Fame, fortune, or even the more esoteric career ambitions of top-notch
software professionals do not seem to motivate Dr. Kay, now a "research
fellow" for Apple, formerly "chief scientist" at Atari Corporation.
Becoming another Silicon Valley
millionaire or accepting the offer of an endowed chair at MIT have not
interested him as much as the prospect of putting the power to
imagine into the hands of every bright kid who got thrown out of a
classroom.Ever since he learned to read at the age of two and
a half, Alan Kay has been accustomed to doing things his own way and
letting the rest of the world catch up later. At the same time he was
close to flunking out of the eighth grade, primarily for
insubordination, he was one of television's original "Quiz Kids." Ten
years before he coined the term "personal computer," before Atari or
PARC existed, and before another pair of bright insubordinates named
Wozniak and Jobs created a new meaning for that good old American word
"Apple," Alan Kay was demonstrating FLEX, a personal computer in all but
name, to the ARPA graduate students' conference.
Alan is now in his early forties, and is acknowledged by his peers,
if not yet the general public, as one of
the contemporary prophets of the personal computer revolution. Now his
goal is to build a "fantasy amplifier," a "dynamic tool for creative
thought" that is powerful enough, small enough, easy enough to use, and
inexpensive enough for every schoolkid in the world to have one. He has
the resources and the track record to make you believe he'll do it.
Alan Kay doesn't fit the popular image of the arrogant, antisocial
hacker, the fast-lane nouveau micromillionaire, or the ivory tower
computer scientist. He wears running shoes and corduroys. He has a
small, meticulous moustache and short, slightly tousled dark hair.
He's so imageless you could pass him in
the halls of the places he works and not notice him, even though he's
the boss. Which isn't to say that he's egoless or even modest. He loves
to quote himself, and often prefaces his homilies with phrases like
"Kay's number one law states . . . ."
When I first encountered him, between his stint as director of the
legendary "Learning Research Group" at Xerox PARC, and his present
position as a kind of "visionary at large" for Apple, Dr. Kay and his
handpicked team at Atari were working under tight secrecy, with a budget
that was rumored to be somewhere between $50 million and $100 million,
to produce something that nobody in the corporation ever described to
anybody outside the corporation. But anybody who has ever talked to him,
or read something he has written about his dreams, can guess the general
thrust of Kay's Atari project, and the probable direction of his current
work at Apple. He has been moving toward realizing his dream, project by
project, prototype by prototype, innovation by innovation, ever since he
was a graduate student.
Being the kind of person he is didn't make it easy for Alan to get an
education. At the beginning, he knew more than all of his classmates and
most of his teachers, and he didn't mind demonstrating it aloud -- a
trait that got him thrown out of classrooms and beaten up on
playgrounds.
Fortunately for him and for all of us who may benefit from his
creations in the future, Alan was already well armored in his mind and
imagination, where it really counted, by the time his teachers and
classmates got ahold of him. For Alan, being ahead of everybody else
started out as a pleasure and quickly turned into a survival trait --
which meant he didn't do too well in school, or anyplace else, until the
summer of his fifteenth year, when "a music camp in Oneonta, New York,
changed my entire life."
Music became the center of his life. In many ways, it still is.
He commutes to Silicon Valley from his
home in Brentwood, 300 miles away, mostly because he doesn't want to be
away from his homemade pipe organ for too long. And he still goes
to music camp every summer. He never understood why his two favorite
toys -- books and musical instruments -- could not be combined into a
single medium capable of dealing with both sounds and symbols. He worked
as a professional jazz and rock guitarist for ten years. When it looked
like he was about to be drafted, Kay joined the U.S. Air Force as a
navigational cadet. In the Air Force, he "wore out a pair of shoes doing
insubordination duty," but he also learned that he had a knack for
computer programming.
After he finished his Air Force duty, the National Center for
Atmospheric Research was eager to use Kay's programming talent to pay
his way through the University of Colorado. He earned a degree in
biology, but his college grades were as mixed as they had always been,
because of his habit of concentrating intently on only those things that
interested him. Through what Alan now calls "sheer luck,"
he came to the attention of somebody
smart enough to actually teach something to a smartass like Alan Kay --
and bold enough to admit a student with an undergraduate record that
read more like a rap sheet than a transcript.
The man who gambled on Kay's checkered history in academia was David
Evans, the chairman of the computer science department at the University
of Utah, a place that was to become one of the centers of the
augmentation community by the mid-1960s. Like so many others who assumed
positions of leadership in the field of interactive computer systems
design, Evans had been involved in early commercial computer research
and with the ARPA-funded groups that created time-sharing.
