Review Essay: The Invisible Computer by Donald A.

Norman, 4(11)
John Mueller

mueller@ucalgary.ca
University of Calgary

Norman, D. A. The Invisible Computer: Why Good Products Can Fail, the Personal Computer Is So Complex,
and Information Appliances Are the Solution. (Book).
Cambridge, MA: MIT Press, 1998, 302 pages (cloth) $38.95 (Cdn., at Chapters.ca) ISBN 0-262-14065-9
(Also available in paperback, 1999, $21.95 (Cdn., at Chapters.ca)

Norman, Donald A. don@jnd.org

URL: http://www.jnd.org/, Donald Norman's home page

The Invisible Computer (1998), is subtitled "Why good products fail, the personal computer is so complex, and
information appliances are the solution," and the book proceeds to examine these three points. Norman has
discussed aspects of these questions in two previous books, notably The Design of Everyday Things (1990) and
Things That Make Us Smart (1993). The previous books need not be read to appreciate The Invisible Computer,
but they do reinforce the depth of the argument. The The Invisible Computer is somewhat different from the first two
books, in that it follows Norman's involvement in the private sector, at Hewlett-Packard and as a Fellow and VP of
Research at Apple Computer, and thus draws some insights from the non-academic, industry perspective as well as
from academic psychological principles. As a result, the book is more grounded in the present than those in the more
general cyberculture genre (Silver, 1998; URL: http://otal.umd.edu/~rccs/biblio.html). Others have taken the
computer industry to task, but usually just by focusing on why profits do not reliably increase following
computerization (e.g., Gibbs, 1997) -- in fact, these often draw upon Norman's trilogy, implicitly or explicitly. And,
thankfully, this is not yet another book on how to do business on the Internet.

The preface makes it clear that Norman believes that problems with the personal computer are yet another instance
of a more general phenomenon: problems associated with the transition of a product (and industry) from being
technology-centered to being human-centered, that is, to a product that is usable by the average consumer rather
than an object to be revered and mastered for its own sake; a means to an end rather than a goal. This perspective
provides far more than yet another bow at the altar of "user friendliness:" "We have even been told that "being digital"
is a virtue. But it isn't: People are analog, not digital; biological, not mechanical. It is time for a human-centered
technology, a humane technology" (Preface, viii).

Originally to have been titled Taming Technology, this book embraces a simple but profound philosophy: "Tools
should be noticed only when they break" (p. 243 and elsewhere). This point is illustrated by examining several
previous innovations, not just computers. For example, it is amusing (and a little frightening) to realize that Edison's
interest in developing the phonograph was the paperless office -- how many devices have failed to yield that result
(including e-mail)! But I suppose we should keep the faith, as there must have been a similar transition period earlier,
lost in the mists of antiquity, when there were repeated promises of the stone-tablet-less office.

A common thread among new devices is that in the salad days of a new technology, the early adopters are those
who actually enjoy the challenge of dealing with the breakdowns, but the potential buyers who would rather not be
bothered always outnumber the early adopters. However, for other prospective users, Norman's philosophy applies:
"Difficult tasks will always have to be taught. The trick is to ensure that the technology is not part of the difficulty" (p.
182). Humans are very adaptable, but at some point it is more appropriate to design the product to minimize the
training need. As it now stands, users have had to adapt for over 50 years as computers have evolved to suit
engineers and early adopters. So much for the rapid rate of technological innovation.

Norman's book is not so much about computers per se, as it is about this diffusion of innovation throughout society,
and the social and economic forces that expedite or impede the diffusion. The technology life-cycle has been
discussed by others, but usually in terms of how to impose the values of the early adopters on the masses (e.g.,
Moore, 1995, Rogers, 1995). Historically a "blame and train" orientation (p. 168) has characterized the computer
industry. Early adopters find the product adequate, so all the industry needs to do is coax the reluctant masses to buy
in and accept the responsibility for learning to live with expensive, inadequate solutions. Marketing is seen as the way
to reach the masses, not assessing user needs and redesigning the product -- build it, then convince them they need
it. From this perspective, the computer industry controls innovation, with user-friendliness defined to satisfy the early
adopter, at best.

