Communication in Context: A Stimulus-Response Account of Agile ...

4 downloads 40581 Views 164KB Size Report
and how to enhance the planning game to take account of requirements ... Cao and Ramesh [5] provide evidence that the RE process for agile software de-.
Communication in Context: A Stimulus-Response Account of Agile Team Interactions Nik Nailah Binti Abdullah1, Helen Sharp2, and Shinichi Honiden1 1 GRACE Center, National Institute of Informatics, Tokyo, Japan Center for Research in Computing, The Open University, United Kingdom {bintiabd,honiden}@nii.ac.jp, [email protected]

2

Abstract. Previous research has indicated that work artefacts commonly used by agile teams capture progress information, while functional aspects such as requirements are developed and sustained through the team’s social interactions and communication channels. This paper reports an initial empirical study to investigate the relationship between agile work artefacts and communication during stand-up meetings and pair programming sessions, specifically focusing on gathering and clarifying requirements. Using Bateson’s communication theory, we found that the work artefacts, and other individuals form an external event system which supports Agile teams during the gathering and clarifying of requirements. Using this communication theory together with Clancey’s situated cognition, we predict that if the two do not exist together throughout the interactions, then teams members will form discoordinated actions together. Keywords: Empirical study, stand-up meetings, communication, situated cognition, contexts.

1 Introduction Agile practice emphasizes communication and downplays documentation. Most agile teams therefore rely on simple artefacts such as story cards and the Wall, together with frequent communication to help create and sustain a shared understanding of the product’s requirements. Analyses of agile practice using distributed cognition [10, 14] and cognitive dimensions [8, 15] suggest that story cards and the Wall focus on capturing and displaying progress information rather than requirements issues such as problems, goals or functionality. Instead, requirements of the product reside in the social context and are sustained through communication and collaboration activities. This paper progresses these earlier analyses by investigating what is achieved during an agile team’s communication, and what role is played by the physical artefacts. We present the preliminary findings from a field study of one co-located agile team, focused on their communication relating to the activities of gathering and clarifying requirements. The paper is organized as follows. In the next section we introduce some relevant related work in agile team communication and agile requirements. Then we describe the study setting, data gathering and analysis. Then we present our findings, followed by conclusion and future work. A. Sillitti et al. (Eds.): XP 2010, LNBIP 48, pp. 166–171, 2010. © Springer-Verlag Berlin Heidelberg 2010

Communication in Context: A Stimulus-Response Account of Agile Team Interactions

167

2 Related Work Work focused on requirements engineering in an agile context often aims to integrate existing requirements engineering tools or processes into an agile environment. For example, Nawrocki [12] describe how to include requirements documentation into XP and how to enhance the planning game to take account of requirements engineering concerns. Grunbacher and Hofer [9] integrate requirements negotiation into agile processes. Gallardo-Valencia and Sim [7] investigate how agile teams handle requirements and conclude that requirements are constantly being validated which in turn strengthens team bonds. Cao and Ramesh [5] provide evidence that the RE process for agile software development differs from that of traditional development – it is an iterative discovery approach – and they identify seven agile RE practices, the most important of which is the intensive communication between the customer(s) and developer(s). Yet few studies of communication in an agile team have been reported. For example Bryant et al [4] investigate collaboration in pairing by analyzing the verbalizations of pairs working together. In this work, the focus was on understanding whether pairs worked collaboratively and how collaborative they were. Moløkken-Østvold [11] focused on the role of the customer and found that where there was daily communication between developers and customers, there were fewer overruns. This latter study did not investigate the nature of those communications.

3 The Study and Research Methods Our study was designed following an ethnographically-informed approach [13]. In particular, this approach aims to study practice without interference, treats all data as ‘strange’, and asks ‘how’ and ‘why’ and ‘what are the characteristics of’ questions. In this case, we intended to study the communication process of the agile team to understand the relationship between physical artefacts and communication during requirements activities. We therefore collected detailed data relating to the team’s verbal and non-verbal interactions and analysed it. 3.1 The Study Context The agile development team observed is based in a large telecommunications and media company in the UK; the observation lasted for four days in March 2009. The team had been together for over a year and several members of the team had worked together before. The team consisted of five developers and two customers. All the studies were conducted at the agile development team’s office. The team’s project was to deliver a prototype portal for cloud services. The portal needed to include some additional features, which were to be achieved by combining it with other systems that the team had been developing. It had to be demonstrated to the company’s clients the following week. The request for additional features was made at the beginning of the observation period and it had to be delivered the day after observation ended. In this team stories were expressed as abbreviated labels such as ‘USER ACCOUNTS’. The Wall was a filing cabinet, as shown in Figure 1.

