Joe's Jottings

Jottings Number 55, by Joe Podolsky:

From: uunet!HP-PaloAlto-om4.om.hp.com!JOE_PODOLSKY

Date: Tue, 14 May 96 17:30:45 -0700

Subject: Satisfying Systems, or, Film Versus Software

What we call things matters.  We humans are programmed to
respond to labels, often ignoring substance.  For example, Nass
and Reeves (see Jottings #52) found that people related more
positively even to standard computers that were designated (even
self-proclaimed) as specialists or as teammates.

What about labels we use in information technology?  IT never
claimed to be a science; we don't at all try to discover things.
IT is fundamentally based on software, and software creation
is often labeled "engineering"; we do, after all, build things. 
But engineering is based on well-proven theories and practice. 
Engineering is a _discipline_, but most software is not
disciplined.  Algorithmic constructs and formal methods,
notwithstanding, most software is, at best, crafted.

Craft implies art that is created following a structured
process.  Organizations like the Software Engineering Institute
and gurus like Tom Yourdon, Capers Jones and Barry Boehm have
long preached the benefits of software process discipline, but
even their methods have not solved our perennial and
continuously painful problem.  It may be a bad rap, but our
customers still complain that we don't meet their needs and,
besides, we're too late and too expensive.

I used to think that the answer was simply to "become more
disciplined."  That's what the Japanese tried to do.  Their
culture is much more amenable to structure, and they spent loads
of money and energy to build software factories, to certify
their programmers, to do software "right."  This approach, after
all, worked with VCRs, cameras, and TVs, industries which they
captured through their focus and disciplined, low-cost
manufacturing processes.

But the Japanese failed with software.  The counterfeiters in
China steal U.S. software, not Japanese. We in the West may be
process slobs, but we seem to be building what the marketplace
wants.

So, we seem to be left with software as an art.  Fine art,
painting, sculpture, etc. seems a little extreme as a model,
since it implies the self-expression of the artist with little
regard to customer wishes.  But we might look at commercial art
for some metaphoric lessons.

The two most visible forms of commercial art are films and
advertising.  Analogies to software creation are abundant. 
Films, advertising, and software all are strongly based in
various physical technologies but also depend a lot on sociology
and psychology for their success.  All are judged ultimately by
their success in the marketplace.  All require the participation
of teams of people.

But there are some differences.  First, and foremost for this
discussion, films and advertising acknowledge their existence as
an art form.  Each film and advertising project has a single
creative director.  No such role is explicitly designated in
most software projects, although each project may _de facto_
have such a person.

Secondly, software has one attribute not present in films and
advertising.  Software has a life.  Software, and especially
information systems, require on-going support.  Furthermore, the
software becomes entwined with its environment and adapts and
evolves, so that the identical code placed in different settings
soon takes on completely different operating characteristics. 
Software is more like an old Volkswagen Beetle or a DC-3
airplane...or a well-worn shoe.

There is, however, one area in IT and in software technology has
long recognized its root in art.  That is the "graphic unit
interface," the tool which both links us to and masks from us
the true complexity of the computer.   _Interactions_, the ACM's
journal on human computer interfaces, decided to confer Design
Awards to the computer-based products that produce the best
"quality of experience" for the users. 

The June 1996 issue of _Interactions_ reported the results of
the first years entries.  The awards jury decided, given the
immaturity of their process, to honor six finalists rather than
selecting a single winner.  The six were: _Apple Guide_ from
Apple Computer; _Graphing Calculator_ from Apple Computer;
_Meeting Manager_ from the City of Edmonton; _The Muse_, a
digital music stand from a group based at Carnegie Mellon
University; _Nokia Feature Stereo TV, submitted by IDEO London;
and NYSE Hand-Held Terminal_ submitted by Mauro/Mauro Design.

What was most interesting to me, however, were the selection
criteria that the jury evolved.  They wanted criteria that would
help them "pin down that feeling of satisfaction you get when
you encounter a tool or toy that clicks with you on a deep
level."  The jury decided that "to create a great user
experience, a design must be all these things:

"_Needed_:  Is the product needed?  (Here is where we can
include financial/market success.)

"_Grounded in understanding_:  Did the designers understand the
needs and characteristics of the people who would be using the
product?

"_ Learnable and usable_: Is the product easy to learn and use? 
Are its features self-evident?

"_Appropriate_:  Does the design solve the problem at the right
level...injecting just the right balance of technology and
people into a situation?

"_Product of effective process_:  (Did the process involve)
collaboration between different disciplines, a deep involvement
of end users, and lots of little steps towards the goals?

"_Aesthetic experience_:  (Does the product evoke a) pleasing,
sensually satisfying experience ... (considering) information,
hardware, software ... the whole context?

"_Manageable_:  What is it like to use (the product) over longer
periods of time?  Does it fit well into the work culture?

_Mutable_:  Can people ... change the product to suit themselves
... (and) tailored for new uses?"

I think this is a great list.   Like any set of attributes, such
as those offered by Boehm many years ago and HP's own venerable
FURPS, these can be used as a checklist for product planning up
front and for evaluation at release.

But what's nice about this particular list is that it calls for
a visceral communication through the product between the
designer and the user, a linkage that evokes pleasure as well as
productivity.

HP's products often win design awards, so these ideas are not
foreign to our culture.  But we have a long way to go before our
business systems can join this competition.  But, building
"pleasing" systems seems like a super way of using all those
MIPS and all that bandwidth that we now have to play with.  Way
back in Jottings #24, we quoted PARC's John Seeley Brown who
suggested we want "beautiful" systems," but these criteria are
much more specific and useful.

Do think that we should use these criteria to design and
evaluate our systems?  If so, how shall we start?  Do you know
of any systems that can meet these criteria, either from
internal groups or from vendors?

Regards,


Joe

Back to Joe's Jottings