Faculty of Computer Sciences, Østfold University College, Norway
Department of Informatics, Faculty of Mathematics and Natural Sciences, University of Oslo, Norway
Social applications have become ubiquitous tools for managing contacts and for sharing of information such as images, bookmarks and statuses. We argue that ideas from social applications can play a role in sharing of accessibility information, and use OurWay, a collaborative route planner, as a case in this argument. By learning from research on well-known social applications, augmented by our own work on aspects of personal navigation with OurWay, we re-frame OurWay as a social application, and look at benefits and potential pitfalls when introducing ideas from social software into the area of accessibility mapping.
Accessibility is a wide term, which applies in different ways to a wide variety of resources, be it information and communication technologies (e.g. the European Union's focus on eAccessibility and eInclusion, the 2006 Riga declaration, and The World Wide Web Consortium's Web Accessibility Initiative (WAI), and the Web Content Accessibility Guidelines (WCAG)) or physical access in public areas (e.g. the United Nations Convention on the rights of persons with disabilities). In this paper, we're concerned with sharing of location-based accessibility information in the physical world.
Accessibility information is frequently shared between users, either directly through friends or contacts, or indirectly through formalized initiatives undertaken by organizations such as The Norwegian Association of Disabled. Collecting, verifying and distributing accessibility information is a formidable task, especially taking into consideration the wide variety of disabilities and the temporal aspect of many obstacles in the urban environment. It is not sufficient to collect the data once and then distribute it, it needs continued maintenance. Furthermore, the commercial actors in the navigation market are primarily targeting car drivers. The potential market for pedestrian navigation is not yet considered big enough to attract the focus of these companies.
Through OurWay (Holone, Misund & Holmstedt 2007), we have introduced a concept in which end users play a central role in creating and maintaining accessibility information. The idea is simple. Based on a geographical network representing roads, paths and other traversable features of the urban landscape, a central route planning server suggests a route to a destination. The calculation of this route takes into consideration feedback received from previous use of the system. The feedback consists of ratings of route segments, and is shared between the users through the use of the central route planning server. Thus, the system learns, and has been shown to produce more accessible routes over time. The OurWay concept is described in more detail in the Research Setting section.
Involving the end users makes sense both from a resource perspective and maintenance perspective. After all, these are real citizen sensors (Goodchild 2007) experiencing the urban accessibility issues on the street level, day by day.
The OurWay concept has evolved from a technical prototype to an accessibility-based routing concept. It was initially inspired by collaborative services on the web, such as Wikipedia and Flickr, however it is only recently that we have started to discuss it as a "proper" social application. By bringing together research on social applications on the web with our experiences from the OurWay prototype, we establish a way of discussing accessibility mapping and routing as a problem that may be solvable by organized or ad-hoc groups of end users. Existing literature on accessibility focuses mostly on the interaction between a single user and a computer system, with some notable exceptions, e.g. (Winberg 2006).
Context, time, place and space have been of interest to the Computer Supported Collaborative Work (CSCW) research community for more than two decades. At the end of the 1990's, the term Social Navigation of Information Space was coined by some CSCW researchers to discuss information sharing on the Internet (as opposed to collaboration in work environments, which had so far been the main focus of CSCW). Of special interest to us are the studies on asynchronous sharing of traces of on-line activity (e.g. (Dourish & Chalmers 1994)). Metaphors from navigation in the physical world were used to frame discussions of navigation in the wast information space on the Internet. In OurWay, we have taken ideas from Social Navigation of Information space and brought them back to the physical world.
Twitter and other micro blogging services have now enabled geo-tagging of status messages, thus providing the possibility of using the statusphere as a vehicle for exchanging and aggregating accessibility information related to place. The web is going mobile, and today's mobile phones come with GPS receivers, and Internet connectivity. This suggests great opportunities for collaborative, on-line, mobile systems such as OurWay. We will briefly mention some possible implications of this when we discuss the application of social software for accessibility mapping.
In this paper, our main research question is: What opportunities and challenges exist when applying insights from social software to the domain of accessibility mapping? As a background for this discussion, we first re-frame OurWay as a social application in the context of existing literature on social software.
