A centre of expertise in digital information management www.ukoln.ac.u k UKOLN is supported by: Salience, Focus and Bandwidth Emma Tonkin – Research Officer Strathclyde Workshop on Distance Conferencing – 22 nd October 2009
Jun 27, 2015
A centre of expertise in digital information management
www.ukoln.ac.uk
UKOLN is supported by:
Salience, Focus and Bandwidth
Emma Tonkin – Research Officer
Strathclyde Workshop on Distance Conferencing – 22nd October 2009
A centre of expertise in digital information management
www.ukoln.ac.uk
Why remote participation?
• Simple premise:
Travel costs £money, £time and £patience
Remote participation
= Cut down on travel
=> Cheaper, faster, less irritating?
A centre of expertise in digital information management
www.ukoln.ac.uk
Really?
• Remote access has penalties
• Limited opportunity to interact/contribute
• Sense of presence limited – doesn’t feel like being there
• ‘Incidental’ interactions limited – doesn’t work like being there
A centre of expertise in digital information management
www.ukoln.ac.uk
Plan…
I will discuss:
1 Brief review of factors
2 Scenarios + Experiences
3 Some of our research in the area
A centre of expertise in digital information management
www.ukoln.ac.uk
Factors
• All the obvious questions:• What? (are you trying to achieve?)• Who? (will be taking part?)
• How many people? • What are they used to?
• How much? • Money – equipment – bandwidth – time?
• How best to emulate and support?
A centre of expertise in digital information management
www.ukoln.ac.uk
User-centered
• Participant buy-in is vital
• How good is the user experience?
• Obligatory pseudo-formula:
€
Opt rp = Funcsemulated relevance +immediacy +presence( )
cost proc * cost hw * cost sw * bandwidthreq( )
A centre of expertise in digital information management
www.ukoln.ac.uk
Relevant theory from CMC
• Cues filtered out – is CMC poorer, because [non-verbal] cues aren’t transmissible?
• Cues filtered in – despite the lack of inherent [non-verbal] cues, people will find a way?
A centre of expertise in digital information management
www.ukoln.ac.uk
A centre of expertise in digital information management
www.ukoln.ac.uk
Scenarios
• Conference keynote/paper presentation
• SIG/board meeting
• Social conventions and etiquette:• Well-understood for each scenario
• Potentially easier to support than arbitrary interactions!
A centre of expertise in digital information management
www.ukoln.ac.uk
Conference keynote: rough rules• During presentation:
• Watch quietly, even if you disagree (but it’s ok to mutter darkly to your neighbour)
• After presentation:• Signal intent to ask question to session
chair (who explicitly directs focus of audience attention)
• When asked, ask question; discuss.
A centre of expertise in digital information management
www.ukoln.ac.uk
Problems for remote viewer
• Boring video streams: Not enough detail, too few cues to guide gaze – faded into background
• Some nuances are lost
• No sense of presence; no incidental interaction
• Signaling intent to contribute
A centre of expertise in digital information management
www.ukoln.ac.uk
Strategies (1)
• Use low-bandwidth channels (twitter, irc, chat) for external contributions
• Alter conventions for the chair to accommodate remote participants
• Direct gaze through video stream; qualitatively appears more immersive
A centre of expertise in digital information management
www.ukoln.ac.uk
Strategies (2)
• Obvious approach: improve quality of transmitted data
• Problem: typically taken to mean ‘higher resolution video’
• Much more expensive. • Most additional data irrelevant, so
filtered out – by viewer! (visual attention)
A centre of expertise in digital information management
www.ukoln.ac.uk
Strategies (3)
• What really makes a difference?
• In our case, we found: • High-quality audio*• Relevance of data transmitted; shot,
focus, framing… camera directs interpretation?
• Simple mechanisms for contribution• Fielding social s/w: twitter, etc.
A centre of expertise in digital information management
www.ukoln.ac.uk
* For certain definitions of ‘high quality’
• Psychoaccoustic/psychovisual modeling• Perceived quality… depends on perception
A centre of expertise in digital information management
www.ukoln.ac.uk
Further aside about audio…
• ‘Cocktail party effect’ – focusing listening attention; source separation
• Permits us to locate & focus attention on a speaker; as accurate as visual localization, but less efficient
• Required info to achieve this is difficult to retain – but very important
A centre of expertise in digital information management
www.ukoln.ac.uk
Board meeting: rough rules
• Session chair controls discussion, explicitly directs attention
• Signal intent to contribute
• Meeting minutes taken by nominated individuals; actions identified, declared during discussion
A centre of expertise in digital information management
www.ukoln.ac.uk
Problems-remote participant
• Cues and signals
• Gaze/attention (what’s so funny? What are we looking at/talking about?)
• Identity of participants; visual name tags?
• Fault-tolerance: Time zone, jet lag and burnout rate
• Handling language difficulties
• Bandwidth limits
A centre of expertise in digital information management
www.ukoln.ac.uk
Strategies (1)
• Build effective conventions between chair and remote participants
• Create fault-tolerant procedures• Circulate notes/actions within minutes • Take turns to complete tasks, so that
departure of one participant is minimally disruptive
A centre of expertise in digital information management
www.ukoln.ac.uk
Strategies (2)
• Focus on what matters:• Video dispensable, given desktop
sharing or even async file sharing• High-quality audio is very important• Back-channel enables informal
discussion between participants, and can mediate requests to contribute
A centre of expertise in digital information management
www.ukoln.ac.uk
Research
• Best use of limited bandwidth/dev time in various contexts
• What data to collect/use/represent?
• How to represent it?
A centre of expertise in digital information management
www.ukoln.ac.uk
Services shared by local and remote participants• Long tradition of hackery –eg. graffiti
wall (bluetooth & web), etc.
• Cheap, fun, simple to set up – but little active interaction sparked as a result
A centre of expertise in digital information management
www.ukoln.ac.uk
Conceptual/domain model development scenario
A centre of expertise in digital information management
www.ukoln.ac.uk
Usual solution - committee
• Coffee in, specification out
• Goal: shared reference, ‘common ground’
• But: Expensive, limited attendance
• CMC problem. Multi-touch – Where should I be looking? What just changed? Is someone else about to edit what I’m looking at?
A centre of expertise in digital information management
www.ukoln.ac.uk
Strategies
• Versioning/diffs
• Simplified visualisation – remote user relative position; distance from screen
• Audio significant; conversation analysis a useful tool in eliciting design issues.
A centre of expertise in digital information management
www.ukoln.ac.uk
Discussion
• Smarter hardware, software, etc.• collect relevant, pre-filtered information• Cheap to transmit• But how easy is it to interpret?• Prior work - reaction times to filtered
data can be very fast. Significance+ equivalence of data must be taught, learned, ’intuitive’
A centre of expertise in digital information management
www.ukoln.ac.uk
Overall Conclusion
• More detailed data not necessarily better• Getting the right data
• Lots of design time goes into figuring out which data matters most
• Representation may differ significantly from initial form of information
• Hearing gaze and seeing speech? – ambient + intuitive
A centre of expertise in digital information management
www.ukoln.ac.uk
Conclusion (2)
• Handy buzzword list:• Emergent, self-organising, etc…• Agile, user-centred design.
• Imagination is free
A centre of expertise in digital information management
www.ukoln.ac.uk
Comments, questions, rotten tomatoes?