Chapter 5 • Study 2: Mutual Location Awareness in Pervasive Computing
5.5 Coordination devices
5.6.2 Two ways of playing
We already mentioned that players from the Control condition and players in the MLA conditions played differently. Indeed, even though their task performances were similar, the differences lay in the cognitive processes involved during collaboration as well as the employment of manifest elements in the environment. Since there were no important differences between the two MLA conditions, we will consider them as equivalent in the discussion below. Discussing these differences is a good way to describe the influence of mutual location awareness on coordination.
The inhibition of communication
One of the most striking differences we described in the quantitative analysis was certainly the communication pattern. We have seen how players in the Control groups exchanged overall more shared map annotations, mostly about positions, direction, strategy and questions. This had been confirmed by the qualitative analysis in which we saw more explicit agreement and discussion for this group.
These results show how the presence of the MLA interface (be it synchronous or asynchronous) led players to take for granted that it would be sufficient to achieve their joint activity. This can be explained by a variant of the least collaborative effort principle (Clark, 1996): participants used the coordination devices that were available (mostly the plan and the MLA) and were less bothered to send more information through shared map annotations. Players in the Control condition only had the plan as a coordination device. The cost to exchange messages was the same but the necessity was higher for players in the Control group: one accepts this cost if there is a real need to exchange coordination devices.
It also means that Control players anticipated something: they had to send more information otherwise the room for interpretation would be too great for the other players. That is why they sent messages about their direction and about strategy: the
other teammates could then better infer what to do, and consequently build a more accurate mutual model. This is then bound to the importance of the perception of what is relevant or not and how it should be communicated to the partners: the capacity of agency.
The importance of human agency
By agency, we refer to the capability of an individual to decide whether or not to act. In the context of coordination, agency describes the process by which a partner does or does not make manifest a sign of mutual intelligibility by exchanging a coordination device. This is where the main difference lay between the different experimental conditions: automating MLA is different than sending one’s position in space. Indeed, participants from the Control condition chose the information they perceived as relevant (position, direction and strategy) for the time being and sent them to their partners. This fact raises an important issue regarding communication and spatial information: compared to automatic positioning in which location is just information, self-declared positioning is both an information and an act of communication act, intentional by definition. If A tells B where he or she is located, not only B knows A’s location but he or she also knows that A considers that it is useful for B to know it. It appeared that players without MLA took better advantage of the annotation capabilities, using it to express their path and their strategy. The players with the awareness tool were able to annotate as well but did not use this opportunity. As we mentioned, there was a certain inertia caused by the presence of location awareness information. We can then conclude that in the context of this experiment it was better to leave users without the location-awareness tool, and instead with a broad channel of communication. Players in the Control groups chose the information they perceived as relevant (position, direction and strategy plus asking questions) and sent them to their partners at the moment they wanted it to be known by the others. They perform more acts of ostensive communication as described by Sperber and Wilson (1986): the self- expressed position is both an attractor for others’ attentions and a way to show the communicator’s intent through messages about strategy or directions. Users could indeed express what they found relevant for the current task: with regard to the content (their position, direction, strategy messages) and to the pragmatic level (questions). Using Clark’s words, this result can be explained with the principle of joint salience. According to this principle, players deployed the coordination devices that would be the most salient and prominent with respect to the common ground of all participants. And, as we will see in the next section, there are differences between the groups in terms of the common ground.
A decreased quality of the common ground
The qualitative analysis of the data showed the different set of coordination devices used by players from the various experimental conditions. The accumulation of coordination devices over time by the team-mates of the Control groups was richer than that of the MLA groups. Since players without the MLA exchanged more map annotations, they collected more information about the situation, which would be why the richness of communication had a positive on the mutual modeling index for them. The location awareness of others (conveyed by an interface in the MLA condition and by shared map annotations in the Control condition) was wrapped in more information and context in the Control groups. This location information made sense for them in
conjunction with other information: mostly the readings each of them noted and, the decision they made during the plan with explicit agreement about what strategy to apply. This helped in constituting a richer common ground. It appeared that automating MLA could be problematic in performing mutual modeling acts based on this common ground.
Moderating the criticism of MLA
Another factor of differentiation between the two different ways of playing is to look at the performance of each group for each phase. This discrimination allows us to moderate the critique of the location-awareness tool.
Actually, we saw that when looking at the task performance for each phase, the MLA interfaces had different effects. For phase 1 and 2, the giving of one’s location could be replaced by verbal communication, which was not the case in the last phase where people had needed a tight coordination. Therefore, as Espinosa et al. (2000), we conclude that awareness information impacts differ depending upon the task. This is obviously a starting point for new field experiments in order to investigate the articulation between tasks and levels of automation with awareness information.
The omnipresence of the plan
Though we insisted on the different ways of playing in the different experimental conditions, we have to conclude by putting forward a pertinent common aspect between them. For both MLA and Control players, the presence of the 5 minutes plan was a very important feature that we can describe through the lens of our framework. The qualitative analysis of the coordination device showed that players from all the conditions heavily relied on this planning phase since it provided them with an important source of information to perform mutual modeling acts. Indeed, during this planning phase, they discussed explicitly how the group should act, when and why. This discussion then allowed each group to set conventions that we can define as “local” (i.e. only for the local context of this game). The plan could thus be defined as the encapsulation of explicit agreements and convention and coordination devices.