The rest of the paper is organized as follows: First we review relevant accessibility and social software research, and take a look at different forms of cooperation. Then we describe the OurWay concept, our previous prototype work, and then attempt to re-frame OurWay as a social application. In the discussion we take on the main research question in the light of social software research. Finally we conclude the paper with suggestions for further work to bring social software into the realm of accessibility mapping.
In this section we briefly review relevant literature on accessibility, social software including forms of collaboration, and trust and privacy.
As mentioned in the introduction, accessibility is a wide term that applies in different ways to different situations. It is interesting to see the different ways the term can apply even when focusing on accessibility in the physical world. For instance, in results from the AUNT-SUE project (Evans 2009), social aspects like fear of crime are mentioned as important factors when users assess the overall accessibility of an area. Similarly, the AMELIA project (Titheridge, Mackett & Achuthan 2009) identify surrounding factors such as places to rest and public toilets as important in the overall accessibility picture.
Völkel et al. (Völkel, Kühn & Weber 2008) describes requirements for accessibility annotation of geographic data, with the aim to collect detailed information for different needs and preferences. This information is then to be used in multi-criteria routing systems.
The research on accessibility in the built environment is extensive, however, we have not been able to identify any previous work explicitly discussing on-line collaboration between users in an accessibility context, or the use of social software for this purpose.
Jonathan Grudin, although not using the term social software, puts forward four characteristics he considers key in these kinds of technologies (Grudin 2006):
... they 1) can be extremely lightweight, 2) make information and activity highly visible, 3) provide individual and group benefits, and 4) are grassroots, self-organizing phenomena.
Grudin refers to studies of using project blogs and wikis as well as the use of hashtags for coordinating and maintaining project-related content and activities.
Recently, the potential application of user generated content created by Internet based communities have become visible, for instance in the role of OpenStreetMap activities after the Haiti earthquake. These types of collaborative efforts are made possible by the low cost of establishing a community, the ease of organizing activity and coordinating work, and the existence of an open infrastructure for creating and publishing the content, in this case maps. These phenomena are covered convincingly by Clay Shirky in his book "Here Comes Everybody: The Power of Organizing Without Organizations" (Shirky 2008). Shirky also discusses the political and organizational pressure that can be put forward by self-organized groups using the new technological advances such as blogs and mobile phones.
Through our studies we have looked at how the users relate to the tools and technology, and how they relate to each other. To describe this in context, we start with a short introduction to important work on groups and cooperation. This is by no means an extensive review, however it helps frame the discussion of how we interpret the usage patterns displayed by the participants in the OurWay studies.
Coming from the work by social scientists to understand learning processes, the term Communities of Practice (CoP) refers to ways in which apprentices learn from taking part in a communities that share a common goal (Wenger 1999). To quote Etienne Wenger, one of the originators of the term:
Communities of practice are groups of people who share a concern or a passion for something they do and learn how to do it better as they interact regularly.
Later, Lave and Wenger described situated learning in CoP's as Legitimate Peripheral Participation, LPP (Lave & Wenger 1991). The main contribution of LPP is an understanding of how the learning process gradually turns an apprentice into an expert. Starting out with important although non-critical tasks, the apprentice is introduced to the Community of Practice. Step by step, the importance of the tasks and the responsibility is increased, until the apprentice has acquired expert skills within the community.
Building on Communities of Practice, Gerhard Fischer has introduced the term Communities of Interest (CoI) to focus on cross-domain collaboration (Fischer 2001). The main concern here is the communication between groups from different domains, the challenges introduced by different cultures and vocabularies and how to capture this in knowledge management systems.
Communities of Interest often have a more temporal nature than Communities of Practice, which are typically long-lived although not static.
Dourish and Chalmers introduced the concept of Social Navigation in 1994. They described it as
... navigation towards a cluster of people or navigation because other people have looked at something (Dourish & Chalmers 1994)
Social Navigation takes an information centric approach, whereby information left behind by users because of their activities form places where people interact. Sharing of information then, is not necessarily a result of participation in a Community of Practice, rather it is a by-product of use, which might well be without any thought for other users.
