Charting past, present, and future research in ubiquitous computing Export

@article{citeulike:201593, abstract = {The proliferation of computing into the physical world promises more than the ubiquitous availability of computing infrastructure; it suggest new paradigms of interaction inspired by constant access to information and computational capabilities. For the past decade, application-driven research on abiquitous computing (ubicomp) has pushed three interaction themes: natural interfaces, context-aware applications, and automated capture and access . To chart a course for future research in ubiquitous computing, we review the accomplishments of these efforts and point to remaining research challenges. Research in ubiquitious computing implicitly requires addressing some notion of scale, whether in the number and type of devices, the physical space of distributed computing, or the number of people using a system. We posit a new area of applications research, everyday computing, focussed on scaling interaction with respect to time. Just as pushing the availiability of computing away from the traditional desktop fundamentally changes the relationship between humans and computers, providing continuous interaction moves computing from a localized tool to a constant companion. Designing for continous interaction requires addressing interruption and reumption of intreaction, representing passages of time and providing associative storage models. Inherent in all of these interaction themes are difficult issues in the social implications of ubiquitous computing and the challenges of evaluating> ubiquitious computing research. Although cumulative experience points to lessons in privacy, security, visibility, and control, there are no simple guidelines for steering research efforts. Akin to any efforts involving new technologies, evaluation strategies form a spectrum from technology feasibility efforts to long-term use studies\—but a user-centric perspective is always possible and necessary}, address = {New York, NY, USA}, author = {Abowd, Gregory D. and Mynatt, Elizabeth D.}, citeulike-article-id = {201593}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=344988}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/344949.344988}, comment = {Do people want a continuously present computer interface? Such tools present interesting cultural and social issues that need to be addressed. For example, blackberries have changed how accessible people are. While some individuals appreciate and find a sort of comfort knowing they can check their email at all times, other people (myself included) prefer and enjoy the occasional freedom from communication. This is not to say that all ubiquitous computing technology is an attempt to maintain constant accessibility but it does introduce such social dynamics and also policy questions such as concerns for privacy. If people are connected to a mobile device constantly, their movements can be tracked, etc. This does pose significant privacy concerns for users. However, the lack of discussion of privacy concerns with respect to technology is alarming because the lay person who is the end user of the technology may not understand the consequences of using such technology. ---=note-separator=--- One of the social concerns mentioned is control of information the system collects about you. However, I rarely am worried about systems "seeing" me, but rather I am more often worried about what it will do with that information or who will see that information. Should we as researchers be focusing less on giving the user control of their information and focusing more on finding ways of ensure the information is used appropriately? How can be build "trustable" systems? Any system that uses recognition (natural input, machine learning, etc) will inherently have errors. There applications where errors are tolerated to a certain extent; for example, a spam filter. But errors are only tolerated because the consequence is usually not all that dire? Is recognition only useful for applications where the consequences are trivial or easy to recover? The goal of ubicomp is to "promote a unified and continuous interaction between humans and computation services." This seems to suggest that ubicomp applications must be integrated with each other. If so, how can we ensure when we build new applications that they can be successfully integrated with the plethora of existing applications? It seems to me that many applications involve somewhat new interactions. Is it a failure (at least from the ubicomp perspective) if you develop a great, useful, and popular app that isn't well-integrated with other ubicomp apps? Is it a failure if the app isn't well-integrated with people's existing lives? ---=note-separator=--- I agree with Aruna's concern about the social implications of ubicomp. Even though it might be possible to be able to capture all the information about our past, is that something we want? Will people be preoccupied and trapped in the past? If everything is available for replay, will it lead to a world were we cannot stop ruminating over our past interactions? Laptops, cell phones, and the internet have caused fragmented attention. The Internet and other communication technologies have enabled this "always connected" sensation. People are overwhelmed with input and there is almost no limit to of amount of information, music, movies, podcasts, messages, etc etc that we can get or possibly consume. How might this effect on society been