Norman reviews the arguments that have been used to justify this strategy, but then challenges them. An alternative
view emerges here, the user as victim, rather than the problem. He identifies the real solution as designing the
products to meet the needs of the masses in the first place; used-centered design, rather than educating the masses
to tolerate the inconveniences that excite early adopters. In other words, don't try to just market immature technology
to the masses, but instead change the product to meet the actual needs of the masses. Norman's user-centered
design says that the lack of acceptance is because the technology doesn't address a user's needs or does so in an
intrusive fashion. So, instead of the computer industry just designing cool things to suit engineers, and then brain-
washing and blaming the end user, the end-user's needs must enter the design process very early. In truth,
considering the haphazard historical development of the personal computer (e.g., Cringely, 1996), it should not be
surprising that user-centered design has only belatedly become a concern of the industry.

The three components of the book's subtitle form an outline for Norman's user-centered perspective.

Why do good products fail?

From one perspective this is about why industries do not adopt new and superior technologies (e.g., Christensen,
1997; previous IEJLL review by Angus, 1999). Edison turned down the radio. Kodak turned down the Polaroid, and
then later the Xerox machine. The phonograph industry ignored radio. Western Union saw no use for the telephone.
In the computer industry, perhaps the best example is the difficulty that Xerox had making the transition from a
photocopier company to a computer company (Hiltzik, 1999). Presently we are watching recording companies play
the ostrich with regard to electronic distribution of music. And why this computer industry preoccupation with
delivering television to my computer screen? Changing from what has worked is a difficult step for a business,
perhaps the economic equivalent of what gestalt psychology labeled "functional fixedness" -- we're making money,
why innovate? Once a technology becomes entrenched in an industry, it is difficult to change; it creates infrastructure
problems and a business model problem.

The other perspective on this, however, is why the mass market doesn't accept a new technology when industry is
promoting it. Norman uses computer and non-computer examples to illustrate that cutting edge technology alone is
not enough for mass acceptance. The Beta videotape format was technically better than VHS. Edison's cylindrical
phonograph lost out to the flat disk. It is not enough to "design it, they will come," but rather "what is it they need, then
we design a solution." It is not just a matter of being patient while the engineers improve the technology (more RAM,
larger hard drive, faster processors and connections, and so forth). User-centered success instead requires matching
the tool to the human activity to be done in the first place, and this applies to both the hardware and the software. "It
isn't enough to be first. It isn't enough to be best. It isn't even enough to be right. ... it is essential to understand why
customers buy products ..." (p. 47). And that understanding does not mean how to market to reluctant customers, but
how to solve actual customer problems -- user-centered design.

The problem for Norman is not that the user has to be blamed and then trained, but instead that the user needs to
educate or inform the design process from the beginning. This derives from the fact that changing technology is
actually easier than changing social, cultural, and organizational practices -- which is why we still have the paper trail
instead of the paperless office. Psychologists have made similar arguments before (I am reminded of the work of
Chapanis, 1982, and Nickerson, 1982), though not with the visibility of Norman's trilogy. Nonetheless, the computer
industry has generally paid more attention to engineers and thus we still had the technology-centered product stream
and the self-serving dependence on human adaptability.

Occasionally the end user gets enough satisfaction from a lesser technology to forego the problems associated with
the superior technology. This serves to illustrate that the heart of the mass user is accessed by technology that is
"good enough" to satisfy an actual need, rather than superior technology that is not suited to actual user needs. That
a product with fewer features would be more useful is blasphemous to the technology enthusiast. Very seldom does a
product emerge with enough novelty and value that users will ignore primitive technology to pursue it; the VisiCalc
spreadsheet was such an exception, but it is hard to think of many others.

Why is the personal computer so complex?