Social Navigation of information uses the way we navigate in the real world, e.g. by visiting restaurants with many customers and avoiding the ones without them, as a metaphor for navigating information places on the web. Paul Dourish describes Social Navigation of Information Space as a type of awareness system, with some significant differences compared to "traditional" CSCW awareness technologies. The information can be aggregated over time from the use of many different users, which means that the communication of awareness can be asynchronous. Decoupling of the activity that produces the information and the situation in which this information is used (perhaps as part of an aggregate) poses challenges to the design of these systems, e.g. for presentation of awareness information (Dourish 1999).
Suggesting a social software approach to accessibility mapping requires consideration of the important issues of trust and privacy. We'll briefly review some relevant work here, and return to it in the discussion.
Palen et. al (Palen & Dourish 2003) presents a privacy framework for discussing the negotiation of private and public information in different social situations. They introduce the idea of boundaries for disclosure, identity and temporality, and use this to discuss how we negotiate what digital footprints we leave behind. In the context of social applications for accessibility mapping, the balance between privacy and trust is one that needs careful consideration.
Information trust and dependability in Wikipedia has been the subject of several studies. Priedhorsky (Priedhorsky, Chen, Lam, Panciera, Terveen & Riedl 2007) identifies types of damage in Wikipedia, as well as assess the impact of that damage. Extending the work of Viégas et al. (Viégas, Wattenberg & Dave 2004), it is in particular the introduction of metrics for damage impact that interests us. Whilst important in most cases where user generated content is involved, understanding the potential consequence of vandalism and erroneous information becomes paramount when applying social applications for turn-by-turn routing instructions.
Other initiatives to make user activity and information validity in Wikipedia more visible include WikiDashboard (Suh, Chi, Kittur & Pendleton 2008) and WikiTrust (Adler, Benterou, Chatterjee, De Alfaro, Pye & Raman 2008). Both of these efforts are working to make Wikipedia more socially transparent, by visualizing author credibility and article revision history and other relevant information to assess trustworthiness of Wikipedia content.
Having briefly reviewed literature relevant to our main research question, we now turn to our case: OurWay. First, we give a short summary of previous work, before we attempt to re-frame OurWay as a social application.
OurWay is a collaborative route planner, where users are providing feedback on accessibility through their interaction with the system. Users are equipped with mobile phones running a client application, which connects to a central route planning and feedback server over a mobile Internet connection. The client displays a map of the area, and allows the user to ask for a route between two locations, which is then displayed on the map. The user can at any time provide feedback to the central server about the part of the route being traversed. To keep the threshold for contribution low, we have used only three levels of feedback: good, uncomfortable, and inaccessible. The submitted ratings are attached to route segments, and are used as weights by the route planning server when calculating new routes.
OurWay started out as a technological prototype demonstrating the possibilities with mobile Internet, location based information, digital maps and route planning. We have shown that the technology is mature and that the concept works on a technological level (Holone et al. 2007). Further, we have suggested different stakeholders in the concept, including end users who use the tool primarily for navigation, and surveyors who use the tool in campaign style operations to map the accessibility of a region or a building (Holone & Misund 2008). More recently, the focus of our research has been on the user experience and concept viability (Holone, Misund, Tolsby & Kristoffersen 2008). We have shown that the routes suggested by the route planner improves over time, despite users being mostly focused on their own navigational tasks, and not thinking explicitly about other users' potential benefit of their actions.
The suggestion that it is the by-product of use that is beneficial to others fits nicely with some of the core ideas of Social Navigation of Information Space. Viewing OurWay as a social navigation tool, several design limitations become obvious, especially with regards to awareness of other users, their actions and identities. This gives us an incentive to attempt to re-frame OurWay as a social application, and re-assess some of the design choices made in the current prototype.
To discuss OurWay as a social application, we use Grudin's four characteristics (Grudin 2006) as a starting point. This re-framing provides the background for the later discussion of the main research question.