168

N.N.B. Abdullah, H. Sharp, and S. Honiden

Fig. 1. The team discussing story cards on the Wall during a stand-up meeting

3.2 Data Gathering and Analysis The data gathered from our observations consisted of field notes, photographs, meeting audio recordings, short video recordings of group interactions in front of the wall, audio records of four pairing sessions, and post-observation informal interviews via email. In this paper we report our analysis of audio, video, interview and field note data from two stand-up meetings from the first and second day observations (22 minutes and 30 seconds in total), and four pair-programming sessions (3 hours 5 minutes in total) also from the first and second day of observation. The audio data was transcribed using symbols commonly employed in discourse analysis [1, 16]; in total we have transcribed about 7,410 lines of dialogue. We then aligned the transcriptions to field notes, photographs, and video interactions based on a timeline, and contacted the team members via email for clarification when necessary. To analyse our transcripts, we used an approach that combines Bateson’s communication theory [2] and Clancey’s situated cognition [6]. Bateson’s communication theory provides a framework for identifying stimulus and response patterns from everyday communication in which the patterns are associated to the notion of context. This theory focuses on the set of stimuli (called the external event system) that is responsible for a person’s response(s) in a repeated manner during a specific context. Clancey’s situated cognition was developed from the observation of work practice and explains how people form knowledge and experiences. Specifically, it looks into how people formulate coordinated actions by relating what people see and what they interact with in their environment to memory. Our analysis approach [3] extracts stimulus and response patterns to uncover physical artefact-communication relationships. A pattern in our work refers to repeated responses exhibited by team members during similar contexts of communication [2]. The patterns are in the form: associated with a context. where the tuple {external stimulus, internal stimulus and focus} is known as a feature, and each feature is then associated to a context. In our work, we were interested in contexts related to requirements External stimulus refers to some specific detail(s)

Communication in Context: A Stimulus-Response Account of Agile Team Interactions

169

from the environment (including artefacts) that caused the individual to react in the observed way at the present moment. There can be several external stimuli at any one time [2]. Internal stimulus refers to dialogue contents or thoughts associated with the external stimulus (i.e., associative thoughts), which can be inferred as remembering [2, 6]. Focus refers to the issue which a set of utterances is about [2]. Once we have extracted the physical artefact-communication relationships, we identify the basic stimuli that constitute the external event system of the agile team, and use situated cognition to explain what the basic stimuli represent for the team.

4 Findings In this section we introduce the eight patterns that we found associated with the contexts of gathering and clarifying requirements. Gathering requirement is defined as when the utterance(s) indicates a response to customer’s user story or developer’s progress. Clarifying requirement is defined as when the utterance(s) indicates a question or an issue raised by the developer to another stakeholder. The eight patterns are listed below, in feature format, followed by an explanation: Patterns from the stand-up meetings Pattern 1 (8 instances): associated with clarifying requirements. When an Agile member is about to start clarifying what he did yesterday, he refers to the relevant story card, and then he communicates. Pattern 2 (22 instances): associated with clarifying requirements. When an Agile member is clarifying what he did yesterday, his pair will contribute to his clarifications. Pattern 3 (6 instances): associated with clarifying requirements. When members as a group confirm that there is no longer the need to clarify what they did yesterday, they look at the physical artefacts together, and then they communicate. Pattern 4 (9 instances): associated with gathering requirements When a team member is about to start gathering details from a new user story, he refers to the physical artefacts, and then he communicates. Pattern 5 (12 instances): associated with gathering requirements When a team member is contributing details to the new user story, another member will contribute to his communication. Pattern 6 (2 instances): associated with gathering requirements When the team as a group decide that they have finished gathering requirements for the new story card on the wall, they look at the physical artefacts together, and then they communicate.

170

N.N.B. Abdullah, H. Sharp, and S. Honiden

Patterns form the pairing sessions Pattern 7 (29 instances): associated with clarifying requirements When pairs get stuck during pairing, another member will contribute to the pairs' communication. Pattern 8 (6 instances): associated with clarifying requirements When pairs get stuck during pairing, one of them refers to the physical artefact, and then they resume their pairing. In each of the eight patterns, the external stimulus was either the physical artefacts (i.e. story card or the wall), or another individual’s communication. We can generalize that the physical artefacts and other individuals were the two fundamental stimuli that allowed the team members to start, sustain and end their requirements contexts throughout time. According to Bateson’s definition, we say that the external event system for the team comprised the physical artefacts and other individuals.

5 Limitations Our analysis is based on a very small set of data from one co-located team, studied for only 4 days. Our data came exclusively from a large commercial company in the UK, and the extent to which our results are valid for other companies or contexts is not clear without replication and further iterations. Additional data and further iterations are needed to verify, refine and expand this initial set of patterns.