"Those career pathways of ARPA project leaders and their graduate
students repeatedly intertwined," Kay recalls. "An enormous amount of
work was done by a few people who kept reappearing in different places
over the years. People frequently migrated back and forth from one ARPA
project or another. ARPA funded people rather than projects, and they
didn't meddle for an extended period. Part of the genius of Licklider
and Bob Sproull was the way this moving around contributed to the growth
of a community."
One of the people Evans managed to recruit for the Utah department
who had an impact, not only on Alan Kay but on the entire course of
personal computing was
Ivan
Sutherland, the graduate student and prot�g� of
Claude Shannon
and J. C. R.
Licklider who single-handedly
created the field of computer graphics as a part of his MIT Ph.D. thesis
-- the now legendary program known as "Sketchpad."
People like Alan Kay still get excited when they talk about
Sketchpad: "Sketchpad had wonderful aspects, besides the fact that it
was the first real computer graphics program. It was not just a tool to
draw things. It was a program that obeyed laws that you wanted to be
held true. So to draw a square in Sketchpad, you drew a line with a
lightpen and said: 'Copy-copy-copy, attach-attach-attach. That angle is
90 degrees, and these four things are to be equal.' Sketchpad would go
zap! and you'd have a square."
Another computer prophet who saw the implications of Sketchpad and
other heretofore esoteric wonders of personal computing was an
irreverent, unorthodox, counterculture fellow by the name of
Ted Nelson, who
has long been in the habit of self-publishing quirky, cranky, amazingly
accurate commentaries on the future of computing. In The Home
Computer Revolution Nelson had this to say about Sutherland's
pioneering program, in a chapter entitled "The most important computer
Program Ever Written":
You could draw a picture on the screen with the
lightpen -- and then file the picture away in the computer's memory.
You could, indeed, save numerous pictures in this way.
For example, you could make a picture of a rabbit
and a picture of a rocket, and then put little rabbits all over a
large rocket. Or, little rockets all over a large rabbit.
The screen on which the picture appeared did not
necessarily show all the details; the important thing was that the
details were in the computer; when you magnified a picture
sufficiently, they would come into view.
You could magnify and shrink a picture to a
spectacular degree. You could fill a rocket picture with rabbit
pictures, then shrink that until all that was visible was a tiny
rocket; then you could make copies of that, and dot them all
over a large copy of the rabbit picture. So that when you expanded
the big rabbit till only a small part showed (so it would be the
size of a house, if the screen were large enough), then the
foot-long rockets on the screen would each have rabbits the size of
a dime.
Finally, if you changed the master picture --
say, by putting a third ear on the big rabbit -- all the copies
would change correspondingly.
Thus Sketchpad let you try things out before
deciding. Instead of making you position a line in one specific way,
it was set up to allow you to a number of different positions and
arrangements, with the ease of moving cut-outs around on a table.
It allowed room for human vagueness and judgment.
Instead of forcing the user to divide things into sharp categories,
or requiring the data to be precise from the beginning -- all those
stiff restrictions people say "the computer requires" -- it let you
slide things around to your heart's content. You could rearrange
till you got what you wanted, no matter for what reason you wanted
it.
There had been lightpens and graphical computer
screens before, used in the military, but Sketchpad was historic in
its simplicity -- a simplicity, it must be added, that had been
deliberately crafted by a cunning intellect -- and its lack of
involvement with any particular field. Indeed, it lacked any
complications normally tangled with what people actually do.
It was, in short, an innocent
program, showing how easy human work could be if a computer were set
up to be really helpful.
As described here, this may not seem very useful,
and that has been part of the problem. Sketchpad was a very
imaginative, novel program, in which Sutherland invented a lot of
new techniques; and it takes imaginative people to see its meaning.
Admittedly the rabbits and rockets are a
frivolous example, suited only to a science-fiction convention at
Easter. But many other applications are obvious: this would do so
much for blueprints, or electronic diagrams, or other areas where
large and precise drafting is needed. Not that drawings of rabbits,
or even drawings of transistors, mean the millennium; but that a new
way of working and seeing was possible.
The techniques of the computer screen are general
and applicable to everything -- but only if you can adapt
your mind to thinking in terms of computer screens.
