains, and on the items currently discounted at the local supermarket. It could also accrue reward points every time goods of a promoted brand are stored in it. With the help of ambient-intelligence technology and the detailed profiles it en-ables, prices of everyday goods could even be adjusted to suit individual custom-ers. In the vendor’s ideal world “each consumer would be quoted an individual price, which […] exactly corresponded to his readiness to pay” [SkS02] – econo-mists call this “perfect price discrimination.” Such individual prices have to be in-troduced carefully, however: the online-bookseller Amazon discovered that such perfection might not be universally appreciated, after their trial of individual DVD prices had to be suspended almost immediately due to massive customer criticism [Sie02, USA00].
The ultimate in shopping may be achieved when all the decision-making is re-
moved from people, and things do the shopping themselves (see the section on so-cial challenges below for a discussion of ubiquitous computing and consumer con-trol). The business consultancy Accenture has already coined a phrase for this – “silent commerce” [Acc01]. Their vision of “autonomous purchasing objects” not only includes photocopiers responsible for ordering their own paper, but also Bar-bie dolls1 that delight children (and their parents…) by ordering new clothes with their own pocket money: “Barbie detects the presence of clothing and compares it with her existing wardrobe – after all, how many tennis outfits does a doll need? The toy can buy straight from the manufacturer via the wireless connection. She
1 Apparently, the toy manufacturer Mattel, owner of the brand, was not at all happy about
this and sent in the lawyers right away – Accenture’s Web site now refers only to “a doll”, with no mention of Barbie.
Social, Economic, and Ethical Implications 7
can be constantly and anonymously shopping, even though the owner might not know it” [Mae02].
Apart from such rather questionable ways of increasing sales revenues, how-
ever, the “smartness” of objects might also facilitate trading at a more fundamen-tal level, by reducing information asymmetries [Ake70]: goods could not only “talk” about their price, ingredients, and availability, but also provide a detailed history of their production, use, and repair. A used car could give a detailed list of the parts that have been replaced or repaired over the course of its lifetime, thus reducing the amount of trust the buyer must have in the seller. Organic food could provide consumers with a comprehensive history of its cultivation, fertilization, and processing, potentially increasing their willingness to pay a premium for it. Obviously, in order for consumers to trust this information, it would either require verification from trusted third parties, or some sort of technical means to prevent tampering (much like odometers today). However, by increasing the overall mar-ket transparency, both for new and used goods, an environment supporting trust-worthy ambient-intelligence systems could potentially increase overall market ac-tivities by reducing the uncertainty inherent in many transactions today.
In a future full of ubiquitous computing, we may not only be tempted to buy (and have the ability to shop) just about anywhere at any time, we may also have to buy just about everywhere, all the time. Digital rights management systems that have recently been developed for distributing digital music and video are a first step in this direction. These systems make it possible for owners of digital content to ex-ert control over the access to digital information even after it has been sold – Buena Vista’s Disney home video unit, for example, recently started selling DVDs that render themselves unplayable 48 hours after unpacking [Reu03]. Digital rights management systems could even be programmed to require a continuous payment while listening to digital music or watching digital video. Such pay-per-use models have traditionally been used for phone calls or public utilities (e.g., electricity, gas), but could now – in an ambient-intelligence world – be imple-mented using everyday objects equipped with sensors and communications capa-bilities. Furniture, for instance, could monitor its usage (e.g., a sofa could count the number of persons that sit on it, the persons’ weight and seating time) and cre-ate a monthly itemized billing statement. While private homes might still prefer owning their beds rather then being billed for sleeping in them, corporate buyers with a high turnover of furniture (such as hotels or offices) could potentially pro-vide a sufficiently large customer base for such a business model to develop. However, as with the market transparencies described above, mechanisms would need to be in place in order to assure both parties that the sensor readings had not been tampered with.
8 Social, Economic, and Ethical Implications
No Risk, No Premium
Information asymmetry is not only a hindrance for trading, but also affects the risk management (moral hazard) of insurance companies. Typically, insurers use broad categories such as age or location in order to judge the risk involved in providing individual coverage, e.g., for health or car insurance. An ambient intelligence en-vironment, however, would allow for a much more fine-grained calculation of the individual risks involved. A smart car, for example, could provide detailed infor-mation about the driving style and parking habits of its owner, thus providing the insurer with a much better assessment of the likelihood of an accident or theft. For “safe” drivers, this would result in reduced insurance premiums as they would be exempted from paying for the insurer’s uncertainty regarding their driving style.
Insurance rates could even become entirely dynamic, reducing or increasing the
premium in a “pay-per-use” fashion in real time, e.g., during driving. Taken to the extreme, the current insurance rate, similar to today’s indicators showing current gas consumption, could be calculated on the spot and displayed in an “insurance meter”, based on the time of day, the route traveled, weather and road conditions, and of course driving style. Similar adjustments could be made for home owners’ insurance (new furniture automatically registers its net worth when placed inside the home) or health insurance (smoking a single cigarette would increase the rate, a walk in the park decrease it). Even in the case of an actual insurance claim, smart objects could significantly lower the insurer’s cost by observing the circum-stances of an accident or by noticing which items were actually missing after a break-in.
Obviously, such detailed activity records would constitute a serious privacy
hazard for many consumers. While technical remedies might be possible (e.g., data would be encrypted before being sent to the insurer, who would need an au-thorized key from the customer – or law enforcement – to read it in case of an ac-cident), both the increased complexity and the limited usefulness to the insurer of such encrypted information would most likely favor the privacy-threatening alter-natives. Even if customers had the choice of opting out, drivers who did not wish to pass their details on to their insurer would most likely have to pay a considera-bly higher premium, as the insurer’s risk would be spread among fewer and fewer non-participating customers. This would not only limit the desired freedom of choice, particularly for low-income drivers, but also implicitly accelerate the move away from fixed rates to cheaper dynamic (but privacy-invading) solutions.