One part of the answer, according to Norman, is that the personal computer tries to be all things to all people, and at
some point ends up doing nothing very well, just as a Swiss Army knife is hardly a match for a toolkit of specialized
tools each designed to do a specific task (p. 71). Obviously this position can be applied to integrated "Office"
products, and do-it-all web browsers: at some point they require more time to master and then use than the benefits
they provide to the typical user. Rampant featurism has been an important benefit for marketing and for product
reviewers, but has little to do with providing the right features for a given activity in a transparent manner. Products
are defined and marketed by a "bigger/more is better" model, whereby if enough features are provided a user can
surely find something they need. Sure, but in the manner of a needle in a haystack; busy users may not find this
attractive at all. In fact, it isn't just the computer industry that makes this mistake, but modern telephones vs. older
models, for example (p. 165). Sadly, many of the features requested in the next upgrade often exist in the present
version but users didn't know they were there (Head, 1998)!

Another part of the answer, according to Norman, is that the graphical user interface does not scale to the present
level of complexity (p. 74). One can have more icons on the desktop or menu bar than there are buttons in the space
shuttle cockpit! Yet now each one has its popup cue to indicate what it does, so much for transparency!

And complexity is important to the present business model in the industry. For one thing, the industry's revenue
stream presently derives from the need for regular "upgrades," or the subscription model. A complex, buggy product
is replaced with a more complex, buggier product, with the assurance that the upgrade will solve the problems in the
former. This has at least assured a generation of consultants and trainers that there will be steady work. The
computer industry did not invent planned obsolescence, but it has turned it into a fine art. The transition to an
alternative route to revenue, from selling a product once (or infrequently, e.g., a camera) and then regularly selling
services (e.g., developing) or supplies (e.g., film) has been very slow to emerge in the computer industry.

Another strategy in the computer industry is to avoid user feedback by differentiating the "user" from the "customer"
or buyer (e.g., p. 45). For example, in the workplace, the buyer is often not the end user, but a procurement officer
who buys for the entire organization (a person who often is neither a user nor an early adopter, rather a bean
counter). What this means is that the industry may sell hundreds or thousands of copies with one contact. Bulk
purchases are very efficient, however, this also means that feedback to the industry comes from that one person,
even though hundreds or thousands of users are involved. The end user is not in a position to vote with their
pocketbook and feet by choosing a different product, so the market forces that might otherwise drive inappropriately
complex or useless products out of the market are blunted by separating the user from the buyer. When individual
users can choose alternatives, bad products do disappear: remember Bob, and PC Junior? Push technology? Barbie
computer?

To some extent, this purchasing strategy may in part be a legacy of the mainframe era, where a company bought a
single machine to be stored in a vault distant from many of its users, and only the computer priesthood (early
adopters) provided feedback. Furthermore, this isolation of the end users from the industry also applies to the
integration of computers into schools as well as business offices, where bulk purchases are made. Kids complain,
teachers blame and train; teachers complain, school boards blame teachers, but the industry is sheltered from
feedback. No wonder it is so hard to show productivity benefits from computer investments (e.g., Landauer, 1995),
as bulk purchases do not include user feedback regarding problems, so engineering continues to be the loudest voice
for development.

Norman also considers and rejects many of the stock answers for dealing with complexity. For example, we have
often been assured that speech recognition will solve all these problems. However, Norman points out that "speech
recognition" is not "language understanding," much less "mind reading" by the computer, and that the latter is really
what's needed to reduce human-computer misunderstandings (p. 97-100)! As a solution to complexity, speech
recognition seems to share many features with the paperless office -- I don't expect either to happen in my lifetime.

Even common business strategies are implicated in answering why we continue to get complexity rather than
simplicity. For example, instead of focus groups with users of your product to see what they (early adopters) want in
the next "upgrade," the computer business should talk to non-users to see what it would take for them to buy in (p.
245). In this context, as an update, perhaps one might note the apparently large percentage of iMac buyers who are
first-time computer owners? There is an entire chapter (10) that discusses how organizational structure can and must
be changed to accomodate user-centered design -- including an endorsement of the way the US Navy operates
effectively using an informal chain of command that is very different than the formal hierarchy (Chapter 7). Best
procedures dictate outcomes, not methods. Quality means letting much of the control go to employees. Technology
permits both, but management practices still resist such devolution.

Why information appliances are the solution

"In the appliance model of computing, ... Learning how to use it is indistinguishable from learning the task"
(p. 57).