Sutherland was Twenty-six when he succeeded Licklider as director of
ARPA's Information Processing Techniques Office. Then he was succeeded
by Bob Taylor when he left for Harvard in the mid-1960s, to work on 3-D
head-mounted displays (like miniature televisions in eyeglass frames)
and other exotic graphics systems. When David Evans tried to lure him to
Utah, Sutherland said he would do it if Evans agreed to become a
business partner -- and thus the pioneering computerized flight
simulation and image generation company of
Evans & Sutherland was born.Kay showed up at Utah in November of
1966. His first task was to read a pile of manuscript Evans gave him --
Ivan Sutherland's thesis. The way Evans ran the graduate program, you
weren't supposed to be around campus very long or very much. You were
supposed to be a professional and move on to high-level consulting jobs
in industry. The job he found for Alan Kay was with a hardware genius
named Ed Cheadle. Ed had an idea about doing a tabletop computer. Kay
worked on FLEX -- his first personal computer software design -- from
1967 to 1969. While some of the founders of today's personal computer
industry were still in high school, Kay was learning how to design
personal computers.
Technically, Cheadle and Kay were not the first to attempt to build a
personal computer. Wes Clark, from Whirlwind and Lincoln Lab's TX-2 and
"imps," had constructed a desk-size machine a few years before, known as
"LINC." FLEX was an attempt to use the more advanced electronic
components that had recently become available to bring more of the
computer's power out where the individual user could interact with it.
FLEX was a significant innovation technically, but it was complicated
and delicate, and in Kay's words, "users found it repellent to learn."
The problem wasn't in the machinery as
much as it was in the special language the user had to master in order
to command the power of the machine to accomplish useful tasks.
That was when Kay first vowed to make sure his personal computer would
come at least part of the way toward the person who was to use it, and
when he realized that software design would be the area in which this
desire could be fulfilled.
Although he didn't fully realize it
yet, Alan Kay was beginning to think about designing a new programming
language. The kind of language he began to yearn for would be a tool for
using the computer as a kind of universal simulator. The problem
was that programming languages were demonically esoteric.
"There are two ways to think about
building an instrument," Kay asserts. "You can build something like a
violin that only a few talented artists can play. Or you can make
something like a pencil that can be used quickly and easily for anything
from learning the alphabet to drawing to writing a computer program."
He was convinced that 99 percent of the problem to be solved in making a
truly usable personal computer program were software problems:
"By 1966, everyone knew where the
silicon was going."
Besides FLEX, Kay's other project at Utah was to make some software
work. He got a pile of tape canisters on his desk, along with a note
that the tapes were supposed to contain a scientific programming
language known as Algol 60, but they didn't work. It was a maddening
software puzzle that was still far from solved when Kay figured out that
it wasn't Algol 60 but a language from Norway, of all places, called
Simula. In a 1984 interview, Kay described what happened when he
finally printed out on paper the program listings stored in those
mysterious canisters and figured out what was on those tapes:
We couldn't understand any of the papers, they were
sort of transliterated from the Norwegian. . . . We spread out the
program listings and actually went through the machine code to try
to figure out what was happening -- and I suddenly realized that
Simula was a programming language to do what Sketchpad did. I had
never really understood what Sketchpad was. I get shivers now
thinking of it. It rotated my point of view through a different
dimension and nothing has been the same since. I suddenly understood
the purpose of higher level languages.
Alan was one of the enthralled audience at
Engelbart's 1968
media show. He was excited by it because it demonstrated what you
could really do with a computer augmented representation system. It also
made it clear to Alan what he didn't want to do. "The Engelbart
crew were all ace pilots of their NLS system," Kay remembers. They had
almost instant response -- like a very good video game. You could pilot
your way through immense fields of information. It was, unfortunately
for my purposes, something elegant and elaborate that these experts had
learned how to play. It was too complex
for my tastes, and I wasn't interested in the whole notion of literacy
as a kind of fluency.
In the course of preparing his Ph.D. thesis, Alan began to explore the
world of artificial intelligence research, which brought him into closer
contact with two more computer scientists who were to heavily influence
his own research --
Marvin Minsky
and
Seymour Papert, who were then codirectors of MIT's pioneering
artificial intelligence research project. In the late 1960s, Papert in
particular was doing something that irrevocable influenced Alan's goals.
Papert was creating a new computer
language. For children.Papert, a mathematician and one of the
early heroes of the myth-shrouded
Project Mac, had
spent five years in Switzerland, working with the developmental
psychologist Jean Piaget. Piaget had triggered his own revolution in
learning theory by spending time -- years and decades -- watching
children learn. He concluded that learning is not simply something
adults impose upon their offspring through teachers and classrooms, but
is a deep part of the way children are innately equipped to react to the
world, and that children construct their notions of how the world works,
from the material available to them, in definite stages.
Piaget was especially interested in how different kinds of
knowledge are acquired by children, and concluded that children are
scientists -- they perform experiments, formulate theories, and
test their theories with more experiments. To the rest of us, this
process is known as "playing," but to children it is a vital form of
research.
