[PREV - THE_SO_CALLED_HOAX]    [TOP]

ENGLEBARTS_BARD


                              Thu Jul 18 11:33:27 2002

A review of "Bootstapping" by Thierry Bardini

   Douglas Englelbart, Coevolution, and the
   Origins of Personal Computing.

   Stanford Press, 2000

   Bardini's credentials: Associate Professor of
   Communication at the University of Montreal

The good stuff here: the history of the
thinking behind the development of 
"the personal computer", getting into 
some of the details of alternate ideas 
they tried, quotations from the original 
players talking about what they were 
thinking about and trying to do, and 
how they feel about the way it played
out. 

   In broad outline, here's the story:
   Englebart at SRI, starting around  
   1959 works on the idea of "Human   
   Augmentation" using computers      
   (though this doesn't get any       
   serious funding until the mid-60s).
   He wants to build a system that can    
   be used first and foremost by the      
   system builders themselves, to make    
   them effectively smarter so they            They came up with a system 
   can continue improving the system           that sounds pretty         
   (this is the "bootstrap" process of         interesting: a five-key    
   Bardini's title).                           "chording" keyboard to the 
                                               left of the QWERTY, and a  
   Around the early seventies, this            three-button mouse to the  
   project starts to fizzle for                right.  
   various reasons and a lot of     
   people start filtering over to                   Ever get annoyed at
   Xerox, where people like Alan Kay                having to switch   
   get to work on developing the                    back and forth     
   Alto and later the Star machines.                between the mouse  
   Unlike with Englebart, their                     and the keyboard?  
   concept is to develop machines                   The original notion
   for "secretaries", so they begin                 was that you would 
   the obsession with "ease of use"                 have an alternate  
   that to this day either energizes                one-handed keyboard
   or corrupts all computer related                 to use while you   
   endeavors, depending on your                     were on the mouse. 
   point of view.  These were the   
   machines that Xerox famously        
   couldn't see how to market, and  
   were later imitated by Steve Jobs
   and co to develop tha Mac-style   
   interfaces we're familiar with       
   now.                             
                                        

Some great historical tidbits:

The original multi-user NLS system
developed at SRI needed about a dozen
displays.  Raster scan displays -- much
like the one that's probably in front of
you now -- were out of the question: the
memory was far too expensive.  There were
some decent vector scan displays (the
descendants of oscilloscopes and radar
screens), but those were pretty expensive
too... so what they did was to buy only
five vector scan displays (only five
inches), and multiplex them with TV
cameras and television displays.  Luckilly
they were interested in collaborative
work: you and a co-worker would both quite
likely be looking at the *same* computer
display, shown on two different TV
screens.

The development of the pointing device:
they tried a number of alternatives,
including things like a "knee pointer"
that mounted under the desk.


One of the things that I particularly like
about Bardini's treatment is that he's
really trying to get at the ideological
sources of Englebart's thinking, and he
touches on disparate sources like Whorf,
Korzybski, and Bateson. There's also some
comparision of Englebart with some
alternate schools of thought (the
"artificial intelligence" guys; and Ted
Nelson gets some mentions also).

In fact I wish he'd gone into this a little
more deeply, and been a little more careful
about tracing references.  It's not
immediately clear if Englebart claims a debt
to Whorf/Korzybski, or if that's just a
parallel that Bardini has noticed.  Did
Englebart know enough about Ted Nelson's
writings to have a chance of being
influenced by them? (Or vice versa?)  I
don't think Bardini spells this out --            ((check this))
though considering that Englebart was
hanging out with guys like Stewart Brand
at the time, it wouldn't be a huge
surprise if he'd read some of the
material that later ended up collected in
Nelson's "Computer Lib/Dream Machines".

But the downside of Bardini is that he
periodically gets lost in ungrounded
academic blather that I imagine would
sound really good to humanities majors
educated in the pomo eighties, but leaves
me almost entirely cold.  It often either
doesn't seem to mean very much, or is
pretty clearly just wrong.

The main idea that Bardini is tracking
here is that different visions of the
"user" produced different kinds of systems
(Englebart had expert "knowledge workers"
in mind; Kay was thinking of novice office
workers; the Apple guys has total newbie's 
in mind, hence the one button mouse). 

A simple enough notion, right?  But
Bardini also likes to point out that
there's a cycle in technical development
where developers recruit (or "create")
users, and then the users in turn
influence the developers.

In other words you try something, you see
how it goes over (and how well you can put
it over), and then you try something else.  