An Economy on Autopilot
Despite the number of potential economic advantages described above, there are also substantial risks involved when relying on ubiquitous-computing technologies for large parts of an economy. The increasing automation of economically relevant aspects and the exclusion of humans as decision makers could certainly become a cause for concern. Under “normal” circumstances, automated control processes increase system stability – machines are certainly much better than humans if they
Social, Economic, and Ethical Implications 9
have to devote their whole attention to a particularly boring task. But situations that have not been anticipated in the software can easily have disastrous conse-quences if they are not directly controlled by humans (as tragic accidents involv-ing airplane autopilots have shown in the past). Other problems might arise from the intricate interplay of several automated processes, which might quickly esca-late into an unanticipated feedback loop that gets out of control. For example, the stock market crash of 1987 was partly caused by newly implemented trading soft-ware [Sie02a], which was designed in such a way that it would trigger the sale of shares whenever a certain pattern appeared in the daily fluctuations of their prices. Since a large number of traders were using the same software, the first appearance of this pattern caused a flood of sales, which in turn reinforced the software’s sell-ing pattern and thus triggered the crash.
Another risk stems from the increased efficiency of an economy supported by
ambient-intelligence technologies. While detailed tracking and inventory systems, along with smart, “self-conscious” products, allow a company to trim production and stock keeping as much as possible, this lack of “slack” in the production proc-ess also increases the risk that unforeseen interruptions can have grave conse-quences. In the case of supply chain management, for example, the attenuation of the bullwhip effect permits a reduction in storage capacity. However, if all com-panies along the delivery chain drastically reduce their stocks, one small unex-pected interruption in supply by the weakest member would immediately halt all production along the whole chain.
The automation and acceleration of the economy increases not only the poten-
tial for possible savings, but also influences the associated risks in these types of complex and sensitive systems. Nevertheless, the increased flexibility and control that a world of ambient intelligence offers will certainly be an important incentive for businesses to drive its deployment in the near future. Therefore, it is important that the ambient-intelligence landscapes we build should be both reliable and so-cially acceptable. We will come back to these issues in a later section. For now, we want to take a closer look at what probably constitutes the most visible impli-cation of an economy driven by ambient-intelligence technologies, namely its threat to personal privacy.
Privacy in Ambient Intelligence
Intelligent fridges, pay-per-use scenarios, and dynamic insurance rates paint a fu-ture in which all of our moves, actions, and decisions are recorded by tireless elec-tronic devices, from the kitchen and living room of our homes to our weekend trips in our cars. Not surprisingly, many critics see this as “an attempt at a violent technological penetration of everyday life” [Ara95], as the “feverish dream of spooks and spies – to plant a ‘bug’ in every object” [Tal00] or even as “a project that aims at totality and, of course, verges on the totalitarian” [Ada00].
By virtue of its very definitions, the vision of ambient intelligence has the po-
tential to create an invisible and comprehensive surveillance network, covering an
10 Social, Economic, and Ethical Implications
unprecedented share of our public and private life: “The old sayings that ‘the walls have ears’ and ‘if these walls could talk’ have become the disturbing reality. The world is filled with all-knowing, all-reporting things” [Luc99]. And with the eco-nomic possibilities described above, such a comprehensive coverage seems more likely to be put into place: shopping without participating in comprehensive profil-ing, buying instead of renting items in a pay-per-use scheme, as well as fixed in-surance schemes that do not constantly transmit information to insurers – all of this might become an expensive luxury for well-off citizens, while the population at large must trade in their privacy for increased productivity and market transpar-ency. This might very well be self-inflicted: given the immediate economic returns of consumer loyalty programs or low insurance rates, the rather vague threats of future privacy violations are easily enough ignored.2 The following sections try to add a differentiated view to this problem, especially with respect to ubiquitous-computing technology, by first examining why personal privacy is desirable, de-scribing when we feel that it has been violated, and then assessing how the de-ployment of future ambient-intelligence systems will affect all that.
The Many Facets of Personal Privacy
Even though critics continue to argue that “all this secrecy is making life harder, more expensive, dangerous and less serendipitous” [Coc00], privacy is still pre-dominantly seen as a fundamental requirement of any modern democracy [Rot01]. It is only when people are free to decide what to do with their lives, according to their interests and beliefs, and without fear of repression from their fellow citizens, that the necessary plurality of ideas and attitudes can develop that will prevent so-ciety being subjugated under a charismatic leader. Harvard law professor Law-rence Lessig [Les99] takes this requirement a step further and distinguishes be-tween a number of motives for the protection of privacy in today’s laws and standards:
• Privacy as Empowerment. Seeing privacy mainly as informational privacy, its
aim is to give people the power to control the publication and distribution of in-formation about themselves [Wes67]. A recent legal discussion surrounding this motivation revolved around the question of whether personal information should be seen as private property (which would entail the right to sell all or part of it as the owner sees fit) or as intellectual property (which would entitle the owner to certain inalienable rights, preventing him for example from selling the rights to his own name to anybody else).
• Privacy as Utility. From the viewpoint of the person involved, privacy can be
seen as a utility providing more or less effective protection against nuisances
2 The continuing media attention that privacy issues are receiving has nevertheless
prompted industry to tread carefully: “A very cautious approach is needed [.] with this kind of monitoring, otherwise newspaper headlines such as ‘Spy in the Kitchen’ would soon appear, killing the intelligent appliance before it takes off” [IGS01].
Social, Economic, and Ethical Implications 11
such as unsolicited phone calls or emails. This view probably best follows Brandeis’ definition of privacy as “The right to be left alone,” where the focus is on minimizing the amount of disturbance for the individual [WaB1890].
• Privacy as Dignity. Dignity not only entails being free from unsubstantiated
suspicion (for example being the target of a wire tap, where the intrusion is usually not directly perceived as a disturbance), but also focuses on the equilib-rium of information available between two people: as in a situation where you are having a conversation with a fully dressed person when you yourself are naked, any relationship where there is a significant information imbalance will make it much more difficult for those with less information about the other to keep their composure.
• Privacy as a Regulating Agent. Privacy laws and moral norms to that extent can
also be seen as a tool for keeping checks and balances on the powers of a deci-sion-making elite. By limiting information gathering of a certain type, crimes or moral norms pertaining to that type of information cannot be effectively con-trolled.