In cars, telephones, microwaves, and so forth, the computer is embedded, hidden, invisible. Norman sees more
special-purpose devices, sharing a common infrastructure, as the solution to problems created by a progressively
more complex single device. By this view, digital convergence will mean multiple, simple devices sharing a standard
infrastructure, the network, not a single device that does everything (in a difficult, inelegant manner). For example, we
standardize on TCP/IP, not the material in the wire carrying it, etc. I'm sure that we will see more special-purpose
networked devices, but I fear that it will be a challenge for the industry to leave them alone, to let them remain
special-purpose.

Norman distinguishes his appliances from the trend toward hand-held computers (e.g., Palm) that merely represent
efforts to "miniaturize the all-purpose device" (p. 57). ... The device should be built for the job, and no more difficult
than it needs to be ... one device per activity (p. 182). I think here of the remarkable early involvement of Federal
Express (UPS, etc.) in using a simple scanner to build searchable databases via the Internet, and other uses of the
bar code scanner. A simple device, to communicate with another device elsewhere, transparently -- not a do-all
device that takes a couple of bulky help manuals and a 24/7 support line.

I am reminded here of a cartoon I saw recently, where a new computer user was calling Information on her
telephone, and asking to get someone's e-mail address. I laughed at the time, but now, well, why is that funny? After
all, the phone company also provides Internet access, aren't the phone-number and e-mail databases linkable?
Novices expect it. Maybe they know something the industry should attend to. This seems a failure to attend to what
users are actually doing, what they would actually benefit from. Or perhaps the telco is waiting to find a way to make
money from it, like charges for calling to ask for a phone number?

In a similar incident, local cell phone companies have recently been promoting wireless e-mail access via your cell
phone. No, I don't know why, I don't know anyone who has asked for it, and I don't see the point of yet another device
(viz. Palm) with no real keyboard or screen. But in one commercial re the virtues of going wireless, part of the pitch
was that you would in addition be able to listen to the radio through your cell phone, yet another feature of this
wonderful new wireless world. I was listening on my car radio at the time, and laughed out loud when I realized that I
couldn't remember a time when radio hadn't been wireless! This complication of the cell phone seems another case
of the strategy of trying to make a device do more than is really optimal. No matter how cool it seems to early
adopters, it becomes like the VCR for most of us -- lots of unused, unwanted "features." Yes, I DO have a cell phone,
but just for outgoing/emergency calls from my car. It gets used very little, thankfully, but it is somewhat comforting to
have, like low-tech jumper cables. I confess that I never owned a wristwatch with a built-in calculator either, though I
have owned chronographs with so many functions I always had to look up which buttons do daylight-savings time and
which ones do stopwatch functions, etc.

Finally, speaking of early adopters, don't you just love it at the end of a meeting, when it's time to schedule the next
meeting, to watch the paper daytimer crowd (myself included) flip open quickly and then wait for the Palm crew to
fiddle with their schedule? Norman speaks hopefully of the Palm, from the perspective of a couple of years ago now.
PDAs may yet satisfy some need, but personally I don't see the niche yet, and in fact the PDA industry segment now
seems to be going in the same direction as cell phones, trying to be a miniaturized full-purpose device to entertain the
techno enthusiasts. I say, wake up folks, the problems with the personal computer are not that it is physically too
large!

As the book was written, circa 1997, Apple Computer, which is known for its efforts to make computers easier to use,
was not winning market share (to put it gently). Norman discusses the failure of Apple at several points, from the
perspective of a former insider, attributing Apple's problems to, among other things, pursuit of short-term gains at the
expense of long-term viability (p. 225). At this point, Apple has staged a stunning recovery, and users choosing
devices by non-technological features such as design and color (p. 36) sounds rather like the iMac world to me. Have
we crossed a hurdle with computers becoming more like information appliances, or at least more consumer-oriented
devices? Time will tell.

Likewise, the confrontation between Microsoft and the USA Department of Justice has occurred since the book was
written. Microsoft may not have invented the upgrade model, but they certainly perfected it, along with sales to
purchasing agents rather than individual end users, the absurdly long feature-list marketing scheme, and other
practices that blame the user and move user-centered design to the bottom of the list. Bill Gates' bank account may
be the one thing to give pause to all the sensibility of Norman's arguments about the importance of simple, well-
designed information appliances. Perhaps, as James Thurber said, "You can fool too many of the people too much of
the time"? Or perhaps a transition to a consumer-driven market will eventually do what Janet Reno and the U.S.
Department of Justice dreams of doing. Time will tell.

