distributed work. 2 2 overview today review evidence about effects of distributed work on task &...
TRANSCRIPT
Distributed Work
22
Overview
• Today• Review evidence about effects of distributed
work on task & social outcomes in teams• Try to understand what problems arise & why
they occur
• Subsequent classes will delve into several of these problems areas in depth
• Common ground & communication effectiveness• The visual element in communication• Why face-to-face communication is distinct• Familiarity• Identity and goal conflicts• Diversity
44
Overview
• Distance collaborations are increasing• Fewer collaborations start & complete as
distance between collaborators grow• Distributed software teams take longer to
complete comparable work• Distributed scientific teams are less
successful in science, education & group maintenance
• Distributed teams endure more conflict
55
Despite problems, distance work is frequent & increasing
• How many of you have been part of distributed teams?
• 67% of companies anticipate increased reliance on virtual teams• 80% for companies with 10,000+ employees
• 35% of respondents rated difficulty of management as top challenge for virtual teams
• 92% said trust is critical for virtual teams• Survey by Institute for Corporate Productivity
66
Rise in long distance collaboration
• International scientific collaboration
Figure 1. Percent of U.S. publications with international collaborators by field, 1981-2001 [Sources: National Science Board (1993-2004),
77
• Small distances make a large difference
Figure: Probability of Joint Research Publication in an R&D Lab
Distance decreases probability of collaboration among scientists
88
Problems
99
Distributed software development takes twice as long
• Distributed software is increasingly common• But software development takes longer when performed by
geographically distributed teams• Compare software development efficiency, when all developers are
at one location or distributed across sites• Two different software development organizations• Time to complete an “MR” (Modification Request)
Study Single site Multiple site RatioEspinosa 48.2 97.2 2.0Herbsleb 5.0 12.7 2.5
Days from start to completion of modification requests
Team type
Espinosa, et al. (2007). Familiarity, complexity, and team performance in geographically distributed software development. Organization Science, 18(4), 613-630.
Herbsleb et al . (2001). An empirical study of global software development: distance and speed. Paper presented at the Proceedings of the 23rd international conference on software engineering.
1010
Cummings & Kiesler (2005; 2007)Study of Large NSF Projects
• Two studies of the outcomes of large NSF funding initiatives• 71 Knowledge & Distributed Intelligence projects
1998/99• 491 Information Technology Research Projects,
2001-2003
• PIs complete questionnaires describing• Collaboration composition • Coordination techniques used• Scientific & educational success
1111
Six types of outcomes
1212
Research Model & Consistent Findings
• Multi-university projects were less successful than single-university projects
• More successful projects used a variety of specific coordination mechanisms
• Multi-university projects used fewer coordination mechanisms than single-university projects
• Reduced use of coordination mechanisms mediated the relationship between # of university and performance
Knowledge Outcomes**
Tools Outcomest
Training Outcomes**
Outreach Outcomes
Collaboration Outcomes
Leverage Outcomest
Number Universities
1515
Why?
• What intervening variables influence collaboration and are harmed by:
• Distance• Difference in location or setting• Technology-mediated communication
Distance
Outputs• Performance• Satisfaction• Group maintenanceInputs Processes
? ?
1616
Why?
• What intervening variables influence collaboration and are harmed by:
• Distance• Difference in location or setting• Technology-mediated communication
Distance
1717
With distance
Coordination/Task
• Fewer communication events
• Less effective communication per episode
Motivation/Socio-emotional
• Differences between distributed groups
• Lack of common identity
1818
Frequency of Communication
•Communication declines rapidly with distance
•Communication frequency halves with doubling of distance with asymptote at ~30 meters
•Shape of curve similar for face-to-face & phone communication.
•Less steep for email & IM communication
Allen, T. (1977). Managing the flow of technology. Cambridge, MA: MIT Press.
1919
Causes & consequences of reduced communication
•Why?
•Consequences
2020
Causes & consequences of reduced communication
Why?Much communication is spontaneous/unintentional
Communication conceived as a Brownian process, enabled by chance encounters when people come together in space & time
ConsequencesLess awareness
Coordination
Information exchange
Liking
2121
Information Sharing
Information sharing is less even & complete across sites than within
• Frequency of communication• Willingness to share• Knowledge of who knows what• Effects of technology
Priorities are better aligned and coordination is more successful within sites than between
2222
Social Identity
Groups split across location tend to be less cohesive than collocated groups
• Pre-existing differences across sites (culture, local loyalties, language) cleavages
• Reduced contact less individual liking• Social categorization effects: Us vs. them
Less cooperation between sites than within sites
2323
Herbsleb et al: Differences between collocated and distributed software development
Herbsleb, J. D., Mock<us, A., Finholt, T. A., & Grinter, R. E. (2001). An empirical study of global software development: distance and speed. Paper presented at the Proceedings of the 23rd international conference on software engineering.
Survey item Collocated Distributed
I lose time trying to figure out who to contact regarding my work Low High
People I need to communicate with are difficult to find Low High
There have been times when I was accidentally excluded from information which was shared byt my coworkers
Low High
I often get useful, worked related information through casual conversation
High Low
My coworkers provide timely information about changes in current plans High Low
I feel like I'm part of the same team as my coworkers High Low
I feel accepted by my coworkers as team members High Low
My coworkers and I share the same team spirit High Low
My coworkers and I have work styles that fit well together High Low
During meetings, my coworkers and I do our best to produce mutually beneficial solutions to problems
High Low
I assist my coworkers with heavy worklaods, beyond what I am required to do
Equal Equal
My coworkers assist me with heavy workloads, beyond what they are required to do
High Low
Mot
ivat
ion
Iden
tity
Info
rmat
ion
>
<<
>>>>>>
>
<
2424
Shared Context: Cramton, Mutual Knowledge Problem
Members confronted by different contexts & pressures• Members in different organizations different value systems• Members in different nations different calendars and deadlines• Members in different units different workloads
Remote members don’t recognize contextual constraints on behavior attribute problems to personal instead of situational causes
• E.g., Absence attributed to unresponsiveness or laziness, instead of competing demands or holiday.
Cramton, C. D. (2001). The mutual knowledge problem and its consequences for dispersed collaboration. Organization Science, 12(3), 346-371.
2828
Interpersonal conflict
2929
Quality of Communication Episodes
• Many telecom technologies lack important features of face-to-face communication• Email: Non-interactive, simultaneous, sequence not preserved• Phone: Looses visual information –both context of discussion &
details of discussants• Video: Delay, asymmetrical fields of view, problems of resolutions &
control
• Technology mediated communication is generally more effortful & less effective than face-to-face communication• More difficult to develop common ground• More difficult to express subtleties of emotion
• Technology mediated communication can offer useful features unavailable in face-to-face settings & improve conversation• Simultaneous input• Archive of the discussion
3030
Resources Available from Collocation
• Shared space
• Development of common ground
• Pick up information from the periphery
• Ease of reconfiguration
3131
Collocation Improves Communication
3232