Depending on what kind of motives one assumes for preserving privacy, ambi-
ent-intelligence technology can become the driving factor for changing the scope and impact of privacy protection as it exists today, and creating substantially dif-ferent social landscapes in the future. This is because ambient-intelligence tech-nology influences two important design parameters relating to privacy: the ability to monitor and the ability to search [Les99].
Ambient Intelligence and Surveillance
The conscious observation of the actions and habits of our fellow men is as old as mankind itself. However, observations using automated systems differ from our nosy neighbors in one important aspect: while in the “good old days”, anything out of the ordinary would attract the attention of our fellow citizens, it is now the ordinary, the everyday routine, that can be (and often is) the sole focus of tireless computerized monitoring. With ambient–intelligence technologies, today’s moni-toring capabilities can obviously be extended far beyond credit-card records, call logs, and news postings. Not only will the spatial scope of such monitoring activi-ties be significantly extended in ambient-intelligence landscapes, but their tempo-ral coverage will also greatly increase: starting from pre-natal diagnostics data stored on babies’ hospital smart cards, to activity patterns in kindergarten and schools, to workplace monitoring and senior citizen's health monitoring.
Such comprehensive monitoring (or surveillance) techniques create new oppor-
tunities for what MIT professor emeritus Gary T. Marx calls border crossings: “Central to our acceptance or sense of outrage with respect to surveillance . are the implications for crossing personal borders” [Mar01]. He goes on to define four such border crossings that form the basis for perceived privacy violation:
12 Social, Economic, and Ethical Implications
• Natural Borders. Physical borders of observability, such as walls and doors,
clothing, darkness, and also sealed letters and phone conversations. Even facial expressions can represent a natural border against the true feelings of a person.
• Social Borders. Expectations with regard to confidentiality in certain social
groups, such as family members, doctors, and lawyers. This also includes the expectation that your colleagues do not read personal fax messages addressed to you, or material that you leave lying around the photocopier.
• Spatial or Temporal Borders. The expectation by people that parts of their lives
can exist in isolation from other parts, both temporally and spatially. For exam-ple, a previous wild adolescent phase should not have a lasting influence on the current life of a father of four, nor should an evening with friends in a bar influ-ence his coexistence with work colleagues.
• Borders due to Ephemeral or Transitory Effects. This describes what is best
known as a “fleeting moment,” a spontaneous utterance or action that we hope will soon be forgotten, or old pictures and letters that we put out in our trash. Seeing audio or video recordings of such events subsequently, or observing someone sifting through our trash, would violate our expectations of being able to have information simply pass away unnoticed or forgotten.
Putting ambient-intelligence systems into place will most certainly allow far
greater possibilities for such border crossings in our daily routines. Consider the popular vision of a wearable “memory amplifier” [May01, Rho97], allowing its wearer to constantly record the events of her daily life in a lifetime multimedia di-ary. While at first sight such technology promises assistance to those of us who frequently tend to forget small details, it also has substantial consequences for our privacy borders stemming from ephemeral and transitory effects: any statement I make during a private conversation could potentially be played back if my conver-sation partner gave others access to her multimedia diary. Even if this information were never disclosed to others, the very thought of dealing with people who have a perfect memory (and thus would never forget anything) would probably have a considerable effect on our interpersonal relationships.
The problem of spatial and temporal borders, on the other hand, is well known
from the field of consumer profiles. Although such profiles are often the subject of public debate, the social and legal attitudes towards them have, until now, been relatively relaxed. Consumer acceptance is also much higher than the frequent negative news coverage might indicate, mostly because their negative conse-quences are often perceived as being rather minor (such as unsolicited spam) com-pared to their advantages (e.g., monetary incentives in the form of discounts or rewards). However, there are well-known risks associated with profiles, and their adoption as the basis for ambient intelligence would only exacerbate such prob-lems. Besides the obvious risk of accidental leaks of information [OHJ01], pro-files also threaten universal equality, a concept central to many constitutions, basic laws, and human rights, where “all men are created equal” [USA98]. Even though an extensively customized ambient-intelligence future where I only get the infor-mation that is relevant to my profile holds great promise, the fact that at the same time a large amount of information might be deliberately withheld from me be-
Social, Economic, and Ethical Implications 13
cause I am not considered a valued recipient of such information, would constitute a severe violation of privacy for many people.
Applying ambient-intelligence technology in areas with primarily social bor-ders – for example where a close social group interacts only among itself, such as families [NKO01, WLS01] or co-workers – might appear to alleviate some of the above concerns. Most participants would already share close relationships and tend to know a great deal about each other, without needing a system to compile a profile of their communication partner. Such systems, however, also raise the ante as to what type of information they handle. While a communication whiteboard for families may facilitate social bonding between physically and temporally sepa-rated members, it would also increase the risk of unwanted social border crossings by accidentally allowing Mum to read a message you left for your sister, or a visit-ing friend to be recorded in the house activity log even though you told grandma you would spend the weekend alone.
Natural borders, then, might be the easiest to respect when designing ambient-
intelligence systems. Here, the concept of surveillance is well known and usually fairly straightforward to spot, after all: if others are able to watch your actions be-hind closed doors, they are most certainly intruding on your privacy. Proponents of wearable computing systems often cite the fact that information could both be gathered and stored locally (i.e., on the user’s belt, or within her shirt) as a turnkey solution for privacy-conscious technologists [Rho99]. Border crossings, however, are not only about who does something, but also what is happening. Even though a context-aware wearable system might keep its data to itself, its array of sensors nevertheless probe deep into our personal life, and the things they might find there could easily startle (and trouble) us, once such systems start anticipating our future actions and reactions. The feeling of having someone (or something) constantly looking over our shoulder and second-guessing us would certainly constitute a natural border crossing for most of us. And the temptation of law enforcement subpoenaing such information not only to determine your physical data (were you at the crime scene?) but also to guess your intentions (by assessing the data feed from our body sensors) would certainly motivate legislation that would make the deletion of such information a crime (just as recent legislation against cybercrime [CoE01] does for computer log files).
