¿ìè¶ÌÊÓƵ

To what does Alan Kay, one of computing’s great innovators, owe his creative
vision? He dreamt up the portable computer in 1968, when computers were
huge, shared mainframes used only by experts. Now, a decade after personal
computers began to revolutionise the way we work, he is devising ways to
bring complex programming for parallel processing within reach of children.

Some of Kay’s peers put it down to his ability to combine common sense and
computing; others, to a Renaissance view of the world. Kay’s father was a
scientist and his mother an artist and musician: ‘I grew up in both
traditions and I have never seen a huge difference between them.’ He has
degrees in mathematics and molecular biology, but has also been a portrait
painter and a professional jazz guitarist.

Now 53, Kay is still trying to realise his vision of 25 years ago, when he
was a postgraduate at the University of Utah. At the time, he described the
portable computer of the future as having wireless communications and being
easy enough for a child to use. He compared his notion of an interactive,
multipurpose portable computer to a dynamic reference book, and called it
Dynabook. He even built a model in cardboard, which he filled with small
weights to check that it would be comfortable to carry around. ‘The
computer’s weight is very much part of the user interface,’ he says.

Kay’s computing skills, plus such attention to everyday detail, impressed
Xerox Corporation. So, two years after first describing the Dynabook, and
following a spell on an artificial intelligence project at Stanford
University, Kay joined Xerox and helped to found its Palo Alto Research
Center. Within four years, a team at PARC had created a prototype personal
computer with an interactive graphical user interface. This used icons,
menus and windows on the computer’s screen to present information, which
could be manipulated using a hand-held pointing device called a mouse.

But, in 1974, such innovations were too revolutionary even for Xerox,
recalls Kay. They so confounded the prevailing notion of a computer as a
large centralised resource that the company decided not to develop the
prototype: it was left for other computer manufacturers to exploit almost a
decade later. Nevertheless, Kay stayed with Xerox until 1981 before moving
to Atari, the home computer maker, for three years as chief scientist. Since
1984 he has been an Apple Fellow, one of a select band of scientists in the
company who is free to pursue their own research interests.

In developing technology for the future, Kay has looked for inspiration to
the past. He acknowledges the influence of a range of computer scientists,
psychologists, educationalists and philosphers. Like many great innovators,
he has synthesised the work of his predecessors to produce ideas that shift
the perspective and open up new possibilities.

Many of the components of the PARC personal computer had been invented in
the previous decade. Douglas Engelbart, a computer scientist at the Stanford
Research Institute in California, invented the mouse in the early 1960s.
Around the same time, Ivan Sutherland, a researcher at the US government’s
Information Processing Techniques Office, created the first interactive
graphics program. While Kay was at Utah, he built a desktop personal
computer called Flex. Although the machine had graphics and multiple
windows it proved unsatisfactory, according to Kay, because the user
interface ‘repelled end users rather than drawing them closer to the
³ó±ð²¹°ù³Ù³ó’.

Then, in the late 1960s, Kay encountered a number of ideas and innovations
that were to shape the course of his thinking. Most computers then, as now,
used bulky cathode-ray tube displays, the technology also used in
televisions. But a team at the University of Illinois pioneered liquid
crystal technology to produce the first flat-panel display. This sparked
Kay’s dream of the Dynabook, in which the electronics of the Flex were
shrunk and attached to a book-sized flat screen.

But Kay did not envision the Dynabook as simply an electronic gadget, like a
calculator. He had read the works of Canadian theorist Marshall McLuhan, who
believed that a medium, such as television, does more than simply deliver a
message; it influences the message and affects the way people think. Kay was
convinced that the computer could do the same if only it were accessible to
all, including children. At that time, the programming languages and
operational screen messages were too obscure for all but specialists. A new,
easily learnt language for ‘reading’ and ‘writing’ on the computer was
required.

Shortly after reading McLuhan, Kay visited Seymour Papert, a researcher who
had helped to develop a programming language for children, called Logo. The
language was designed not so much to teach children how to program, rather
to teach them to think. Above all, it demonstrated that if a computer
language was designed with an understanding of the the user in mind, then
even primary schoolchildren could do quite complicated things with it, such
as write programs to translate foreign languages, produce poetry and drive a
‘turtle’ robot to draw geometric pictures.

In designing Logo, Papert incorporated the ideas of cognitive psychologist
Jean Piaget, who had observed that children go through a series of distinct
intellectual stages in their development. As children get older, what they
learn becomes more abstract. Early childhood, up to around 7 years of age,
Piaget characterised as the intuitive stage. From 7 to around 11, children
learn primarily by manipulating concrete objects. After 11 or so, abstract
reasoning develops. Piaget suggested that learning is most successful when
the material and method match the child’s particular developmental stage.
Papert reasoned that the process could be accelerated if abstract things,
such as geometry, could be made more concrete. Logo took advantage of young
children’s ability to manipulate an object – the turtle in this case – to
perform abstract geometry.