Papert recognized that the
responsiveness and representational capacity of computers might allow
children to conduct their research on a scale never possible in a
sandbox or on a blackboard.
LOGO,
the computer language developed by Papert, his colleague Wallace
Fuerzing, and others at MIT and at the consulting firm of Bolt, Bernack
& Newman, was created for a purpose that was shockingly different from
the purposes that had motivated the creation of previous computer
languages. FORTRAN made it easier for scientists to program computers.
COBOL made it easier for accountants to program computers. LISP, some
might say, made it easier for computers to program computers. LOGO,
however, was an effort to make it easier for children to program
computers.
Although its creators knew that the LOGO experiment could have
profound implications in artificial intelligence and computer science as
well as in education, the project was primarily intended to create a
tool for teaching thinking and problem-solving skills to children. The
intention was to empower rather than to suppress children's natural
desire to solve problems in ways they find fun and rewarding.
"The object is not for the computer to
program the student, but for the student to program the computer,"
was the way the LOGO group put it.
Beginning in 1968, children between the ages of eight and twelve were
introduced to programming through the use of attractive graphics and a
new approach that put the power to learn in the hands of the people who
were doing the learning. By learning how to use LOGO to have fun with
computers, students were automatically practicing skills that would
generalize to other parts of their lives.
Papert had observed from both his computer science and developmental
psychology experience that certain of these skills are "powerful ideas"
that can be used at any age, in any subject area, because they have to
do with knowing how to learn
. This is the key element that separated LOGO from the "computer
assisted instruction" projects that had preceded it. Instead of treating
education as a task of transferring knowledge from the teacher to the
student, the LOGO approach was to help students strengthen their ability
to discover knowledge on their own.
One of the most important of these skills, for example, is the idea
of "bugs" -- the word that programmers use to describe the small
mistakes that inevitably crop up in computer programs, and which must be
tracked down before the program will work.
Instead of launching students on an
ego-bruising search for the "right" answer, the task of learning LOGO
was meant to encourage children to solve problems by daring to try new
procedures, then debugging the procedures until they work.
The first revolutionary learning instrument introduced in LOGO was
the "turtle," a device that is part machine and part metaphor. The
original LOGO turtle was a small robot, controlled by the computer and
programmed by the child, that could be instructed to move around,
pulling a pen as it moved, drawing intriguing patterns on paper in the
process. Alan Kay was one of several software designers who realized
that this process was more than just practice at drawing pictures, for
the ability to manipulate symbols -- whether the symbols are turtle
drawings, words, or mathematical equations -- is central to every medium
used to augment human thinking.
The abstract turtle of today's more advanced display technology is a
triangular graphic figure that leaves a video trail behind it on a
display screen. Whether it is made of metal and draws on paper, or made
of electrons and draws on a video screen, the turtle is what educational
psychologists call a transitional object -- and what Papert calls
an "object-to-think-with."
Instead of "programming the computer" to draw a pattern, children are
encouraged to "teach the turtle" how to draw it. They start by
"pretending to be the turtle" and trying to guess what the turtle would
do in order to trace a square, a triangle, a circle, or a spiral. Then
they give the turtle a series of English-like commands, typed in through
a keyboard, to "teach the turtle a new word."
If the procedure followed by the turtle in response to the typed
commands doesn't achieve the desired graphic effect, the next step is to
systematically track down the "bug" that is preventing success. The fear
of being wrong is replaced in this process by the immediate feedback of
discovering powerful ideas on one's own.
After decades of research, Papert summarized the results of his LOGO
work for a general audience in Mindstorms: Children, computers, and
powerful ideas. In this manifesto of what has grown into an
international movement in both the educational and computing
communities, Papert reiterated something important that is easy to lose
in the complexities of the underlying technology -- that
the purpose of any tool ought to be to
help human beings become more human:
In my vision the computer acts as a transitional
object to mediate relationships that are ultimately between person
and person. . . .I am talking about a
revolution in ideas that is no more reducible to technologies than
physics and molecular biology are reducible to the technological
tools used in laboratories or poetry to the printing press. In my
vision, technology has two roles. One is heuristic: The computer
presence has catalyzed the emergence of ideas. The other is
instrumental: The computer will carry ideas into a world larger than
the research centers where they have incubated up to now.
When he came across the LOGO work,
during the time he was meditating about the fact that he had put two
years into the FLEX machine only to find that it wasn't amenable to
humans who tried to use it, Alan Kay recalls that "it was like a light
going on in my head. I knew I would never design another program that
was not set up for children."One of the first things he
understood was that a program or a programming language that can be
learned by children doesn't have to be a "toy." The toy can also serve
as a tool. But that transformation doesn't happen naturally -- it comes
about through a great deal of work by the person who designs the
language. Kay already knew that the most important tools for creating
personal computing were to be found in the software, but now it dawned
on him that the power those tools would amplify would be the power to
learn -- whether the user is a child, a computer systems designer,
or an artificial intelligence program.
