[drupal-docs] Day Two Computer Human Interaction 2005

[Kieran Lal]

This morning I spent a lot of time running around doing guerilla 
marketing for the Open Source meeting I had this evening.  Jump to the 
bottom if you want to learn about that.

I attended the mid-day sessions on methods and usability.  The first 
paper focused on how developers responded to usability problems versus 
redesign proposals.   This is very interesting for the 
CivicSpace/Drupal community because we are interested in learning how 
best to improve our usability.   The results were that developers 
responded much better to redesign proposals then they did to being told 
about usability problems.   Developers indicated they were frequently 
aware of the results of user testing that highlight usability problems. 
  Instead they appreciated being presented with usability designs which 
gave them ideas, helped them support decisions, and helped prioritize 
efforts.   Developers also said that redesigns were more concrete and 
constructive then being presented with usability problems.   The 
interesting conclusion of the study was that developers still wanted to 
be presented with both usability problems and usability redesigns.  My 
inclination based on this is to recommend CivicSpace hire another 
designer who can provide more redesigns to all the Drupal developers to 
help improve the usability.

The second talk was about creating a single score for evaluating 
usability.  We see this often with movies getting 4 out of 5 stars, or 
JD powers giving a rating on a car as a single score.  This effort that 
started at Intuit attempted to combine quantitative metrics as well as 
subjective metrics.  They used task completion, time on task, error 
counts, and post task satisfaction (the subjective rating).   They 
mangled the scores with correlations and other statistical hand waving 
to get a final single score.   I could see this being useful in large 
corporate research groups that have a lot of usability data.

The third talk was on creating limited models of users minds and how we 
can anticipate how skills training will improve the usability of an 
interface.    This is a fascinating area because it bridges real 
science with user interfaces.  But most of the scientific modeling 
infrastructure is written in 30 year old programming languages and 
isn't yet that accessible.   This is very exciting and as these 
research project continue to improve their tools we will see lay people 
being able to rapidly evaluate user interfaces in scientifically valid 
ways.  Search on Cogtools if you are interested.

Short papers on social computing and community.
How do people collaborate in complex shared work?
They chose to study Requests for Proposals (RFPs), a way of asking for 
bids on contracts.  This area of research is known as Activity Centric 
computing.  The first phase of this research was to conduct user 
interviews and then followed with participatory analysis.  Basically, 
they made a structured list of shared tasks to be done.   Here is what 
they found was involved in RFPs: planning, doing, reflecting, re-using. 
  They found that complex tasks involve landmarks.  For RFPs the 
landmarks were: Docs, dates & schedules, systems and databases, roles 
and personas, events.  If this research proves valuable then landmarks 
could be an important concept in supporting complex work.

Relescope: An experiment in accelerating relationships.
This talk was given by Steve Farrell , a former colleague of mine.  It 
was a good example of why most modern social networking systems suck, 
in comparison.  He asked the question "Are these two people going to 
interact with each other?" How do we help people innovate by connecting 
them.  240 of the CHI attendees who had published papers were given 
relationship reports recommending who they should seek out based on 
likely common interests of their past publications.   This was useful, 
but many people who hadn't published were the most likely to want a 
relationship report and didn't receive one.  Linkedin, are you paying 
attention?

How peer photos influence member participation in online communities
I disagreed with his premise that online communities succeed because 
people participate.   I argued that online communities succeed because 
they have good missions, and reflect the off-line community.  It's 
scary how many researchers base their understanding of online 
communities on Slashdot.

The Uses of Personal Networked Digitial Imaging
The talk focused on two reasons we share photos using camera phones.  
In short, we use photos as a relational activity, and as a chain of 
evidence of what we are doing.   Sharing photos is currently difficult 
and needs to be improved.  The value of photos is interpretive and 
flexible.  The researcher said that a casual photo of her nephew on a 
road trip with friends would have been invaluable, because he died on 
the trip.

I saw this photo and thought of you
This is important because the revenue hopes of camera phones has been 
disapointing in the US.   What is the emergent behavior of camera 
phones?  We take networked photos to bring people into the experience, 
it was coined social touch.  In particular top uses were to share with 
absent friends and family.  The second top use was capture remote 
tasks, i.e. which shoes should I buy?  The challenges are to develop 
web based forms of sharing, categorization by value,  and to help 
photo's integrate into larger dialogues.  For example, I could argue 
with Drupal developers that we need more user experience designers.  Or 
I could send a picture of the large number of young women who seem to 
dominate graduate work in user experience education to make my point.

Human Computer Interaction in the office
I stayed on topic and stuck to the short paper track on topics that I 
felt would most greatly improve my understanding of research relevant 
to CivicSpace.   The first talk was on Don't take my folders away.   
Basically, we need to be able to tag folders so they can be ordered and 
labeled.   Also, the structure of folders was important and frequently 
should be reused.   i.e. The subfolders under CHI2004, are basically 
the same as CHI2005.   This talk was followed by two talks on email, 
which are really important from a practical standpoint.   The biggest 
problem I had with these studies was that basically the user studies 
were very biased towards the use of these products, because their 
company made them, or did research on them.   The questionable results 
from IBM showed that managers had twice as many top level categorized 
folders and three times as many second level folders as regular users.  
  More importantly they showed that only 40% of emails were task related 
and much of the rest had to do with relationships.   So look for 
relationships categorization not just to-do lists in your next email 
client.

Email Triage from Microsoft
Email triage is not very well supported in most email clients and is 
frequently done on personal time at home.   There are two main 
techniques for triage.  First, is single pass which involves 
sequentially triaging mail.  The second is multi-pass which involves 
scanning mail to fill a category, i.e. mail from bosses.  The biggest 
complaint again was that email clients lacked the ability to indicate 
social context.  So email from one team should indicate that's it's 
from a team you are working on, and email from peers you are meeting at 
an upcoming conference should indicate that it's related to meeting 
them soon.  Instead we get information like mail size, and time mail 
was sent.

Open source Birds of a Feather.
I got 25 attendees and any one of them are of the caliber to make a big 
difference in any open source project.   We got a wide sample from 
interested HCI grad students who were curious, to former mozilla 
development managers, to evil empire employees who were sitting on a 
treasure trove of remote usability testing tools they had worked on at 
stint at university.    My intent was to get some momentum going among 
the HCI crowd again, now that Firefox is the poster child for usability 
in open source.   I hope to launch a mailing list in the next few days. 
  One take away seemed to be that we should start arguing that open 
source projects don't need more engineers to add functionality, but 
they need usability designers to help reduce the complexity presented 
to the user instead.

Cheers,
Kieran
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/enriched
Size: 8491 bytes
Desc: not available
Url : http://drupal3.drupal.org/pipermail/documentation/attachments/20050407/c0a7e3b7/attachment.bin