See also: IRC log
<Steven> [Is the UML appropriate for Interaction Design?]
<Steven> Giorgio: Usability can be made better by providing tools that allow early evaluation
<Steven> ... you have to go into the details
<Steven> ... at the abstract level there are problems that cannot be caught
<Steven> Scribe: Steven
Giorgio: There are several
approaches to model-based development
... most miss one important aspect
... they focus on the main model
... whis is not the right way to go when you want to focus on
usability first
... state machines can help you here
... it is important to focus on details that affect
usability
... at abstract and concrete levels
... one possible platform is wireframes, for debugging and user
testing
... you can focus on accessibility
... such as testing without mouse access
... forcing you to show that you can navigate from the
keyboard
... you could easily identify anti-patterns
... generate test scripts automatically
... analyse the interaction graph, prove reachability, and
distance betwen states
[case study]
[Slide with state machine]
[Domain model]
[Wireframe prototype]
[Interaction graph]
Giorgio: The graph is complex, reflecting the complexity of the interface
Jose: I disagree that
... I disagree that this is helpful
Giorgio: This allows you to test assumptions
Jose: In the real world no
Giorgio: It allows you to test for consistency
[Usability problems]
Giorgio: There is for instance an
unnecessary action for the user to return from the error
message to the login screen
... potential usability problems like this can be extract4ed
from the state graph
... State machines offer no expressivity limits
... don't ask you to express things you don't want to
Fabio: dynamic construction of interaction elements may be an expressivity limitation
Giogio: You could activate tasks
Fabio: You don''t know how many times it might be activated
Jochen: The URC has been an ISO
standard since 2008
... there is a pilot store of 'apps' already running
... openurc.org is the consortium for the urc
Fabio: How do you connect the task model
Jochen: Done manually
... from task model to socket elements
Fabio: In what language?
Jochen: In the abstract
description
... there is a task engine that plugs in
... the clients can be in any language, phones, PCs, java apps,
c++
... the protocol is the important bit
aaa: How does the user integrate descriptions from resource server, to make updates on the UI
Jochen: This is still open
aaa: I have seen some RDF descriptions from the server
Jochen: The resource description
uses RDF, but I am not an expert in this
... we are thinking about integrating semantics
Javier: MOSKitt is a set of
modelling tools
... there is no standard UML, and we need that really
... so we have just chosen one
bbb: The model-based approach is
for richer applications of modern applications
... richer interfaces
... not for simple login screens
... practitioners don't see why you need a model-based approach
for a login screen
... look at facebook and similar
Javier: We need to know how to
model things, but for the new things there is no consensus
yet
... we should have at least one standard language for the basic
things
bbb: We should take in the new requirements when thinking about standardisation
dsr: We have a number of areas of
discussion (see the agenda)
... What criteria determine whether a model-based UI design
framework is appropriate to a given application? What are the
main scenarios and domains where MBUI is valuable? Provide
counter examples where it isn't appropriate.
Jose: The problem is to identify when the model based approach can be applied
Javier: I think that modelling can be used in practice
dsr: Are there counter examples?
Fabio: MB approach is good for
multiple devices
... Not needed if you are only using a desktop system, except
perhaps in safety critical systems
Daniel: Is this the right
question to ask?
... all methods are model-based in some sense
... even when just programming
... sometimes using low-level machinery
... but the programmer must have some kind of model
dsr: So the model is
implicit
... no reasoning about it
... is possible
Daniel: So what level should we go to
Pablo: The question is which level is appropriate
Daniel: We have to be more specific that just saying "model-based"
Pablo: When environments change, you need a model
Giorgio: It allows you to hide detail, and expose detail
dsr: For Dynamic adaptation,
having a runtime model is important
... Accessibility is an example of this
... but going back to my question, are there examples that are
not appropriate?
Jose: [Scribe missed]
<fpatern> models useful for accessibility, changing contexts, multi-device interfaces
Jose: something about the difference between concrete and abstract specification
ccc: How is the finalization of the interface reflected in the model
Jose: Good question
ccc: We need to address how you create a fabulous-looking website
ddd: It is hard to reverse
engineer when there is no task
... no related task in the task model
ccc: Some approaches decouple both layers
Daniel: Most do
ccc: The final appearance is really important
Daniel: Most approaches allow for
that, I am sure
... since you can generate arbitrary style markup
... I understand that many designers don't know how to use
it
... but look at css Zen Garden
... it is possible
ccc: That is a question of best practices
Daniel: I mean that most
approaches can decouple task process from style
... from look and feel
... some good designers are good at look and feel and
interaction design
... but you can use different people
... Artistic designs are examples of where you don't need task
designs, where there is no task per se
ddd: A real example is expression planet (?)
Giorhio: Another question: how usable are the tools?
Steven: You mean they haven't
applied their own tools to their own tools?
...: -)
[Coffee break, 20 mins]
http://homepages.cwi.nl/~steven/vandf/2002.5-pixelpoint.html "Sooner or later it will get home to web designers not to design in pixels anymore, because each page will fit on a postage stamp if they do. "
"But at what point will they realise they shouldn't be designing their sites with fixed sizes at all, pixels or points, but using percentages, and letting the user decide on the base size, so that users can increase the base size to make the fonts larger and therefore more readable? Probably when their own sight starts to fade, around retirement age I suppose. "
<MichaelN> :-)
<kaz> scribenick: kaz
Q&A:
* modeling is needed but difficult
* which is easier, programming itself or modeling?
* usability testing?
- how to structure usability testing?
- different level of analysis could be useful to usability testing as well
(Fabio will take note online)
This is scribe.perl Revision: 1.135 of Date: 2009/03/02 03:52:20 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/dias/dis/ Succeeded: s/r, /r / Succeeded: s/,J/, J/ Succeeded: s/Are the/Are there/ Succeeded: s/nant/ant/ Succeeded: s/ate/ate?/ Succeeded: s/plaet/planet/ Succeeded: s/./?/ Succeeded: s/@@@/wrapping up discussion/ Found Scribe: Steven Inferring ScribeNick: Steven WARNING: No scribe lines found matching previous ScribeNick pattern: <kaz> ... Found ScribeNick: kaz ScribeNicks: kaz, Steven Present: See_day_1 WARNING: Fewer than 3 people found for Present list! Agenda: http://www.w3.org/2010/02/mbui/program.html Got date from IRC log name: 14 May 2010 Guessing minutes URL: http://www.w3.org/2010/05/14-mbui-minutes.html People with action items:[End of scribe.perl diagnostic output]