Although he knew he had a monstrous software task ahead of him if he
was to create a means by which even children could use computers as a
simulation tool, his FLEX experience and his exposure to LOGO convinced
Kay that there was far more to it than just building an easy-to-operate
computer and creating a new kind of computer language.
It was something akin to the problem of
building a tool that a child could use to build a sandcastle, but would
be equally useful to architects who wanted to erect skyscrapers.
What he had in mind was an altogether new kind of artifact: If he ended
up with something an eight-year-old could carry in one hand and use to
communicate music, words, pictures, and to consult museums and
libraries, would the device be perceived as a tool or as a toy?
Kay began to understand that what he
wanted to create was an entirely new medium -- a medium
that would be fundamentally different from all the previous static media
of history. This was going to be the first dynamic medium -- a
means of representing, communicating, and animating thoughts, dreams,
and fantasies as well as words, images, and sounds. He recognized the
power of Engelbart's system as a toolkit for knowledge workers like
editors and architects, scientists, stockbrokers, attorneys, designers,
engineers, and legislators. Information experts desperately needed tools
like NLS. But Kay was after a more universal, perhaps more profound
power.
One of the concepts that played a big part in Papert's LOGO project,
and thus influenced Alan Kay and others, was derived from the thinking
of John Dewey, whose work encouraged generations of progressive
educators. Dewey developed a theory that Piaget later elaborated -- that
the imaginative play often mistakenly
judged by adults to be "aimless" is actually a potent tool for learning
about the world. Kay wanted to link the natural desire to explore
fantasies with the innate ability to learn from experimentation, and he
knew that the computer's power to simulate anything that could be
clearly described was one key to making that connection.
Alan wanted to create a medium that
was a fantasy amplifier as well as an intellectual
augmentor. First he had to devise a language more suited for his
purposes than LOGO, a "new kind of programming system that would attempt
to combine simplicity and ease of access with a qualitative improvement
in expert-level adult programming." With the right kind of programming
language, used in conjunction with the high-powered computer hardware he
foresaw for the near future, Kay thought that an entirely new kind of
computer -- a personal computer -- might be possible.
Such a software advance as the kind Kay envisioned could only be
accomplished by using hardware that didn't exist yet in 1969, since the
computing power required for each individual unit would have to be
several hundred times that of the most sophisticated time-sharing
computers of the 1960s. But at the end of the 1960s, such previously
undreamed-of computing power seemed to be possible, if not imminent. The
year 1969 was pivotal in the evolution of personal computing, as well as
in Alan Kay's career. It was the year that the ARPAnet time-sharing
communities began to discover that they were all plugged into a new kind
of social-informational entity, and enthusiastically began to use their
new medium to design the next generations of hardware and software.
After he finished his thesis on FLEX, Kay began to pursue his goal of
designing a new computer language in one of the few places that had had
the hardware, the software, and the critical mass of brain power to
support his future plans -- the
Stanford Artificial
Intelligence Laboratory. He had a lot to think about. There were
many great programmers, but very few great creators of programming
languages.
The programming language for the eventual successor to FLEX was his
primary interest, not only because he
knew that the hardware would be catching up to him, but because he knew
that programming languages influence the minds of the people who use
computers. In 1977, after the task of creating his new
programming language, Smalltalk, was accomplished, Kay described
the importance of this connection between a programming language and the
thinking of the person who uses it:
The particular structure of a symbolic language is
important because it provides a context in which some concepts are
easier to think about than others. For example, mathematical
notation first arose to abbreviate concepts that could be expressed
only as ungainly circumlocutions in natural language. Gradually it
was realized that the form of an expression and manipulation could
be of a great help in the conception and manipulation of the meaning
for which the expression stood. . . .
The computer
created new needs for language by inverting the traditional process
of scientific investigation. It made new universes available that
could be shaped by theories to produce simulated phenomena.
The "inverting" of "the traditional process of scientific
investigation" noted by Kay was the source of the computer's power of
simulation. And the ability to simulate ideas in visible form was
exactly what a new programming language needed to include in order to
use a computer as an imagination amplifier. If Piaget was correct and
children are both scientists and epistemologists, a tool for simulating
scientific investigation could have great impact on how much and how
fast young children and adult computer programmers are able to learn.