The Power of Searching and Combining Information Bits
All these examples serve to show that ambient-intelligence landscapes, even when created for the greater good and with the best of intentions, will run a high risk of involuntarily threatening the personal borders that separate our private and public lives, simply because their monitoring capabilities will facilitate more of the bor-der crossings described above. However, whether such crossings ultimately occur, given the opportunities created, will depend very much on the type of searching capabilities such ambient-intelligence systems might offer.
Search technology is traditionally a topic in the fields of information retrieval
or databases, rather than that of ambient intelligence. However, the chances are
14 Social, Economic, and Ethical Implications
high that such technology will be a basic building block of future ambient-intelligence landscapes, as many of the envisioned applications in the field require precisely these capabilities. An automated diary collecting 24/7 audio and video data would not be much use unless it was combined with powerful search and re-trieval technology that allowed us to comb large amounts of data for very specific information. And the ability to combine different information sources, especially large, innocuous ones such as walking patterns or eating habits, is the backbone of any “smart” system, which must make the best use of a large variety of different sensor input to take decisions that make it appear to understand what is happening around us.
Having thus both monitoring and search capabilities at the very core of their ar-
chitecture, ambient-intelligence systems will very likely provide their developers, owners, and regulators with a significant tool for driving the future development of privacy concepts within society. Depending on the actual systems deployed, some of the motivating aspects of privacy as discussed above might become more or less prominent, thus influencing corresponding legal and social norms. How-ever, as important as privacy is, it is merely the tip of the iceberg that constitutes the social implications of ambient intelligence. In the following section we want to explore a bit more of the edges that lurk just below the waterline, where they seem to be of no immediate threat but where they could potentially become much more dangerous as the deployment of ambient intelligence picks up speed.
Social Challenges and Implications
Life without computers is unimaginable for most of us today – embedded proces-sors monitor the condition of high-risk patients around the clock, they control cen-tral heating in buildings, air conditioning in tunnels, and they safely guide air-planes between continents. The potential economic benefits of ubiquitous computing are certainly key factors for the further proliferation of information technology, such as novel indoor and outdoor positioning systems, ubiquitous communication platforms, and unobtrusive monitoring installations. This technol-ogy will form and shape the foundations of future ambient-intelligence land-scapes.
As more and more objects and environments are being equipped with ambient-
intelligence technology, the degree of our dependence on the correct, reliable functioning of the deployed devices and microcomputers including their software infrastructures is increasing accordingly. Today, in most cases, we are still able to decide for ourselves whether we want to use devices equipped with modern com-puter technology (e.g., by choosing manual control for our central heating, or by deciding not to carry a mobile phone if we dislike the constant accessibility its us-age implies). But in a largely computerized future, it might not be possible to es-cape from this sort of technologically induced dependence, which leads to a num-ber of fundamental social challenges for future ambient-intelligence systems. Privacy is just one of these challenges, though probably the most prominent one.
Social, Economic, and Ethical Implications 15
However, the more thoroughly “computerized” our environment becomes, the more basic attributes of the world we live in will subtly change, such as its reli-ability, accessibility, and transparency. In the following, we attempt to identify these concerns and try to address additional ethical and social implications of fu-ture ambient-intelligence landscapes.
The vision of ambient intelligence describes systems that work completely in the background, discreetly and unobtrusively helping us to carry out our tasks. Since our needs and circumstances can change over time, such systems must be able to adapt themselves dynamically to the current situation. In doing so, one crucial ba-sic requirement is reliability in the broadest sense of the word. In addition to en-suring dependability from a technological point of view, a complex and highly dynamic system must also remain manageable and controllable, and must retain the ability to predict (and, to a certain extent, verify) that the system is behaving correctly:
• Manageability. It is far from clear that implementing large-scale ambient-
intelligence scenarios involving potentially millions of smart, adaptive devices, is simply a question of scaling up existing toy examples. Will these services and applications still be able to meet their original requirements, even with a massive increase in the number of tiny interacting objects? And, above all, how will we be able to understand and control such a highly dynamic world involv-ing such large numbers of individual objects?
• Predictability. Today’s technical infrastructures, such as the phone system,
television, and electricity, are relatively easy to use, even for people with no special qualifications. This also entails the ability to detect malfunctions: for example, if you lift a telephone receiver and do not hear a dial tone, it is imme-diately evident that the phone (either the handset or the landline) is not working properly. However, this type of predictability of system behavior can no longer be taken for granted in an ambient-intelligence landscape, as systems are ex-pected to function without users noticing their presence. This will make fault detection and diagnosis fundamentally difficult, especially for the layman [ECP02]. Additionally, users might continue to rely on a failed service (e.g., an automated backup service or the self-diagnostics of a smart product) without noticing, thus increasing the damage done until the problem is finally discov-ered.
• Dependability. Incorporating computing and communication technology into
everyday artifacts requires ever-decreasing form factors and minimal energy consumption. This makes it difficult to use hardware redundancy in such sys-tems, even though their envisioned unobtrusive and ubiquitous use implies much harsher surroundings than, say, an everyday indoor environment. This calls for alternative concepts and mechanisms in order to overcome service in-terruptions and device failures, such as an explicit diversification of system
16 Social, Economic, and Ethical Implications
functions. Such a diversification can be achieved by providing fully independ-ent ways of carrying out the same task, preferably based on separate sets of sys-tem resources wherever feasible. A communications connection, for instance, can be diversified if the system provides different communications mechanisms in parallel, such as GSM, Bluetooth, and wireless LAN.
The power outages that affected not only large parts of the USA and Canada
but also Italy and some other countries in 2003 have demonstrated our dependence on existing technical infrastructures, in this case the power grid. With the constant goal of saving costs, any industry-built ambient-intelligence infrastructure will run a high risk of forgoing safety for the sake of efficiency, resulting in brittle systems that will work only sporadically.