However, Simple language like this is not
Bardini's forte, and instead you need to
plow through a lot of pomo jargon that
appears to be the air that academic
intellectuals breath these days.

Here, I submit is the absolute low-point 
of this book:


   p. 178/179 - Ch 6, "The Arrival of the
   Real User and the Beginning of the End"

   The user is socially constructed and
   socially situated in the processes by
   which technology is developed and
   diffused, and then the user is
   progressively realized in a social
   setting.  

   The relations between designers and
   users are organized in the negotitions
   about the future uses of the
   technology, starting from the abstract
   or virtual representations of the user
   in the mind of the designer and
   progressively approaching
   confrontations with real users.  


   In this process, the kind of testing
   initiated at PARC [...]  was a
   fundamental, but nevertheless limited
   move.  This move, according to their
   claim, shifted the focus from a
   comparison between devices to a study
   of the human-device interaction.
   However, although this move
   demonstrated the interest of
   introducing the human aspect of the
   problem, it still fell short of fully
   realizing the user.

      Instead, it reduced the user to a
   subject, in the scientific, not the
   philosophical sense -- an object of
   study in which most of the qualities of
   the human being were deemed not to be
   of interest and bracketed out by the
   experimenters.  The introduction of
   cognitive science as a way to introduce
   real users into the processes of
   technological development thus also
   limited conceptions of what real users
   might be and might do with the
   technology being developed. 

   [...] This choice of novice subjects was
   the obvious limit of this process, not
   because these subjects lacked experience
   with the devices, but because they were
   constructed as novices and as nothing
   else.  Attention was paid to differences
   of sex, but their social identity was no
   concern of the experiment [...]
   
   These subjects were only "half-real" or,
   in other words, they were model subjects,
   subjects who could be read as embodiments
   of the generic eye-hand system.  The
   experiments were able to prove that using
   the mouse was as efficient as pointing,
   but people who point usually don't develop
   repetitive-stress injuries (RSIs). The
   limitations of the definition of the user
   imposed by the cognitive-science
   conception of the user as an experimental
   subject thus laid ample groundwork for
   subsequent unintended consequences of the
   technology.  


If you're the kind of person who's
impressed by this stuff, this is probably
all very impressive, but what actually
strikes me is that it totally misses the
point.  The fact that they didn't notice RSI
problems in the handful of Stanford
kids they tested is because that they
didn't work them for ten hours a day for
ten years.  It had nothing to do with any
sort of failure to come to grips with
their humanity.  And how exactly you're
supposed to do that in a preliminary
useability study remains a complete
mystery to me ("So tell me, how do you
*feel* about that pointing device?  Does
it remind you of your mother?").

Keep in mind that they were already doing
things like video taping the users faces
to get an idea of what their reactions
were.  They didn't just look at
statistical summaries of reaction times, 
they sat down and watched people in action
using the prototypes.  

And further, the person who designed the
study under discussion actually was a fan
of cursor keys... the fact that the users
seemed to prefer the mouse came as a
surprise.  Claiming that there was no real
contact between experimenters and subjects
here seems very peculiar... in fact this
experimental result is the *one* place I
can think of where Bardini documents the
iterative development process he likes to 
talk about. 

(Note: the above quote is also the one and
only mention of RSI in the entire book.)


While the low points of this book are low,
the high points are pretty high. It has
many saving graces.  First, as he mentions
in the preface, he does not think that his
ideas are more interesting than those of
the people he's writing about, so much of
the book can be read as "historical
realism".  Second, as I mentioned above, I
think he casts his intellectual net wider
that a more computer-oriented person
would, and gets at some of the deeper
sources.  And third, what I would take to
be his main thesis actually makes a good
deal of sense once it's striped of the
compulsive pomosity:

Decisions were made about the standard
computer user interface based on some
assumptions about the user that deserve to
be reviewed. There is no reason to assume
that the the course we've taken is the
right course, and perhaps the paths not
taken can still be taken, and should be
taken.

The point being that experts need
interfaces too, and maybe you're doing            YOU_KNOW_TOO_MUCH
them a disservice if you force them to use        
the same simplified controls that might           
be appropriate for a beginner or                  
infrequent user. 
   
        But then this is a notion that
        you're starting to hear from a
        number of sources -- Jakob
        Nielsen uses the example of
        slashdot: possibly too
        complicated for a totally
        non-technical audience, but it
        works well enough for the slash
        crowd....
                              
What I actually would recommend this
book for is it's historical review
(e.g.  Englebart and his group at SRI
got lost in "est" in the early 70s!)

--------
[NEXT - FURTHER_REFLEXION]