Grudin's perspective is obviously the end users perception of the technology at hand. In this respect OurWay can be considered extremely lightweight. One of the key design elements of the concept is that of annotation, or accessibility feedback provided by the user. When the user wants to provide feedback to the system, she simply pushes a button and selects either good, uncomfortable or inaccessible from a menu, and the information is sent back to the server and stored for later use. The context in which the information is captured, along with social relations to other users should also be considered in further explorations of the concept. Today, no explicit context is captured, and we have previously suggested that the implicit context created by the group one associate with can be sufficient for this purpose. This assumption needs to be challenged, however we should keep in mind that capturing and using meaningful context is not a trivial endeavor (see for instance (Agre 2001) ).
Typically, the user re-requests a route at this point, receiving an alternative route from the current location to the destination. This is a vast simplification of the typically form based inquiry (door widths, ramp angles etc) taking place in traditional accessibility mapping. It also contrasts with the requirements suggested by Völkel et al. We do not propose that the OurWay concept should replace such detailed approaches, rather that the information generated by users can augment existing information, if it exists.
Perhaps the main drawback of the current OurWay design is the way the activity of other users is shared in the group. The current prototype only indirectly provides this information, through the resulting route suggested to the user. In other words, there are no clues provided as to who provided the annotations that led to the suggested route, or what context they were captured in. Albeit a shortcoming of the current prototype, especially seen in light of the literature we're now relating to, this design decision has led us to some interesting observations in terms of cooperation and system robustness, as we will see in the next section.
All our test users have been introduced to the OurWay concept and the idea of "the others" before taking part in our experiments (see (Holone et al. 2008) for details). Even so, they soon focus on their own navigational tasks, and tend to forget the existence of the other users. They are concerned with their individual benefit of using the system: the immediate incentive for providing feedback to the system is the new and alternative route the system provides. However, the group benefit is obvious, even though it might not be a conscious act by each participant to annotate in the group's best interest. The OurWay concept has been demonstrated to be fairly robust despite this self-centered use, perhaps mostly because users tend to annotate only when they themselves need a new route alternative from the system, that is when the encounter an absolute obstacle like a staircase.
The Social Navigation way of thinking about interaction fits very well with the way users of OurWay share their experiences. They leave behind traces of their use, which in turn is used to provide better quality routes for other users. The OurWay concept also differs slightly from the core ideas of Social navigation, by the fact that it is the removal of obstacles, not identification of non-obstacles which is key to the concept.
OurWay has so far only been tested with pre-recruited participants in controlled experimental settings (outdoors and indoors). Making OurWay available to a larger (and less controlled) group would be necessary to draw conclusions with regards to the nature of community build-up. This also raises the interesting question, also addressed both by Grudin and Shirky, of the interplay between structured organizations and self-organizing communities, and the role of lightweight social software in this context. We will return to this issue in the main discussion of this paper.
In the previous section we discussed the OurWay concept as a social application. Based on this, we now take a broader perspective, and take on the main research question: What opportunities and challenges exist when applying insights from social software to the domain of accessibility mapping? We continue to apply Grudin's four characteristics for this exercise, this time emphasizing the relevance of each characteristic when applied in an accessibility mapping context.
By utilizing Web 2.0 technologies and the understanding of community building around these technologies, more complete and better coverage of accessibility information can be achieved. It might not be information of rigorous detail, rather it may take the form of status messages with ratings and locations. It is also possible to leverage the use of hashtags for grouping of related information, and as a way for users to provide contextual information. This in turn can be used for route planning trough aggregation services facilitated by lightweight integration with existing Web 2.0 services.
One obvious challenge is the use of mobile phones for people with special needs. Although mobile phones are ubiquitous among all user groups, the necessary HCI challenges related to use in the field must be considered carefully. One key argument for using an open, lightweight infrastructure for a social accessibility mapping application is that the opportunities for adaption to different needs and requirements are moved away from the core of the system and towards the perimeter. This removes the dependence on a central agency to provide customized solutions to each combination of such requirements, and allows for innovation on the user side of the system.
