Of Deadlocks and Peopleware - Collaborative Work Practices in Global Software Development

1. Of Deadlocks and Peopleware - Collaborative Work Practices in Global Software Development Dr. Gabriela AvramUniversity of Limerick

2. Outline • Why this paper? • The research team – the observers • The distributed SD team – the achievers • The crisis situation – what we observed • What we made of it • Why do we think it’s important

3. Why this paper? • software development - a human, social and organizational activity as well as a technical activity • work practices in software engineering • People - “the” most important factor • the major problems - “not so much technological as sociological in nature." • software and hardware are made of replaceable modular components, software development teams are not.

4. Why this paper? • How does the global distribution of software development affect collaborative work practices in a particular work setting? What can we learn from the particular challenges and the ways people use for coping with them in this specific situation? • Building on: • CSCW studies (articulation and coordination work, organizational memory, knowledge management practices ) • KM studies (situated learning, knowing in practice, software engineering as knowledge work)

5. The research team – the observers • The project - Social, Organizational and Cultural aspects of Global Software Development (socGSD) • field studies of workplace activity over an extended period of time • participants are studied in real work circumstances and are engaged in continually evolving working arrangements. • data collection and analysis methods - informed by ethnography (observation, document analysis, in-context interviews, audio recording, focus groups, workshops)

6. The distributed SD team – the achievers • CROWOLF – a project in one of the Irish sites of a MNC involved in SD • Distributed team with co-located development; relatively high staff turnover; • Duration: Jan 05 – June 07 • Developing a product on top of an enterprise portal and relying on a runtime engine, both developed in house- high interdependencies • First release – June 06 – preview • System testing following – 90 days planned

7. The crisis situation – what we observed We don’t know what was wrong in the first place, how we fixed it, and now it came back! Matt Pat Claude Sean Ian

8. We’re trying to do surgery here with a butter knife to get this working again! Matt Pat Claude Sean Ian

9. Were we better a few weeks ago? The other cluster seemed to be working… only seemed to be… Matt Pat Claude Sean Ian

10. The problem might be due to a configuration error! Maybe we should set it up again – the way we did when you guys were here! Matt Pat Claude Sean Ian

11. Felix Let’s ask Felix what could be the cause for the portal behaving the way it does! Any hints? Is there anyone who tested this previously? is there any other group using the runtime engine yet?! Matt Pat Claude Sean Ian

12. Felix We’re 10 days behind today; tomorrow we’ll be 11 days behind and we’re overlapping our end date; we’re red, and we have no idea when we’ll get green again! Matt Pat Claude Sean Ian

13. We got it to work; we documented everything- there’re lots of manual steps! Pat Claude

14. And now it’s broken, and we have to put the pieces back together! Maybe I’m getting paranoid here, but … We fix it, they break it! We already got it to work two times there, and two times here… Pat Claude

15. Would they allow us to help them deploy again? Organize a meeting - and ask Matt to share what he’s doing? Pat Claude

16. You mean…us sitting on his shoulder and watching while he’s doing it again? If you can get it…but I’m sure they won’t like it! Pat Claude

17. Possible solutions: • applying a workaround - using the “portal knife” - it is only a temporary solution and it is lethal – no upgrade would be possible after that. • a re-install they would be able to assist to (“sitting on Matt’s shoulder” and monitoring his steps); it might solve the situation, but, as Claude mentioned, it could damage their relationship badly. • getting support from the German team by getting Felix involved; Felix’s team wrote the code causing the trouble when interacting with CROWOLF; they should know a workaround ( “they should have their own butter knife”.) • flying someone over to the US location– either Claude or Ian; obviously, they will need approval for this and it might collide with their personal priorities at this moment.

18. What we made of it • Knowledge management practices • “How to” – the situated learning aspect • “Who knows what?” – the transactive memory aspect • Awareness maintaining – the mutual knowledge aspect • Adding more resources to the project • The importance of social networks

19. Why do we think it’s important • For the practitioners: getting new insights, reflecting on what they’re doing • For the managers: there are other factors that need to be taken into account • For the research community: an approach that could complement quantitative studies, providing more granularity • The value of studying real people, in real work settings • The paramount role of human actors, of their values and social connections in getting work done!

20. Thank you! Questions?