predicted before the Internet and can we apply the same methods to ubicomp? How can we safely take advantage of the benefits of ubiquitous computing without feeding our sometimes unhealthy desire for more information? Matt's comment about things magically working together raises several more interesting questions. To have bits of technology talk to each other to figure out the right thing requires either 1) standards or 2) semantic reasoning skills. Standards are hard to get designers to agree upon and even when they exist, they may be suboptimal standards that stick with us for a long time. Semantic reasoning of computer agents requires them to communicate (oh you need standards for that) and to reason about what other devices can do. Part of me wants to throw up my hands and say this is AI-complete (or worse) but then again perhaps a useful research direction is to figure out how to make this problem tractable and push back the boundary of what is AI-complete. Another thing that surprised me was that Abowd and Mynatt said that multi-modal integration would not be discussed in the paper. It's not clear to me that they believe multi-modal technology has nothing to do with ubicomp. One of the "W"'s which is mentioned is "Why?" For a machine to understand why, it needs to be able to model intentionality and since ubicomp is technology "embodied" in the world, it seems to me that multi-modal integration is really important. Another question that this raised for me was, "What forms of modeling human intention are there? And how can machines model and deal with ambiguous cues to intention?" ---=note-separator=--- everyday computing is about scaling computing over time -- what other types of scale need to be considered? (article starts out with device size; physical space; and people) Past work in Ubicomp has succeeded in addressing issues of: natural computing (errors an issue); context (mention errors again as an issue, and data fusion as a possible solution). Does error play a role in capture and access too? everyday computing: mention some critical features: rarely have clear beginning or end; interruption is expected; multi tasking is the norm; associative models of information needed differs from symbiosis in focus on "more free-flowing and integrative" computing. Grand challenge -- rather than a killer app just a killer existence -- is the motivation devoid of need in this paper? or is it assumed it will lead to good? they do argue that ubicomp should be more human focused }, doi = {10.1145/344949.344988}, issn = {1073-0516}, journal = {ACM Trans. Comput.-Hum. Interact.}, keywords = {associative, context, cs-mini, hcii-minis, historical, privacy, survey, ubicomp}, month = {March}, number = {1}, pages = {29--58}, posted-at = {2007-01-15 21:47:11}, priority = {0}, publisher = {ACM}, title = {Charting past, present, and future research in ubiquitous computing}, url = {http://dx.doi.org/10.1145/344949.344988}, volume = {7}, year = {2000} }

TY - JOUR ID - citeulike:201593 L3 - citeulike-article-id:201593 N2 - The proliferation of computing into the physical world promises more than the ubiquitous availability of computing infrastructure; it suggest new paradigms of interaction inspired by constant access to information and computational capabilities. For the past decade, application-driven research on abiquitous computing (ubicomp) has pushed three interaction themes: natural interfaces, context-aware applications, and automated capture and access . To chart a course for future research in ubiquitous computing, we review the accomplishments of these efforts and point to remaining research challenges. Research in ubiquitious computing implicitly requires addressing some notion of scale, whether in the number and type of devices, the physical space of distributed computing, or the number of people using a system. We posit a new area of applications research, everyday computing, focussed on scaling interaction with respect to time. Just as pushing the availiability of computing away from the traditional desktop fundamentally changes the relationship between humans and computers, providing continuous interaction moves computing from a localized tool to a constant companion. Designing for continous interaction requires addressing interruption and reumption of intreaction, representing passages of time and providing associative storage models. Inherent in all of these interaction themes are difficult issues in the social implications of ubiquitous computing and the challenges of evaluating> ubiquitious computing research. Although cumulative experience points to lessons in privacy, security, visibility, and control, there are no simple guidelines for steering research efforts. Akin to any efforts involving new technologies, evaluation strategies form a spectrum from technology feasibility efforts to long-term use studies—but a user-centric perspective is always possible and necessary IS - 1 JF - ACM Trans. Comput.