Kay explored further, seeking out the work of psychologist Jerome Bruner
who had developed Piaget’s ideas. Bruner suggested that instead of evolving
through a continuum of stages, we develop a set of ‘mentalities’, or ways of
reasoning, understanding and responding to the world, which exist in
parallel and can at times be in conflict. Bruner described three
mentalities: the ‘enactive’, which is to do with manipulating objects; the
‘iconic’, which is to do with understanding through observation; and the
‘symbolic’, which is to do with abstract reasoning.

From Papert’s work, Kay decided that any interface design must provide a
learning environment to be successful; from Bruner’s, he concluded that the
interface must also take into account the mentalities of the user. As a
guiding design principle, Kay created the slogan: ‘Doing with images makes
symbols’. By this he meant that grounding the user in physical activity,
linked to a visual environment where the human powers of recognition and
interpretation could be harnessed, would lead naturally into abstract
reasoning about how to solve problems on the computer. In practice, it led
to the adoption of the mouse, which was linked to a computer screen
displaying multiple overlapping windows and icons, through which the
computer and its programs could be understood and controlled.

Making computers child’s play

Windows offered a way of showing many resources on the screen at once. They
also avoided one of the problems associated with the syntax of traditional
computer systems. This syntax required the user to shut down one program, or
mode, before using another, which was time consuming and distracting. In a
windows-based system, a new program could be called up without having to
shut down any others; by placing the window of the active program seemingly
‘on top’ of any others it would always be obvious which was the current
activity.

Kay also drew on the ideas of Maria Montessori, the Italian educationalist,
and Tim Gallwey, the American tennis coach and author of The Inner Game of
Tennis, to create an environment on the computer screen that users could
explore for themselves. Montessori believed in providing an interesting
environment in which children learnt through activities they chose for
themselves. Gallwey suggested learning could be more effective if
irrelevant mentalities were distracted while relevant ones were brought to
the fore. Kay built these ideas into the visual elements of the computer
interface so that the way the interface worked could be discovered through
exploration.

‘What you get with the visual is the ability to put up multiple alternatives
so that even if people don’t know what to do next there’s a whole bunch of
things they can try,’ says Kay. ‘So if they are trying to be creative they
generally don’t have the feeling of being stumped.’

Given the parallels Kay drew between the computer user interface and other
learning environments, plus his insistence on extending the new medium into
the realm of the child, it is perhaps not surprising that Kay chose to bring
in children to test the prototypes at PARC. Working with children brought
the problems of interface design into clearer focus, says Kay. ‘There’s an
idea that if you can’t make a system that’s understandable by adults you
just train them. We bent over backwards to try to do something that would be
obvious to the kids. We wanted the kids to be able to explore on their own,
Montessori-like. We wanted them to be able to examine everything. There are
a lot of things we did in the name of completeness and aesthetics that I
think we wouldn’t have done just for adults.’

Developing computer systems in which children could ‘make things and learn
by the making’ was a ‘romantic thing to pursue’, says Kay. ‘Romance’ is a
term that Kay uses repeatedly to explain the motivation for his work, a term
that appears to encompass fun, excitement, creativity, beauty and furthering
the ideals of civilisation. ‘You can go on working on an impossible project
for a long time if it has a lot of romance to it,’ says Kay. The Dynabook
was a romantic vision still not fully realised.

Today’s portable and hand-held computers do not yet have all the facilities
Kay envisioned, such as wireless communications. But it is a measure of the
clarity of his vision that as they begin to match the Dynabook’s original
specification, so the ‘object-oriented’ programming method that Kay also
devised is now seen by computer companies and software developers as the
best approach.

An interactive user interface with icons, menus and multiple windows, each
of which can represent an active process, is a complex, dynamic system.
Traditional computer languages like Cobol, Basic and C were not designed to
support this kind of system. These languages tell the computer how to solve
a particular problem, using operations on essentially ‘passive’ data,
carried out in strictly ordered procedures. These step-by-step procedures to
solve problems are also called ‘algorithms’. For Kay this algorithmic
approach was inadequate and inappropriate for the dynamic nonlinear system
he envisioned with the Dynabook.

