🚀 go-pugleaf

Hello task analysis fans!

Several weeks ago I sent out a message on usenet requesting information from
people regarding task analysis in HCI.  I got several replies, and here is the
summary (sorry for the delay).  

Parts from the original message:
  > Subject: GOMS/CLG/TAG/ETAG/ETIT/TAKD/etc...
  > Newsgroups: comp.cog-eng,comp.human-factors
  > 
  >  I would like to see complete analysis *and* hear your horror
  >  stories (or your blue-sky and happiness stories).  I.e., what
  >  they (TA Methods) could or could not do in a real-world setting.
  
Several responded that was "more talk than action" in using task analysis
in commercial environments.  Why?  

From: Rob Procter <rnp@dcs.edinburgh.ac.uk>
  > My suspicion is that the lack
  > of documented case studies of TA is due to the fact that TA is unusable for real
  > projects. Diaper admits as much in ``Giving HCI Away'' (in Proc. HCI'89 People and
  > Computers V, (1989), Cambridge University Press, pp. 109-117), though he blames
  > the situation on the lack of "good tool support". At the HCI'91 conference, a
  > speaker frankly confessed that "TA was a mess"; endless papers on the "new, best
  > TA method" yet, but little evidence of its utility. There is a school of thought
  > that the whole TA enterprise is based upon a flawed model of human activity.
 
From: Jon (J.A.) Meads <JMEADS@BNR.CA>
  > The sad truth appears to be that there is more talk about the need
  > to do it than there is the opportunity. In all the major companies
  > I've worked for, there has been a serious reluctance to let people
  > from development (whether designers, software engineers, or human
  > factors types) talk to customers.
  > 
  > My experience with the real world (US industry) is that there is too much
  > of an emphasis on movement, not direction. Managers tend to be happy if
  > they see activity and results. The fact that the results they see may not
  > be relevant to the needs of the users of their products is not necessarily
  > important to them. Top US management is beginning to recognize this. But
  > the focus has been on manufacturing and commodity level products. In high
  > tech companies, there is also a growing awareness but they haven't been
  > able to work the techniques into the system yet. And they won't as long as
  > the reward system is based on schedules and budgets rather than end
  > customer satisfaction.
  > 
  > The bottom line is that between organizational structures which keep the
  > customer/end user removed from the developer, software engineering
  > processes designed for batch~process systems, and development schedules
  > which key off market windows, you don't have much task analysis being done
  > in the real world. It will change (or we'll lose the software industry to
  > Japan, Singapore, India, etc.)
  

These messages paint a bleak picture.  Are TA methods used in industry?
Yes, but obviously there are problems with applying these methods.

Putting aside this problem, what *work* was reported?  Sadly, there were
hardly any reports of on going research.  This does not mean that research
has stopped, far from it, only that I did not get responses.  What I did
get was:


From: borkoles@dcs.qmw.ac.uk (Joe Borkoles; Phd)
  *** ETKS (Extended TKS) Resources:
  > see HICSS 92, 'ETKS : Generative Task Model in User Interface Design', vol 2.,
  > Jan 7 to 10, Kauai Hawaii.
  > The paper includes a summary of an analysis. 

From: G de Haan <dehaan@cs.vu.nl>
  > Two main streams of research involve using ETAG for design
  > purposes and using ETAG for knowledge representation (and
  > to base help systems upon).
  > 
  > I am not directly in contact with industry, but Tauber and
  > van der Veer have given Post-Academic courses for designers
  > who are usually working in industrial settings. They report
  > that ETAG is very usable, but also that it may be a bit too
  > large for practical usage. As such, ETAG is taken as a 
  > source for inspiration and parts of it are used (I consider
  > it as a big leap forwards when we have got designers to
  > consider the conceptual elements with which they confront
  > the user).

  *** Resources:
  > de haan, van der veer and van vliet (1991) formal modeling techniques
  > in human-computer interaction. Acta Psychologica 78, 27-67. a review of
  > various formal models and what's good/bad about them.
  >
  > de haan and muradin (1992) a case study on applying ETAG to the design of
  > a human-computer interface. Zeitschrift fuer psychologie. a case study on
  > ETAG applied to task-analysis.
  >
  > de haan and van der veer (1992) analysing user interfaces: ETAG validation
  > studies. A quite (10 pg's) extended abstract which reviews several student
  > projects on applying ETAG to tasks analysis, submitted (and probably 
  > accepted by) to the EACE/NCGIA workshop on task analysis in human-computer
  > interaction, June 9-11, Schaerding, Austria.

From: Lars Oestreicher <larsoe@csd.uu.se>
  > We are currently closing a three
  > year project focusing on task analysis and system design as complementary
  > research areas to achieve workable task support  at work places. The
  > project has been a closely connected interdisciplinary research work
  > involving both computer scientists (of which I am one), as well as
  > psychologists.
  > 
  > However, regarding your initial question we found that the currently
  > available methods you listed (yes, I know all of them except ETIT, more or
  > less to the bone) do not really constitute a help in system design oriented
  > task analysis. We are therefore testing a symbiotic approach between task
  > analysis, formal descriptions and rapid prototyping. Just to make sure that
  > we would not merely be advocating a "yamta" (yet another method for task
  > analysis) we did some investigations of the more advanced of the methods,
  > ETAG, TAKD, and HTA just to see the weak points for the purpose of system
  > design.

Honestly, I did not get a big response to my primary mail message.  But, I have
included cuttings from email that I did receive. I appreciate all who responded,
and I look forward to future correspondence.  Also, I will be working at CHI'92
(Ph.D. student volunteer) so perhaps we could cross paths.

Thank you again for your responses.  Hope to hear from you in the future.
--
+-----------------------------------------+-------------------------------+
| Kevin A. Mayo (mayoka@vtopus.cs.vt.edu) | The shortest path between any |
| Computer Science Department / VA Tech   |   two points is through hell. |
| Blacksburg, VA  24061 USA               |                               |
+-+------------------------+--------------+----------+------------------+-+
  | 703-231-5475/6931 (W)  | My thoughts are my own. | 703-951-4337 (H) |
  +------------------------+-------------------------+------------------+