According to the rules of scientific induction, first set down by
Francis Bacon three hundred years ago, scientific knowledge and the
power granted by that knowledge are created by first observing nature,
noting patterns and relationships that emerge from those direct
observations, then creating a theory to explain the observations.
With the creation of a machine that
"obeyed laws you wanted to be held true," it became possible to specify
the laws governing a world that doesn't exist, then observe the
representation created by the computer on the basis of those laws.
Papert called these simulated universes "microworlds," and used
LOGO-created microworlds to teach logic, geometry, calculus, and
problem-solving to ten-year-olds. Part of the fascination of a good
video game lies in the visual impact of its microworld representation
and the amount of power given to the player to react to it and thus
learn how to control it. In Smalltalk, every object was meant to be a
microworld.
Computer scientists talk about
computational metaphors in computer languages -- alternative
frameworks for thinking about what programming really does.
The most widespread and oldest metaphor
is that of a recipe, the kind of recipe you create for a very
stupid but obedient servant -- a list of definite, step-by-step
instructions that could provide a desired result when carried out by a
mindless instruction-following mechanism. The sequence of instructions
is an accurate but limiting metaphor for how a computer operates. It is
a reflection of the fact that early computers were built to do just one
thing at a time, but to do it very fast and get on to the next
instruction.
This model, however, is not well
suited to computers of the future, which will perform many processes at
the same time (in the kind of computation that is called parallel
processing). Languages based on the dominant metaphors of numerical,
serial procedures are much better suited for linear processes like
arithmetic and less well suited for exactly those tasks that computers
need to perform if they are to serve as representational media.
Parallel processing is also a better
model of the way human brains handle information.
Starting from concepts set forth in LOGO and in Simula, Kay began to
devise a new metaphor in which the string of one-at-a-time instructions
is replaced by a multidimensional environment occupied by objects that
communicate by sending one another messages.
In effect, he started out to build a
computer language that would enable the programmer to look at the host
computer not as a serial instruction follower, but as thousands of
independent computers, each one able to command the power of the whole
machine.
In 1969 and 1970, the growing impact of the Vietnam war and the
pressure by congressional critics of what they interpreted as "frivolous
research" contributed to the death of the "ARPA spirit" that had led to
the creation of time-sharing and computer networks. The "Mansfield
Amendment" in 1970 required ARPA to fund only projects with immediately
obvious defense applications. Taylor was gone. The AI laboratories and
the computer systems designers found funding from other agencies, but
the central community that had grown up in the sixties began to
fragment.
The momentum of the interactive approach to computing had built up
such intensity in its small following by the late 1960s that everybody
knew this fragmentation could only be a temporary situation. But nobody
was sure where, or how, the regrouping would take place. Around 1971,
Alan began to notice that the very best minds among his old friends from
ARPA projects were showing up at a new institution a little more than a
mile away from his office at the Stanford AI laboratory.
By the beginning of 1971, Alan Kay was a Xerox consultant, then a
full-time member of the founding team at the
Palo Alto Research
Center. By this time, the hardware revolution had achieved another
level of miniaturization, with the advent of integrated circuitry and
the invention of the microprocessor. Xerox had the facilities to design
and produce small quantities of state-of-the-art microelectronic
hardware, which allowed the computer designers unheard-of power to get
their designs up and running quickly. It
was precisely the kind of environment in which a true personal computer
might move from dream to design stage. Alan Kay was already thinking
about a special kind of very powerful and portable personal computer
that he later came to call "the Dynabook."
Everybody, from the programmers in the "software factory" who
designed the software operating system and programming tools, to the
hardware engineers of the Alto prototype machines, to the Ethernet
local-area-network team who worked to link the units, was motivated by
the burning desire to get a working
personal computer in their own hands as soon as possible. In
1971, Alan wrote and thought about something that wasn't yet called a
Dynabook but looked very much like it. Kay's Learning Research Group,
including Adele Goldberg, Dan Ingalls, and others, began to create
Smalltalk, the programming "environment" that would breathe
computational life into the hardware, once the hardware wizards
downstairs cooked up a small network of prototype personal computers.
One of the most important features of the anticipated hardware was
the visual resolution of the display screen.
One of the things Alan had noticed when
watching children learn LOGO was that kids are very demanding computer
users, especially in terms of having a high-resolution, colorful,
dynamic display. They were accustomed to cartoons on television and
70-mm wide-screen movies, not the fuzzy images then to be found on
computer displays. Kay and his colleagues knew that hardware
breakthroughs of the near future would make it possible to combine the
interactive properties of a graphical language like Sketchpad with very
high-resolution images.