Remembering the techniques of the biological world he had studied as an
undergraduate, Kay observed that evolution had long since solved the problem
of managing complex dynamic systems. Biological systems maintain control by
confining processes and their data in self-contained cells. These cells act
on each other by sending ‘messages’, carried by chemical messengers. Kay
used a direct analogy to biological cells in devising Smalltalk, the name he
gave to his object-oriented programming language. The computer system was
broken down into component parts. Their operations and their data were
confined, in cell-like units called ‘objects’. This in effect surrounded an
object with the equivalent of a membrane to protect its data from
inappropriate processes. The processes within objects were activated by the
receipt of a message from another object.

Smalltalk is more than a programming language. It was designed to act also
as the computer’s operating system – the software in the computer that
controls the machine’s resources and supervises the programs. Smalltalk was
conceived as inherently dynamic – every object is always ready to receive
and send messages, and many objects can be active simultaneously, or in
parallel. ‘The view that we (at PARC) had of objects was that they were
never going to be put away on the disc. Some part of Smalltalk was always
ticking away on the computer regardless of what you were doing. The objects
were never thought of as being static,’ says Kay.

Smalltalk, the graphical user interface and the personal computer were just
three of the inventions to come out of PARC in the 1970s. Out of the same
fertile laboratories during this decade came the laser printer, which would
become the most popular personal computer printer in the 1980s, and
Ethernet, a local area networking software that connects several hundred
computers via a single length of cable. The basic research at PARC for these
technologies cost around $60 million – and generated an industry
that is currently worth around $95 billion a year. Ironically, says
Kay, Xerox was unable to recognise the innovation taking place and did
nothing to exploit the 10-year lead it could have had in the personal
computer industry.

He attributes the extraordinary productivity of PARC largely to the fact
that almost all the researchers were experts in both hardware and software.
They were never diverted by problems caused by having to integrate other
people’s technology with theirs. ‘We built every bit of hardware and
software ourselves. The only stupidities we had to worry about were our
´Ç·É²Ô.’

Kay laments the schism that has appeared in computer science between
hardware and software, symbolised by what he calls the ‘outrageous’ teaching
of them as separate disciplines at universities. ‘At PARC we would create
the software architecture first and then we would build the hardware to
optimise whatever the software needed to do. That’s the exact opposite to
the way things are done today. But it’s hard to imagine any use for
hard-ware other than to run software so why put the hardware first?’

Kay and his fellow researchers at PARC also built on the work that had been
done in the 1960s under the auspices of the Advanced Research Projects
Agency of the US Department of Defense, which became the Defense Advanced
Research Projects Agency, or DARPA, a decade later. This was a golden era in
US computer science research, with ARPA distributing a legacy from the space
programme to a number of brilliant scientists who were given their head. But
as the Vietnam War escalated in the late 1960s, ARPA research became
confined to projects related to more immediate military needs. Kay decided
that the only place he would find the funding to pursue his ‘romantic’
projects was inside commercial companies. At Apple he has been working on
the next generation of the user interface.

Eyes on the next horizon

Kay has long been concerned about what he describes as the lack of symmetry
in the computer interface. By this he means the fact that a quite different
language is used to ‘read’ the computer from the languages that are used to
‘write’ to it. Reading gives access to materials and tools created by
others. Writing means generating materials and tools for oneself and others.
True computer literacy requires the ability to do both. The invention of the
graphical user interface has made it considerably easier for nonspecialists
to read the computer, but writing, or ‘authoring’, is still in the realm of
the professional. The nirvana of personal computing is, according to Kay,
‘when end users can change their tools and build new ones without having to
become professional-level programmers’.

But developing a symmetric interface appears to be an order of magnitude
more challenging than the creation of the user interface. ‘It took us only
four or five years to design (Apple’s) Mac-style interface,’ says Kay, ‘but
it has taken us 20 years to not quite do end-user authoring.’ A system has
been developed that is ‘the closest thing to a symmetric read/write system
that we have done’, says Kay. Apple, however, has so far released few
details of the system beyond the fact that it is object-oriented and
supports the development of dynamic systems. Kay describes it as ‘parallel
processing for kids’.

It is characteristic of Kay that while the symmetric authoring system
appears to be coming to fruition he has his eyes on the next horizon. The
next major development in the computer interface should create a
fundamental shift in the relationship between the user and the system, he
says. ‘The icon-based interfaces we developed in the 1970s were oriented
toward being able to easily teach people about the system. The new
interfaces will require the system to learn from the user.’

Kay calls this approach ‘agent-oriented programming’ where an agent is a
piece of software that carries out some task or pursues some goal on behalf
of the user. Agents become more important as the size and complexity of
electronic information systems increase. The idea of agents has been around
for some time and a number have already been created. Kay himself helped to
design an agent called NewsPeek more than 10 years ago. So, on past
experience, it’s about time the computer industry thought seriously about
exploiting the development.

Clive Davidson is a freelance journalist.