REACTIONS TO READINGS DUE NOV 15TH
                                       
    Christian mogensen@cs.stanford.edu
    http://www-pcd.stanford.edu/mogens/377/reaction-1115.html 
    
    
     _________________________________________________________________ 
   
                               QUALITATIVE DATA
                                       
   Robson
   
     * Everything you ever wanted to know about qualitative analysis and
       didn't want to ask. 
     * If he could restrain himself to just one paragraph per heading, it
       would be a much better guide. i.e. it might be short enough to be
       useful. 
     * There is a zeal in documenting and listing everything here:
       listing some of the possible memos you could write yourself while
       performing a case study; guidelines, memos, for the writing of;
       tactics, conclusions, for the drawing of; ad nauseam. 
     * If the chapter was reorganized, the amount of detail would make
       sense: start with high level overview, then describe each method
       in one paragraph. Finally let the reader refer to the back of the
       chapter for the gory details on, say, causal networks. 
     * It's interesting that the suggested file structure for qualitative
       analysis is hyper-textual. "All files are cross-referenced." Any
       reason for this is not given however. 
     * The qualitative methods described would be very useful in any
       process of understanding. 
     * The editor still hasn't discovered page breaks. 
     * I liked: the first paragraph of each section. 
     * I liked: the discussion of displaying qualitative data. 
     * I liked: the
       credibility/dependability/confirmability/transferability
       distinction in the conclusion. 
     * I liked: the brief discussion of computer based tools. I wished
       there were more discussion of general types of tools. Knowledge
       management software like Lotus Notes or the early Lotus Organizer
       would be dynamite here, since they generate links on the fly based
       on keyword weightings. 
       
                            STORYBOARD PROTOTYPING
                                       
   Madsen and Aiken
   
     * A basic paper. Not much quantitative data there. Not even much
       qualitative summary of tests. On first reading it seemed neat
       enough, but on rereading I felt it was shallow. Ok - it is a
       preliminary report, but it is frustrating not to see any sort of
       data or analysis. 
     * Their definition of Storyboarding (they quote themselves) is at
       odds with my own: storyboarding is a rough pictorial outline of a
       story or interaction. Then again, my definition is coloured by the
       Filmmaking and HCI course I took last year. What they call a
       storyboard I would call an interface prototype or mock-up. You can
       buy special tools to help you build these things. (HyperCard,
       Visual Basic, DemoII, TK are all examples) 
     * Their HyperCard extensions are limited to "limited interaction
       modes": pushing buttons. They take a $1000 Mac and turn it into a
       $15 VCR control panel. Now that's progress! This would be
       admirable had the point been to model and analyse the existing VCR
       interface. Unfortunately the point was to model the user-computer
       interface, where there is no point in limiting oneself to the
       limited modes of teeny buttons. 
     * A better approach (HyperCard is only good for prototyping this
       tool, as they note) is to model types of control, and allow them
       to be rendered in different ways: slider, up/down arrows, scroll
       bar, etc. This is what is known as a UIMS (User Interface
       Management System). What you want to be able to do is say: "these
       buttons should be a slider", not just "these buttons should go
       there". 
     * When they talk of creating families of systems, it is clear they
       haven't thought through their examples: Space stations,
       country-wide real estate systems. They are describing large-scale
       UIMS with embedded semantic knowledge (ie: this requirement means
       display these controls), something that has little to do with
       user-interface mock-up ("storyboards") editing tools. 
     * I liked: the idea of using interaction playback to help debrief a
       user. This apparently led to richer commentary - of course, one
       must be cautious of post-facto rationalizations for actions. 
     * I liked: idea of letting users play with a UIMS-like system. 
     * Note: their On/Off buttons would give Don Norman heart attack. A
       blob in a circle means ON? Or does it mean off?