6 Discussion and Conclusions Using Bateson’s communication theory [2] to account for how individuals react to the external world, we have identified that the external event system of the agile team comprises physical artefacts and other individuals. Bateson tells us that if we remove either one of the stimuli (physical artefacts or collaborating members), we can predict that an individual’s response will be slightly different from what we expect. According to Clancey’s situated cognition [6], a person formulates coordinated action by relating what he/she sees in the environment to memory. During the construction of ongoing activity, knowledge, and experiences are formed, enabling team members to formulate coordinated actions together. In order to ensure that a group of people formulate coordinated actions together during work activity, they need to communicate together where they are able to share what they physically see and do together in the environment situated to their ongoing activities. We then may predict that if one of the team members is absent, or that the team members are distributed, then discoordinated actions may be formulated. On the one hand, these implications are not surprising. However their significance lies in the fact that we have begun to relate the kind of communication that takes place in an agile team to theories of communication and learning. For agile practitioners, these preliminary findings indicate that we may be able to predict when discoordinated actions will occur, and suggest how to avoid or rectify them. For agile

Communication in Context: A Stimulus-Response Account of Agile Team Interactions

171

researchers, we have illustrated how communication theory may be applied to agile practice. Future work will analyse further communication data both from this team and from other teams.

Acknowledgements We are grateful to the agile team members and customers for their participation, and the Agile Alliance who provided partial funding for this work. We are also grateful to Dr. William J. Clancey from NASA Ames Research Center, California and the Florida Institute for Human and Machine Cognition, and Dr. Robert GM Hausmann from Carnegie Learning Incorporation, Pittsburgh for their constructive comments on our study, helping us to move our study forward.

References 1. Atkinson, J.M., Heritage, J.: Structures of Social Action: Studies in Conversation Analysis. Cambridge University Press, Cambridge (1984) 2. Bateson, G.: Steps towards Ecology of Mind. Collected Essays in Anthropology, Psychiatry, Evolution, and Epistemology. University of Chicago Press, Chicago (1972) 3. Binti Abdullah, N.N., Sharp, H.C., Honiden, S.: A Method of Analysis to Uncover Physical Artifact-Communication Relationship. In: Proceedings of the 23rd FLAIRS conference. Special track: Cognition and AI: Capturing Cognitive Plausibility and Informing Psychological Processes. AAI Press (2010) 4. Bryant, S., Romero, P., du Boulay, B.: The collaborative nature of pair programming. In: Abrahamsson, P., Marchesi, M., Succi, G. (eds.) XP 2006. LNCS, vol. 4044, pp. 53–64. Springer, Heidelberg (2006) 5. Cao, L., Ramesh, B.: Agile Requirements Engineering Practices: An Empirical Study. IEEE Software, 60–67 (January/February 2008) 6. Clancey, W.J.: Situated Cognition on human knowledge and computer representation. Cambridge University Press, Cambridge (1997) 7. Gallardo-Valencia, R.E., Sim, S.E.: Agile Validation is continuous and collaborative: A field Study of Agile Requirements presented at Agile. Research-in-Progress Workshop (2009) 8. Green, T.R.G.: Cognitive dimensions of notations. In: Sutcliffe, A., Macaulay, L. (eds.) People and Computers V, pp. 443–460. Cambridge University Press, Cambridge (1989) 9. Grunbacher, P., Hofer, C.: Complementing XP with Requirements Negotiation. In: Wells, D., Williams, L. (eds.) XP 2002. LNCS, vol. 2418, pp. 105–108. Springer, Heidelberg (2002) 10. Hutchins, E.: Cognition in the Wild. MIT Press, Cambridge (1995) 11. Moløkken-Østvold, K., Furulund, K.M.: The Relationship between Customer Collaboration and Software Project Overruns. In: Proceedings of Agile 2007, pp. 72–83 (2007) 12. Nawrocki, J.: Extreme Programming Modified: Embrace requirements engineering practices. In: Proceedings of RE 02, pp. 303–310 (2002) 13. Robinson, H., Segal, J., Sharp, H.: Ethnographically-informed empirical studies of software practice. Information & Software Technology 49(6), 540–551 (2007) 14. Sharp, H., Robinson, H.: Collaboration and Co-ordination in mature eXtreme Programming teams. International Journal of Human-Computer Studies 66, 506–518 (2008) 15. Sharp, H., Robinson, H., Petre, M.: The role of physical artefacts in agile software development: two complementary perspectives. Interacting with Computers 21(1-2), 108–116 (2009) 16. Wood, L.A., Kroger, R.O.: Doing Discourse Analysis. Methods for Studying Action and Talk in Text. Sage Publications, Thousand Oaks (2000)

Suggest Documents