In sum, taming this technology will require: (1) a shift to user-centered design, (2) a shift to simpler information
appliances that can communicate seamlessly, and (3) a shift in the business model of the computer industry from the
subscription (upgrade) model to providing consumables and services as the basis for profits (e.g., selling film instead
of cameras). This implies that the economics of information appliances will be driven by the mass consumption
market, rather than sales to early adopters.

Norman is a Professor Emeritus of Cognitive Science at the University of San Diego, with a long history of research
and publication in human cognition. He is involved in other enterprises that attempt to deal with taming technology.
He is a partner with Jakob Nielsen in the Neilsen-Norman Group, a consulting firm that works with companies on
issues such as human-centered web-site design and other user advocacy issues. He also is President of a
distributed-learning company, UNext.com, using the technology of the Internet as an enabling tool, without
dispensing with books and other media.

Norman has adopted jnd.org as his domain name. JND is an acronym for "just noticeable difference," a
psychophysical concept describing how much a situation has to change before we notice it has changed. We can all
hope that his efforts in this trilogy take us at least a JND toward the invisible computer. Norman does offer solutions
here, as he has for over a decade, not just complaints, and we can only hope that the computer industry takes heed -
- better late than never.

If tools should be noticed only when they break, could we say a good book is noticed only when you finish it? If so,
this one fits that bill for me. Geeks may complain that it isn't enough about computers, but that is precisely the point
for most of us who just want a transparent tool. Obviously I am a fan of most of his arguments, and while this review
is somewhat an homage, nonetheless this is a thoughtful, readable book (with a clever dustcover).

References

Chapanis, A. (1982). Computers and the common man. In Kasschau, R. A., Lachman, R., & Laughery, K. R. (1982).
Information technology and psychology. New York: Praeger.

Christensen, C. (1997). The innovator's dilemma: When new technologies cause great firms to fail. Cambridge,
MA: Harvard Business School Press.

Cringely, R. X. (1996). Accidental empires. New York: Harper Business.

Gibbs, W. W. (1997). Taking computers to task, Scientific American, July, 82-89. URL:
http://www.sciam.com/0797issue/0797trends.html

Head, A. J. (1998). Are Microsoft's animated interface agents helpful? Online Magazine, January.

Hiltzik, M. A. (1999). Dealers of lightning: Xerox PARC and the dawn of the computer age. New York: Harper
Business.

Landauer, T. K. (1995). The trouble with computers. Cambridge, MA: MIT Press.

Moore, G. (1991). Crossing the chasm: Marketing and selling high-tech products to mainstream customers.
New York: Harper.

Nickerson, R.S.. (1982). Information technology and psychology. In Kasschau, R. A., Lachman, R., & Laughery, K. R.
(1982). Information technology and psychology. New York: Praeger.
Norman, D. (1990). The design of everyday things. New York: Doubleday. URL: Amazon.com entry, Chapters.ca
entry

Norman, D. (1993). Things that make us smart. Reading, MA: Addison-Wesley. URL: Amazon.com entry,
Chapters.ca entry

Norman, D. (1998). The invisible computer: : Why good products can fail, the personal computer is so
complex and information appliances are the solution. Cambridge, MA: MIT Press. URL: Amazon.com entry,
Chapters.ca entry

Rogers, E. M. (1995). Diffusion of innovations. (4th ed.) New York: Free Press.

Silver, D. (1998). Cyberculture: An annotated bibliography. [On-line]. Available:

Author Note

John Mueller is a Professor in the Educational Psychology Department at the University of Calgary. He specializes
in cognitive psychology, affect and learning, and computers and learning. He has taught various courses in
educational computing, specializing in accessing and using the global Internet as an educational resource. He can be
contacted through e-mail at mueller@ucalgary.ca. His web site address is http://mueller.educ.ucalgary.ca/