Delegation of Control
In order to minimize the need for human intervention in complex, highly dynamic environments, new concepts for delegating controlare necessary – automatic processes should take care of routine tasks in a dependable manner, but also pro-vide accounting mechanisms for monitoring complex control flows. Control and accounting mechanisms are important tools for determining who is in control of an autonomous system, and who is responsible if something goes wrong. At the same time, however, the autonomy of artifacts is also limited by their reliance on the technical infrastructure:
• Content Control. If smart objects provide information about themselves, this
raises the question of who guarantees the objectivity and accuracy of the state-ments made. For example, smart products might be used to tie customers more closely to traders by recommending they purchase other goods produced by that same trader. In a certain sense, smart objects are becoming media representing a particular “ideology” (e.g., that of the product’s manufacturer, or the politi-cally motivated opinion of a consumer protection organization). Could a con-sumer protection institute use its own electronic directory to map a smart prod-uct label onto information other than that which the producer intended (for example, to warn of allergies to ingredients)? And maybe more importantly: who will decide what a smart toy tells the children, potentially shaping the chil-dren’s opinions without their parents’ knowledge? Tempting children to buy additional toys would only be the most obvious strategy – a much more serious threat would be the moral values induced by smart toys during play.
• System Control. It is similarly conceivable that automobiles or other products,
as components of an ambient-intelligence network, would no longer feel com-pletely “loyal” to their owners, but would instead enforce the guidelines of in-surance companies, manufacturers, or the judiciary. For example, a smart car might refuse to open the door for its driver because he or she has stopped in a no-parking zone. But when should an intelligent device obey human orders, and when should it follow its own “convictions”? While such a no-parking system
Social, Economic, and Ethical Implications 17
might be desirable for congested cities, some kind of manual override mecha-nism would obviously be needed for emergency situations, e.g., when rushing a seriously injured person to hospital (and trying to park in front of it). Even if a system were designed to only make suggestions, it would still find itself tread-ing a fine line between inspiration and frustration, between obliging helpfulness and pig-headed patronization [Sat01].
• Accountability. If autonomous objects such as the previously mentioned smart
doll start taking decisions on their own (e.g., buying new clothes), legal guide-lines need to be drawn up in order to resolve who is ultimately responsible for these business transactions. Smart assistants might order unwanted plane tick-ets, smart fridges excessive amounts of food – in both cases the automated sys-tem might have performed according to specification, though neither the origi-nal programmers nor the layman user would be able to understand its reasoning. Providing the user with a detailed explanation of completed transac-tions is only part of the solution, especially when monetary damages are in-volved. It may look appealing to simply shift the responsibility and liability onto the end user by changing the license agreements of smart objects accord-ingly in the small print, but it is questionable whether such a procedure would prove tenable if taken to court.
Similar discussions involving the questions of accountability and content con-
trol are already taking place in the context of the World Wide Web. For example, questions regarding the right to possess and use certain prestigious domain names [Bit00, DNH03] can be compared to the issue of content control (i.e., who is al-lowed to resolve a certain URL stored in the RFID tag of a product), while na-tional laws trying to control digital copyright as well as freedom of speech [EFF03] might already set standards regarding the future “freedom” of smart de-vices to obey their owners.
Another fundamental challenge for ambient-intelligence systems is their social compatibility. If we, as humans, want to be capable of participating in highly dy-namic systems, their parameters will have to be adjusted accordingly. System be-havior relating to particular aspects should retain a certain transparency and iner-tia, allowing humans to detect and adjust to changes. On the other hand, it should also be taken into account that an all-encompassing ambient-intelligence land-scape must also meet the needs and requirements of as broad a section of society as possible, especially if participation is practically mandatory.
• Transparency. With the pay-per-use model discussed above, perusing an ambi-
ent-intelligence environment might incur a large number of micropayments, e.g., for bus or theater seats we have sat on, pages of books or newspapers we have read, or clothing we have worn. Irrespective of technical feasibility, this prompts the question of how we could keep track of the resulting number of short-term contracts and the countless associated micropayments, let alone ret-
18 Social, Economic, and Ethical Implications
rospectively check the legitimacy of these transactions. Not only would it be extremely tedious and unrealistic to manually check thousands of transactions, it is also questionable to what extent inappropriate items could be identified and rejected, and to what extent legitimate payments could be unambiguously and indisputably allocated to the responsible party. Dynamic insurance rates that vary according to the style of driving (as discussed in an earlier section) consti-tute another example of a potential loss of transparency, especially if the under-lying assessment methods changed dynamically with no warning, or if they were unknown or too complicated to be understood by the user.
• Knowledge Sustainability. Most information in our everyday life today remains
valid for an extended period of time, e.g. food prices in our favorite supermar-ket, or prices for public transport. It is this inertia of information that permits us to use acquired knowledge and prior experiences to cope with future situations and tasks. In a highly dynamic world, the sustainability of knowledge risks be-ing lost – an experience that was valid and useful one minute could become ob-solete and unusable the next. Such a loss or accelerated devaluation of long-term experiences could, in the long term, contribute to an increased uncertainty and lack of direction for people in society.
• Fairness. Detailed cross-marketing based on ambient intelligence promises tai-
lor-made offers that virtually eliminate unwanted advertisements. However, a specific offer may be withheld from a particular consumer for one of two rea-sons: either the offer was not worth the consumer, or the consumer was not worth the offer. David Lyon, Professor of Sociology at Queen’s University in Canada, calls this process “social sorting” – “Categorizing persons and groups in ways that appear to be accurate and scientific, but which in many ways ac-centuate differences and reinforce existing inequalities” [Lyo01]. People not matching a certain ‘desirable’ profile might have to pay much higher prices, as they do not qualify for any of the existing discounts, which might in turn rein-force the non-matching patterns.