The amount of image resolution possible on a video display screen
depends on how many picture elements are represented on the screen. Kay
felt that the threshold number of picture elements needed to most
strongly attract and hold the attention of a large population of
computer users, and give the users significant power to control the
computer, would be around one million dots. (The resolution of a
standard snapshot is the equivalent to about four million dots.) The
Alto computer being constructed for PARC researchers -- which the
Learning Research Group called "an interim Dynabook" -- would have
around half a million dots.
The technique by which the Alto would
achieve its high-resolution screen was called "bit-mapping," a
term that meant that each picture element, each dot of light on the
display screen, was connected to one bit of information in a specific
place in the computer's memory, thus creating a kind of two-way
informational map of the screen. If, for example, a specific bit in the
computer's "memory map" was turned off, there would not be a dot of
light at the location on the screen. Conversely, an "on" bit at a
coordinate in the memory map would produce a dot of light at the
designated screen location. By turning on and off parts of the bit map
through software commands, recognizable graphic images can be created
(and changed) on the screen.
Bit-mapping was a major step toward creating a computer that an
individual could use comfortably, whether the user is an expert
programmer or a beginner. The importance
of a visual display that is connected directly to the computer's memory
is related to the human talent for recognizing very subtle visual
patterns in large fields of information -- undoubtedly a survival trait
that evolved way back when our ancestors climbed trees and prowled
savannas.
Human information processors have a
very small short term memory, however, which means that all computers
and no humans can extract the square roots of thousand-digit numbers in
less than a second, no computers and all humans can recognize a familiar
face in a crowd. By connecting part of the computer's internal processes
to a visible symbolic representation, bit-mapping puts the most
sophisticated part of the human information processor in closer contact
with the most sophisticated part of the mechanical information
processor.
Bit-mapping created more than a passive window on the computer's
internal processes. Just as the computer could tell the human who used
it certain facts about whatever it had in its memory, the user was also
given the power to change the computer by manipulating the display. If
users change the form of the visual representations on bit-mapped
screens, using keyboard commands, lightpens (� la Sketchpad), or
pointing devices like mice (� la Engelbart), they can also change the
computer's memory. The screen is a representation, but it is also a
control panel -- a drawing on a bit-mapped screen can be nothing more
than a drawing, but it can also be a kind of command, even a program, to
control the computer's operations.
If, for example, you were to use a mouse to move a video pointer on
the screen to touch a visual representation of a file folder or an out
basket, and you could call the folder, for example, from the computer's
memory and display a document from it on your screen simply by pointing
to it, or send the contents of the computer-stored out basket to
somebody else's in basket, then a person would be able to accomplish the
kind of work done in offices, even if that person knew nothing about
computer programming. Which, after all, was the potential future market
that motivated Xerox management to create PARC and cut their whiz kids
loose in the first place.
Creating new kinds of computer input
and output devices to help human pattern recognition mesh with
mechanical symbol manipulation is known as "designing the human
interface," an art and science that had to be created in the
1970s in the kind of human-computer partnership envisioned by Licklider
and Engelbart in the 1960s, which could start to happen by the 1980s.
Alan Kay's Smalltalk project played a key role in the evolution of the
Alto interface, and as such was integral to the eventual company goals
in the office automation market. But
even at the beginning, Kay started bringing children into the project.
Part of the Smalltalk project's effect on the early days at PARC was
inspirational. It wasn't long before the rest of the team understood
Alan's desire to bring children into the process of designing the same
instrument that he and all the other computer scientists wanted to use
themselves. Another aspect of Kay's contribution was more concrete: the
absolute conviction that they were designing something meant for people
to use. That might not sound too revolutionary today, but even as late
as 1971, most of the top-flight computer scientists who believed that
this tool was going to be more than just a gadget for computer
programmers were at PARC.
PARC in the early 1970s was a collection of the worlds best computer
scientists, hardware engineers, physicists, programmers . . . which
meant that it was also a collection of people with strong personalities
and definite opinions. Bob Taylor, Alan
Kay, Butler Lampson, Bob Metcalfe, and their colleagues each had his own
unique approach to creating personal computing, but they agreed on one
fundamental assumption -- that their ultimate product should be as
generally useful as a hammer, or pulley, or book. Secretaries and
business executives would one day be able to use the same tool to help
them perform the work. Architects and designers would have the power of
modeling, forecasting, and simulation at their fingertips.
A true personal computer, the diverse
PARC groups agreed, ought to be usable by legislators and librarians,
teachers and children. And a computer that could be commanded by looking
at images on a screen and pointing to them by means of a mouse was
certainly a lot more widely usable than a computer that required arcane
keyboard-entered commands in order to function.
