SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
The UARC/SPARC Experience:1993 - 2002
Tom Finholt
School of InformationUniversity of Michigan
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Outline
SOC functions– Primary– Secondary
Description– System– History– Organization– Funding– Incentives– Collaboration readiness
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Outline (continued)
Access Resource diagram Technology involved Successes and challenges Usage
– Analysis of user behavior• Action logs• Chat logs
– Analysis of user attitudes Conclusions
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
SOC functions
Primary– Shared instrument
Secondary– Distributed research center– Virtual community of practice
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Description: System Functional
– real time remote control of instruments– access to archival and real time data– communication with remote colleagues
Technical– Web-based, thin client model– CHEF
Support– 1-3 FTEs at Michigan (migrating to NCAR)
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Aurora image
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Description: History
UARC v.5– NeXTstep-based system, in use 1992-97
UARC v.6– java applet-based system, in use 1996-98
SPARC– Web-based, thin client system, in use 1999
- present
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
UARC 5.0 interface
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
UARC 6.0 interface
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
SPARC interface
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Description: Organization 14 investigators
– 3 computer science– 2 human-computer interaction– 9 space physics
15 students– 6 computer science– 6 human-computer interaction– 3 space physics
8 staff– 7 programmers– 1 system administrator
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Description: Funding
UARC (IRI 9216848)– cooperative agreement between Michigan
and NSF CISE & ATM– approximately $800 K per year, 1992-97
SPARC (ATM 9873025)– KDI award from NSF CISE– approximately $500 K per year, 1997-01
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Description: Incentives Funding
– scientists recruited through sub-contracts Recognition
– innovative system– highly visible to space science and computer
science communities Novel capabilities
– new ways of combining data– respond to opportunities (e.g., UCB sounding
rocket campaign)
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Description: Collaboration readiness
Multiple authorship the norm Ubiquitous email adoption 1992-2001
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Access: People
Space physicists Concentrated in N. America and Europe
(approx. 300 researchers) Affiliated with the American
Geophysical Union– aeronomy and space physics sub-divisions
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Access: Instruments
Observational– ground-based
• active– incoherent scatter radar (e.g., convection reversal
boundaries)– all-sky cameras (e.g., auroral morphology)– interferometry (e.g., winds)
• passive– riometry– magnetometer chains
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Sondrestrom observatorySondrestrom, Greenland
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Access: Instruments (cont.)
Observational– space-based
• satellite imagery (UV and visible light)• shuttle imagery• sounding rockets
Theoretical– computational simulations
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Polar spacecraft
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Instrument scope, 1993
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Instrument scope, 2001
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Resource diagram
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Technology involved 1992-97
– UARC v.5 built on NeXT machines, migrated to NeXTstep OS
1996-98– UARC v.6 built as java applets (MURAL), Web-
based 1998 to present
– SPARC built as thin client (e.g., HTTP), migrating toward CHEF-based system (e.g., portlet technology)
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Success and challenges Scientific
– first simultaneous real-time display of incoherent scatter radar chain (April, 1998)
– first combined real-time simulation and observation campaign (April, 1997)
– first archival campaign (December, 1993)– Simultaneous views of observational and simulation data
Educational– first student campaign (February, 1995)
Future– model for dozens of collaboratory efforts
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Simulation and observational data
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Social network 2/95 campaign
Michigan
Scientist
Scientist
Student
Scientist
No one
Scientist
FloridaCalifornia B
California A
StudentSite crew
Greenland
Scientist
Engineer
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Usage
UARC v.5– total scientific users = 27– peak usage = 8– 80% of use by four users
UARC v.6– total scientific users approximately 100– peak usage = 40– 80% of use by six users
SPARC– usage data in progress for the workshop
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Analysis of user behavior
Analyzed chat and action logs in order to summarize users’ experience with the UARC and SPARC systems
Focused on:– Hours of use– Frequency of use sessions– Content of use sessions (actions and chat)– Growth in use over time
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Methodology
Extracted data from automated usage and chat logs– UARC 5.0 completed (1993-97)
Standardized data Loaded standardized data into an
Access database Used SPSS to analyze sub-sets of data
from the Access database
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Results from action logs Aggregate use
– Total frequency of sessions per year– Total hours of use per year
Individual use– Frequency of sessions per year– Hours of use per year
Change in system use– Cumulative number of users over time– Number of new users over time
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Total sessions per year
209
739
322
117
33
0
100
200
300
400
500
600
700
800
1993 1994 1995 1996 1997
Year
To
tal n
um
be
r o
f s
es
sio
ns
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Total hours of use per year
700
4586
2058
447 363
0
1000
2000
3000
4000
5000
1993 1994 1995 1996 1997
Year
To
tal h
ou
rs o
f u
se
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Change in system use
1012
2 2 1
10
2224
26 27
0
5
10
15
20
25
30
1993 1994 1995 1996 1997
Year
Nu
mb
er o
f u
sers
New
Cumulative
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Results from chat logs
Aggregate use– Total frequency of turns per year
– Total frequency of words per year Individual use
– Frequency of turns per year
– Frequency of words per year
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
1786
4230
2963
2288
1028
0
2500
5000
1993 1994 1995 1996 1997
Year
Num
ber
of c
hat
turn
sTotal chat turns per year
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
29359
73063
4993640387
18939
0
25000
50000
75000
100000
1993 1994 1995 1996 1997
Total words per year
4.16x 3.17x 9.16x 7.17x 4.18x
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Key findings Rapid start, 1993-94
– Number of sessions triples– Hours of use increases six-fold– Number of users doubles
Dramatic shift after 1994– Use declines– Growth in use stops
Overall use concentrated in a small number of scientists– 25% of users generated 75% of total use
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Analysis of user attitudes
Analyzed longitudinal survey responses, 1993-2001, comparing UARC sample with matched control sample
Focused on:– Communication choices– Use of Web resources– The relationship between communication,
Web use, and outcomes
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Methodology Longitudinal survey
– The 2001 survey administration was the sixth wave of data collection dating to 1993
Response rate– 87% for UARC sample; 60% for matched
control sample Mode of administration
– 90% via a Web-based questionnaire; 10% via telephone interview
Content– 2/3 repeat items; 1/3 new items – focused
on Web use
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Results from longitudinal survey Communication choices
– Change, 1994 to 2001– Difference, UARC vs. Control
Web use– Change, 1994 to 2001– Difference, UARC vs. Control
Effect on outcomes– Relationship of communication to funding– Relationship of Web use to funding
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Key findings, 1994 vs. 2001
Substitution effect of email for other forms of communication– More pronounced for Control
Significant decline in UARC use and satisfaction after 1994– Relationship in the decline of UARC use to
respondents’ use of Web sites – Defectors
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Key findings, 2001 Some use of data intensive websites
– Concentrated in a small proportion of scientists
No predominant Web site Scientists do not read core space
science journals on-line Low use of collaboration technologies
– Data conferencing– Videoconferencing
SCHOOL OF INFORMATION UNIVERSITY OF MICHIGAN
Conclusions
Teleobservation dominates teleoperation Proof of concept – but not killer application Primary success in terms of emulation, not
total use or scientific impact Thin client ensures universal access, but
compromises capabilities The Web has not yet revolutionized the
practice of space physics