• Universal Access. The natural interfaces envisioned in ambient-intelligence
scenarios certainly have the opportunity to overcome many of today’s accessi-bility problems, such as the small screens and keypads of modern mobile phones that often prevent elderly people from using them [Gon01]. Many pro-jects in the field target elderly and physically disabled people in particular, for example with electronic “memory aids,” reading aids and navigation systems [Mak01], which might pave the way for a universal design [Ste01] that consid-ers the needs of minorities and marginal groups early on in the design stage. In-telligent interfaces and the concept of ubiquitous information access are often seen as key developments for bridging the digital divide, where different sec-tions of the population have different abilities to participate in the information society. However, having more information opportunities does not necessarily mean more justice or freedom, simply because the potential dependencies and opportunities for manipulation would be so numerous they could overwhelm individuals, making it even more difficult to assess the trustworthiness of the information’s source. Information that was uncritical or sponsored by advertis-ers (and therefore one-sided) could become available free of charge, while in-
Social, Economic, and Ethical Implications 19
dependent, high-quality information would cost money, thus widening the digi-tal divide even further. Since ubiquitous computing is not just about informa-tion itself, but is inherently linked to real-world objects, these new means of ac-cess and content control could easily lead to the digital divide becoming a real and perceivable rift in our everyday lives.
In history, the development of regulatory, social, and ethical standards tends to
lag considerably behind the rapid proliferation of pioneering technological inven-tions, as was the case with the invention of the assembly line and mass production at the beginning of the 20th century, and with the appearance of the global Internet in the 1980s, for example. In an emerging future of ambient-intelligence systems, one exciting question is whether we will be aware of the impending pitfalls and tackle them in an early (design) phase, where we still have the means to shape the envisaged systems according to fundamental social and ethical requirements, or if there is a need for yet another social revolution that subsequently brings about necessary adaptations by force.
The fundamental paradigm of ambient intelligence, namely that computers disap-pear from the user’s consciousness and recede into the background, is sometimes seen as an attempt to have technology infiltrate everyday life unnoticed by the general public in order to circumvent any possible social resistance [Ara95]. Yet beyond any perceived sinister motives (which might be easy enough to counter), a widespread public acceptance of ambient intelligence also rests on issues of an almost philosophical nature, such as the fundamental nature of smart objects or our changing relationship with our environment.
• Feasibility and Credibility. Many philosophers and social scientists identify a
prevailing self-confident and technophile attitude among scientists in the field of ambient intelligence and ubiquitous computing [Ada03], where the non-critical anticipation of future technological developments almost attains the characteristics of a metaphysical prophecy. Others doubt the credibility of the envisioned scenarios, e.g., when ambient intelligence is said to simplify our lives, help us save time, and relieve us of laborious tasks. While this assertion has been constantly repeated throughout the twentieth century by the consumer goods industry, adding “smart machines” everywhere will not help to overcome the existing pattern of hurry, rush, stress, and separation from other people, but will only increase their efficiency [Win99]. Such criticism may build up and induce a serious credibility gap, reducing the acceptability of ubiquitous com-puting technologies.
• Artifact Autonomy. Networked everyday objects embedded in an ambient-
intelligence landscape lose part of their autonomy and, with this, exhibit an in-creased dependence on the infrastructure. For users, this reduces the “object constancy” of the objects that surround them, as the example of electronic
20 Social, Economic, and Ethical Implications
books made from smart paper shows: reading such a book may presuppose a regular connection to a server (license server, accounts server, etc.). Because of this, it appears to be more error-prone and less autonomous than a “normal” book, which can always be read, whereas the electronic one can only be read if the infrastructure is functioning.
• Impact on Health and Environment. It is hard to predict the impact that a large-
scale use of ubiquitous computing and communication technology would have on our environment in terms of raw material consumption, energy consump-tion, and disposal. For example, if all supermarket goods were equipped with smart labels in the future, billions of these tiny and individually quite harmless chips would end up in the household garbage. On the other hand, the remote identification capability provided by smart labels would enable information on products to be made available throughout their entire lives, permitting the dif-ferent materials in waste products to be efficiently identified and separated. It is also not yet fully understood whether, and to what extent, electromagnetic ra-diation (e.g., produced by wirelessly communicating smart objects) could affect our physical health. A vision involving myriads of everyday objects and wear-able “information appliances” that communicate wirelessly with each other thus gives due cause for concern, as its potential adverse environmental effects could permanently influence the lives of future generations [HBB03].
• The Relationship between Man and the World. From a philosophical point of
view, the vision of ambient intelligence fundamentally changes the environ-ment in which we live: “By this weaving of extensions of ourselves into the surroundings, significant parts of the environment lose important aspects of their otherness and the environment as a whole tends to become more and more a subservient ‘artifact’. This artifact, which the world immediately surrounding us becomes, is almost entirely ‘us’ rather than ‘other’. In this sense, the sur-rounding world has almost disappeared.” [Ara95]. Similarly, Adamowsky stipulates that our inability to handle the physical world in a flexible enough way will force us to replace it by digital surrogates – equivalents of particular aspects of the real world in the digital world, implemented in the form of mod-els, simulations, and virtual counterparts – which will ultimately lead to a trans-formation, dislocation, substitution, and the loss of fundamental properties re-lating to the world [Ada00].
Dryer et al. [DEA99] conducted two empirical studies to examine the theoreti-
cal relationships between system design for mobile computing, human behaviors, social attributions, and interaction outcome. In their conclusion, they express “doubt that our inevitable future is to become a machinelike collective society. How devices are used is not determined by their creators alone. Individuals influ-ence how devices are used, and humans can be tenaciously social creatures.” They conclude “Given the importance of social relationships in our lives, we may adopt only those devices that support, rather than inhibit, such relationships.” With the substantial amount of skepticism related to technology, such findings seem to counterbalance the immediate threat that a thoroughly computerized future ap-pears to hold. However, apart from personal prejudices, the wide range of social
Social, Economic, and Ethical Implications 21
consequences that ambient intelligence may have will certainly need to be ad-dressed in future systems and debates. These challenges are of fundamental impor-tance and may ultimately even have a decisive influence on the large-scale accep-tance of ambient-intelligence technologies and environments.
Brave New World?
“Everything will be connected to everything else,” but “no one has any idea what all those connections will mean” [Luc99]. This criticism can be taken as a per-ceived lack of focus when it comes to ambient-intelligence applications, but also as a deficiency in terms of understanding the consequences of deploying ubiqui-tous-computing systems in the real world: how will we use “smart things” in our everyday lives? When should we switch them on or off? What should smart things be permitted to hear, see, and feel? And whom should they be allowed tell about it? Whether these consequences concern the protection of personal data, the impli-cations for the macro-economy, or social acceptance – developers of ubiquitous computing systems can profit greatly from a careful evaluation of the conse-quences of such technology within the framework of established concepts from the fields of sociology, economics, and jurisprudence.