The concept relies on access to a geographical network on which to base the route calculations. Although less lightweight than the suggested use of status messages on the user side, the OurWay concept has been using open user generated content from the OpenStreetMap project. This makes the process of updating the core geographical information lightweight, in the sense that it removes the dependency on the much slower update cycles of corporations. Further, it allows access to the actual core data, and not only to map images which cannot be used for for routing applications. Finally, the Creative Commons licensing allows for appropriation and invention by all interested parties.
Any navigational tool needs to be reliable from the user perspective. A tool which provides accessible routes must adhere to high standards when it comes to dependable information and social transparency. Finding the balance between transparency and privacy is challenging, and will vary from person to person in different situations. There is, however, little doubt that allowing the user to see the ratings and context which is used to calculate the routes is important. The question of whether to identify the users behind the collected information is more difficult to answer. Some suggestions can be found in the WikiTrust and WikiDashboard projects mentioned earlier, where exposing the source and history of information is a key point.
Using the boundaries from Palen et al., it is obvious that any use of a computer system to collect and disseminate accessibility information diminishes the possibilities for negotiating the disclosure of ones true identity at a later stage. The real identities of users might not need to be exposed, a user pseudonym can also earn a good reputation for providing relevant information, as is the case in Wikipedia. One can argue that this removes the personal relationship between information source and the user, as is often sought in questions about accessibility. On the other hand, this can be compensated for over time by attaching trust in information from user pseudonyms. Tapping into the existing social networks of the users (including groups and lists) and using this as a filter in the route planning is another interesting way forward.
Some individual benefits of a social software system for accessibility mapping are obvious, such as the opportunity to provide feedback on suggested routes and get immediate reward in the form of an alternative route from the system. The benefits for the group(s) are, as we have shown through the work with OurWay, more accessible routes over time, despite use primarily to solve personal navigational tasks. Another benefit for the group as a whole, is that this way of collecting and maintaining accessibility information is a potential resolution to the resource challenge mentioned in the introduction of this paper.
Further, by utilizing existing social software infrastructures, the social awareness of mobility challenges can be raised, also outside of the group of core users. Take for example a location based service such as FourSquare, where integration with Twitter is essential to spread the word about FourSquare to the users' existing Twitter networks. Another lesson to take away from new specialized social services is the use of game metaphors to get people to contribute information, and to achieve personal status. By linking rewards in the virtual world of the service with rewards in the real world, a new set of incentives for contribution has been established.
It is difficult to imagine accessibility mapping tools being appropriated by enough users to appear as a "true" grassroots movement. There is obvious value in the networks, agendas and political power represented by established interest organizations. At the same time, parts of what make social applications work is the ability for users to rapidly form groups and processes, often to influence the organizations or establishments they are taking part in. In other words, there should be a mutual interest from institutions, individual users and user groups to make use of social software tools, however the suitable balance of initiative and power remains to be established.
For a hierarchical organization to suddenly be an active participant in a system largely created and run by a grassroots movement can be challenging, both with regards to competence, resources and a shift in power relationships. Similar issues are discussed by Borchorst et al. (Borchorst, Bødker & Zander 2009), where they look at the application of Web 2.0 technology in the interaction between citizens and municipalities in Denmark. An established institution that becomes a participant in a social application not only provides a new form of communication between users and the institution, it creates a new set of expectations about the system from the users perspective.
We have reviewed accessibility and social software literature, and used this to re-frame of the OurWay concept in the light of Jonathan Grudin's four characteristics of such systems. Further, we have used this as a background to discuss the potential application of social software for accessibility mapping.
Using existing Web 2.0 technologies as a platform to further explore the OurWay concept is an interesting way forward, and may allow for appropriation of the concept beyond what a fixed set of features can give. It also introduces new challenges, like establishing trust in user generated content, and privacy related to digital traces left behind through use.
Using ideas from social software and applying them to accessibility mapping has huge potential, both as a possible solution to the resource challenge for collecting this kind of data, and for raising social awareness about mobility issues in the urban landscape. It also changes expectations and communication forms, both for end users and institutions.
One possible next step in the OurWay project is to apply these ideas, leveraging existing Web 2.0 technologies and social network services as a platform for the OurWay concept.