-Hum. Interact. SN - 1073-0516 AD - New York, NY, USA EP - 58 TI - Charting past, present, and future research in ubiquitous computing PB - ACM VL - 7 SP - 29 KW - associative KW - context KW - cs-mini KW - hcii-minis KW - historical KW - privacy KW - survey KW - ubicomp N1 - Do people want a continuously present computer interface? Such tools present interesting cultural and social issues that need to be addressed. For example, blackberries have changed how accessible people are. While some individuals appreciate and find a sort of comfort knowing they can check their email at all times, other people (myself included) prefer and enjoy the occasional freedom from communication. This is not to say that all ubiquitous computing technology is an attempt to maintain constant accessibility but it does introduce such social dynamics and also policy questions such as concerns for privacy. If people are connected to a mobile device constantly, their movements can be tracked, etc. This does pose significant privacy concerns for users. However, the lack of discussion of privacy concerns with respect to technology is alarming because the lay person who is the end user of the technology may not understand the consequences of using such technology. N1 - One of the social concerns mentioned is control of information the system collects about you. However, I rarely am worried about systems "seeing" me, but rather I am more often worried about what it will do with that information or who will see that information. Should we as researchers be focusing less on giving the user control of their information and focusing more on finding ways of ensure the information is used appropriately? How can be build "trustable" systems? Any system that uses recognition (natural input, machine learning, etc) will inherently have errors. There applications where errors are tolerated to a certain extent; for example, a spam filter. But errors are only tolerated because the consequence is usually not all that dire? Is recognition only useful for applications where the consequences are trivial or easy to recover? The goal of ubicomp is to "promote a unified and continuous interaction between humans and computation services." This seems to suggest that ubicomp applications must be integrated with each other. If so, how can we ensure when we build new applications that they can be successfully integrated with the plethora of existing applications? It seems to me that many applications involve somewhat new interactions. Is it a failure (at least from the ubicomp perspective) if you develop a great, useful, and popular app that isn't well-integrated with other ubicomp apps? Is it a failure if the app isn't well-integrated with people's existing lives? N1 - I agree with Aruna's concern about the social implications of ubicomp. Even though it might be possible to be able to capture all the information about our past, is that something we want? Will people be preoccupied and trapped in the past? If everything is available for replay, will it lead to a world were we cannot stop ruminating over our past interactions? Laptops, cell phones, and the internet have caused fragmented attention. The Internet and other communication technologies have enabled this "always connected" sensation. People are overwhelmed with input and there is almost no limit to of amount of information, music, movies, podcasts, messages, etc etc that we can get or possibly consume. How might this effect on society been predicted before the Internet and can we apply the same methods to ubicomp? How can we safely take advantage of the benefits of ubiquitous computing without feeding our sometimes unhealthy desire for more information? Matt's comment about things magically working together raises several more interesting questions. To have bits of technology talk to each other to figure out the right thing requires either 1) standards or 2) semantic reasoning skills. Standards are hard to get designers to agree upon and even when they exist, they may be suboptimal standards that stick with us for a long time. Semantic reasoning of computer agents requires them to communicate (oh you need standards for that) and to reason about what other devices can do. Part of me wants to throw up my hands and say this is AI-complete (or worse) but then again perhaps a useful research direction is to figure out how to make this problem tractable and push back the boundary of what is AI-complete. Another thing that surprised me was that Abowd and Mynatt said that multi-modal integration would not be discussed in the paper. It's not clear to me that they believe multi-modal technology has nothing to do with ubicomp. One of the "W"'s which is mentioned is "Why?" For a machine to understand why, it needs to be able to model intentionality and since ubicomp is technology "embodied" in the world, it seems to me that multi-modal integration is really important. Another question that this raised for me was, "What forms of modeling human intention are there? And how can machines model and deal with ambiguous cues to intention?" N1 - everyday computing is about scaling computing over time -- what other types of scale need to be considered? (article starts out with device size; physical space; and people) Past work in Ubicomp has succeeded in addressing issues of: natural computing (errors an issue); context (mention errors again as an issue, and data fusion as a possible solution). Does error play a role in capture and access too? everyday computing: mention some critical features: rarely have clear beginning or end; interruption is expected; multi tasking is the norm; associative models of information needed differs from symbiosis in focus on "more free-flowing and integrative" computing. Grand challenge -- rather than a killer app just a killer existence -- is the motivation devoid of need in this paper? or is it assumed it will lead to good? they do argue that ubicomp should be more human focused AU - Abowd, Gregory AU - Mynatt, Elizabeth PY - 2000/03// UR - http://dx.doi.org/10.1145/344949.344988 ER -