Although predicting the future is difficult, if not impossible, the above discus-
sion allows us to guess at a few of the possible implications of wide-scale use of ambient-intelligence technology: new business models will increase profits, possi- bly at the expense of safety margins; the balance of political and economic power could shift; economic developments will accelerate and initiate long-term changes in our social values and motives; personal borders could be violated by new sur- veillance and search technology; and, not least, there is the danger that we will lose confidence in our environment, thus fundamentally and unfavorably changing our attitude towards the world that surrounds us. The intention of this article was to throw light on the interdisciplinary fields of ubiquitous computing and ambient intelligence, in order better to understand how far these visions can and should in- fluence our everyday lives. By identifying and addressing the great challenges of technical and social change, as well as their environmental sustainability, it may be possible to steer this development in a direction that has more in common with Weiser’s optimistic vision of the 21st century than with the depressing mix of consumer terror and police state conjured up by Steven Spielberg in his movie “Minority Report” [Dic56]. Acknowledgements. This work has been partially funded by the Gottlieb Daimler- and Karl Benz-foundation, Ladenburg, Germany, as part of the interdisciplinary research pro- ject “Living in a Smart Environment – Implications of Ubiquitous Computing”. The authors also wish to thank Jochen Jagob, TU Darmstadt, Germany, for fruitful discussions on the economic issues.
22 Social, Economic, and Ethical Implications
[ABS01] Abowd GD, Brumitt B, Shafer S (Eds) (2001) Proceedings of Ubicomp 2001, At-
lanta GA, USA, September 2001. Springer-Verlag
[Acc01] Accenture Technology Labs (2001) Silent Commerce. Available at
[Ada00] Adamowsky N (2000) Kulturelle Relevanz. Ladenburger Diskurs „Ubiquitous
Computing”, February 2000. Available at www.inf.ethz.ch/vs/events/slides/ adamowldbg.pdf
[Ada03] Adamowsky N (2003) Smarte Götter und magische Maschinen – zur Virulenz
vormoderner Argumentationsmuster in Ubiquitous-computing-Visionen. In: Mattern F [Mat03], pp 231–247
[Aho01] Ahola J (2001) Ambient Intelligence, ERCIM News, No 47, October 2001 [Ake70] Akerlof G (1970) The Market for Lemons: Qualitative Uncertainty and the Market
Mechanism, The Quarterly Journal of Economics, 84(3), pp 488–500
[Ara95] Araya AA (1995) Questioning Ubiquitous Computing. In: Proceedings of the 1995
ACM 23rd Annual Conference on Computer Science. ACM Press, 1995. Available at doi.acm.org/10.1145/259526.259560
[Bit00] Bitlaw.com (2000) Some well publicized examples of domain names disputes. See
[Coc00] Cochrane P (2000) Head to Head. Sovereign Magazine, pp 56–57, March 2000 [CoR03] Coroama V, Röthenbacher F (2003) The Chatty Environment – Providing Every-
day Independence to the Visually Impaired. In: The 2nd International Workshop on Ubiquitous Computing for Pervasive Healthcare Applications (UbiHealth2003), Octo-ber 2003. Available at www.inf.ethz.ch/vs/publ/papers/ubicomp2003-hc.pdf
[CoE01] Council of Europe (2001) Convention on Cybercrime. ETS No 185. Available at
conventions.coe.int/Treaty/EN/cadreprincipal.htm, November 2001
[DEA99] Dryer DC, Eisbach C, Ark WS (1999) At what cost pervasive? A social comput-
ing view of mobile computing systems. IBM Systems Journal, 38(4):652–676
[Dic56] Dick PK (1956) Minority Report. Fantastic Universe, January 1956 [DNH03] Domain Name Handbook (2003) Domain Dispute Index. See
[ECP02] Estrin D, Culler D, Pister K, Sukhatme G (2002) Connecting the Physical World
with Pervasive Networks. IEEE Pervasive Computing – Mobile and Ubiquitous Sys-tems, 1(1):59–69, January 2002
[EFF03] Electronic Frontier Foundation EFF (2003) Blue Ribbon Campaign for free speech
[Fer02] Ferguson GT (2002) Have Your Objects Call My Objects. Harvard Business Re-
[Fin03] Finkenzeller K (2003) RFID-Handbook. John Wiley & Sons, 2nd edition [Fle02] Fleisch E (2002) Von der Vernetzung von Unternehmen zur Vernutzung von
Dingen. In: Schögel, Marcus, Torsten Tomczak and Christian Belz (Eds.): Roadm@p to E-Business – Wie Unternehmen das Internet erfolgreich nutzen, Thexis, St. Gallen, pp 124–135
[Gon01] Gonçalves DJ (2001) Ubiquitous Computing and AI Towards an Inclusive Soci-
ety. In: Heller et al [HJG01], pp 37–40
Social, Economic, and Ethical Implications 23
[HBB03] Hilty L, Behrendt S, Binswanger M, Bruinink A, Erdmann L, Fröhlich J,
Köhler A, Kuster N, Som C, Würtenberger F (2003) Das Vorsorgeprinzip in der Informationsgesellschaft. Auswirkungen des Pervasive Computing auf Gesundheit und Umwelt. Studie des Zentrums für Technologiefolgen-Abschätzung TA-SWISS, TA 46/2003, August 2003
[HJG01] Heller R, Jorge J, Guedj R (ed) (2001) Proceedings of the 2001 EC/NSF Work-
shop on Universal Accessibility of Ubiquitous Computing: Providing for the Elderly, Alcácer do Sal, Portugal, May 2001, ACM Press. See also virtual.inesc.pt/wuauc01/
[IGS01] IBM Global Services (2001) Transforming the appliance industry. Switching on
revenue streams in services. White Paper, 2001
[Jos00] Joshi YV (2000) Information Visibility and its Effect on Supply Chain Dynamics.
[Lee97] Lee HL, Padmanabhan V, Whang S (1997) The Bullwhip Effect in Supply Chains.
MIT Sloan Management Review, 38(3):93–102, Spring 1997
[Les99] Lessig L (1999) Code and Other Laws of Cyberspace. Basic Books, New York
[Luc99] Lucky R (1999) Everything will be connected to everything else. Connections.
IEEE Spectrum, March 1999. Available at www.argreenhouse.com/papers/rlucky/ spectrum/connect.shtml
[Lyo01] Lyon D (2001) Facing the Future: Seeking Ethics for Everyday Surveillance. Eth-
ics and Information Technology, 3(3):171–180, July 2001
[Mae02] Maeder T (2002) What Barbie Wants, Barbie Gets. Wired Magazine, 10(1), Janu-
[Mak01] Makris P (2001) Accessibility of Ubiquitous Computing: Providing for the Eld-
[Mar01] Marx GT (2001) Murky Conceptual Waters: The Public and the Private. Ethics
and Information Technology, 3(3):157–169, July 2001
[Mat03] Mattern F (ed) (2003) Total Vernetzt: Szenarien einer informatisierten Welt.
[Mat03a] Mattern F (2003) Ubiquitous Computing: Scenarios for an informatized world.
[May01] Mayo RN (2001) The Factoids Project. July 2001. Available at
[Moo65] Moore GE (1965) Cramming more components onto integrated circuits. Electron-
[NKO01] Nagel KS, Kidd CD, O'Connell T, Day A, Abowd GD (2001) Family Intercom:
Developing a Context-Aware Audio Communication System. In: Abowd GD et al [ABS01], pp 176–183
[OHJ01] O'Harrow Jr R (2001) Prozac Maker Reveals Patient E-Mail Addresses. The
[Reu03] Reuters (2003) This DVD Will Self-Destruct. Wired News, May 16, 2003. Avail-
able at www.wired.com/news/technology/0,1282,58883,00.html
[Rho97] Rhodes B (1997) The Wearable Remembrance Agent: A System for Augmented
Memory. Personal Technologies Journal. Special Issue on Wearable Computing, 1:218–224, January 1997
[Rho99] Rhodes B, Minar N, Weaver J (1999) Wearable Computing Meets Ubiquitous
Computing – Reaping the Best of Both Worlds. In: Proceedings of the Third Interna-
24 Social, Economic, and Ethical Implications
tional Symposium on Wearable Computers (ISWC '99), San Francisco CA, October 1999, pp 141–149
[Rot01] Rotenberg M (2001) Testimony and Statement for the Record. Hearing on Privacy
in the Commercial World before the Subcommittee on Commerce, Trade, and Con-sumer Protection, U.S. House of Representatives, March 2001. Available at www.epic.org/privacy/testimony_0301.html
[Sat01] Satyanarayanan M (2001) Pervasive Computing: Vision and Challenges. IEEE Per-
sonal Communications, 8(4):10–17, August 2001
[Sie02] Siegele L (2002) Tante Emma lebt. Die Zeit, (42):27, October 2002 [Sie02a] Siegele L (2002) How about now? A survey of the real-time economy. The
Economist, 362(8257):3–18, January 2002
[SkS02] Skiera B, Spann M (2002) Preisdifferenzierung im Internet. In: Schögel M,
Tomczak T, Belz C (eds) (2002) Roadm@p to E-Business – Wie Unternehmen das Internet erfolgreich nutzen. Thexis, St. Gallen, pp 270–284
[Ste01] Stephanidis C (2001) Towards Universal Access in the Information Society.
[Tal00] Talbott S (2000) The Trouble With Ubiquitous Technology Pushers, or: Why We'd
Be Better Off without the MIT Media Lab. NetFuture: Technology and Human Re-sponsibility. Available at www.netfuture.org/2000/Jan0600_100.html#3
[Tha01] Thackara J (2001) The design challenge of pervasive computing. Interactions,
8(3):46–52, May 2001. Available at doi.acm.org/10.1145/369825.369832
[USA00] USA Today (2000) Amazon May Spell End for 'Dynamic' Pricing, September
2000. Available at www.usatoday.com/life/cyber/tech/cti595.htm
[USA98] The Declaration of Independence and the Constitution of the United States. Ban-
[WaB1890] Warren S, Brandeis L (1890) The Right to Privacy. Harvard Law Review,
[Wei91] Weiser M (1991) The Computer for the 21st Century. Scientific American,
[Wei93] Weiser M (1993) Ubiquitous Computing. IEEE Computer, 26(10):71–72, October
[Wes67] Westin AF (1967) Privacy and Freedom. Atheneum, New York NY, 1967 [Win99] Winner L (1999) The Voluntary Complexity Movement. NetFuture: Technology
and Human Responsibility, September 1999. Available at www.netfuture.org/1999/ Sep1499_94.html#3
[WLS01] Westerlund B, Lindquist S, Sundblad Y (2001) Cooperative Design of Commu-
nication Support for and with Families in Stockholm, September 2001. Available at in-terliving.kth.se/papers.html
[ZPS04] Zerdick A, Picot A, Schrape K, Burgelman J-C, Silverstone R (eds) (2003) E-
Available online at www.sciencedirect.comJingping Zhang, Man Ye, Haishan Huang, Lezhi Li, and Aiyun YangOdds of major depression have significantly increased among adults withchronic diseases. However, the diagnosis of depression is often unrecognizedin China.To know the prevalence of depression in medical inpatients with differ-ent chronic diseases and to assess the level of unrecognized depres
Presence of Pharmaceuticals in Treated Wastewater Effluent and Surface Water Supply Systems, Metropolitan Atlanta, Georgia, July–September 1999 Deborah M. Moll, Ph.D.1; Elizabeth A. Frick2;Alden K. Henderson, Ph.D., M.P.H.1; Edward T. Furlong, Ph.D. 3; Michael T. Meyer, Ph.D.41Centers for Disease Control and Prevention, National Center for Environmental Health, Atlanta, GA;2U.S. Geologic