The first Alto personal computer prototypes were distributed to PARC
researchers in 1974. As they had predicted, the creation of an
environment in which every researcher had, for the first time in
history, personal access to a powerful computer, and the means to
communicate with all of his or her colleagues' computers, had a profound
effect on their ability to do their job of designing even more powerful
computer systems.
By the late 1970s, yet another generation of even more advanced
hardware and software had been created by a network of nearly a thousand
researchers at PARC equipped with Altos, communicating via Ethernet
networks. But the outside world, and many people in the computer world,
were still unaware of the potential of personal computers. The problem,
as PARC alumnus Charles Simonyi was to point out in 1983, an eventful
decade later, was that Xerox management couldn't be faulted for not
realizing in 1973 that PARC was more than ten years ahead of an industry
that wouldn't even exist until 1975.
Another small cloud on the horizon in the mid-1970s -- the
"home-brew" computer hobbyists who were building their own low-power
microcomputers -- became a gathering storm of popular interest in
personal computing by the end of the 1970s. The microcomputer hobbyists,
who assembled the new microprocessor chips into operational computers,
were for the most part unaware of the far more powerful devices that
were in use in Palo Alto years before a tiny company in New Mexico, the
now-legendary MITS, produced the first affordable, do-it-yourself
computer -- the Altair.
In March, 1977, Alan Kay and
Adele Goldberg condensed a PARC technical report into an article,
the title of which described both the dream and the reality of the
Smalltalk branch of the PARC project: "Personal Dynamic Media" was
published in a magazine named Computer, during a time when
computer magazines were for specialists. Like Bush, Licklider, Taylor
and Engelbart before them, Kay and Goldberg did not talk of circuits or
programs, but of media, knowledge, and creative human thought:
For most of recorded history, the interactions of
humans with their media have been primarily nonconversational in the
sense that marks on paper, paint on walls, even "motion" pictures
and television do not change in response to the viewer's wished. A
mathematical formulation -- which may symbolize the essence of an
entire universe -- once put down on paper, remains static and
requires the reader to expand on its possibilities.
Every message is, in one sense or another, a
simulation of some idea. It may be representational or abstract.
The essence of a medium is very much dependent on the way messages
are embedded, changed, and viewed. Although digital computers were
originally designed to do arithmetic computation, the ability to
simulate the details of any descriptive model means that the
computer, viewed as a medium in itself, can be all other media
if the embedding and viewing methods are sufficiently well provided.
Moreover, this new "metamedium" is active -- it can respond
to queries and experiments -- so that the messages may involve the
learner in a two-way conversation. This property has never been
available before except through the medium of an individual teacher.
We think the implications are vast and compelling.
A dynamic medium for creative thought: the
Dynabook. Imagine having your own self-contained knowledge
navigator in a portable package the size and shape of an ordinary
notebook. Suppose it had enough power to outrace your senses of
sight and hearing, enough capacity to store for later retrieval
thousands of page-equivalents of reference materials, poems,
letters, recipes, records, drawings, animations, musical scores,
waveforms, dynamic simulations, and anything else you would like to
remember and change.
The Learning Research Group introduced students from the nearby Jordan
Middle School in Palo Alto to what they called "interim Dynabooks."
Nearly a decade before keyboards and
display screens became familiar appliances, these children were
introduced to a device no child and only a few computer scientists had
seen before -- an Alto computer set up to run Smalltalk. By using
the mouse and the graphics capabilities provided by the hardware and
software, these students were able use Smalltalk to command the computer
in much the same way that Papert's students in Cambridge, years before,
had learned to program in LOGO by "teaching the turtle new words."The
screen was either a "very crisp high-resolution black-and-white CRT or a
lower resolution high quality color display." High-fidelity speakers and
sound synthesizers, five-key keyboards like Engelbart's, and piano-like
keyboards were also available. The system could store the equivalent of
1500 pages of text and graphics, and the processor was capable of
creating, editing, storing, and retrieving documents that consisted of
words, graphic images, sounds, numbers, or combinations of all four
symbol forms.
The mouse could be used to draw as well as to point, and an "iconic
editor" (another Smalltalk innovation)
used symbols that children who were too young to read could use to edit
graphics; e.g., instead of typing in a command to invoke a
graphics cursor, a child could point to a paintbrush icon.
The interim Dynabook could be used to read or write an old-fashioned
book, complete with illustrations, but it could also do much more: "It
need not be treated as a simulated paper book since this is a new medium
with new properties. A dynamic search may be made for a particular
context. The non-sequential nature of the file medium and the use of
dynamic manipulation allows a story to have many accessible points of
view; Durrell's Alexandria Quartet, for instance, could be one
book in which the reader may pursue many paths through the narrative,"
wrote Kay and Goldberg.
The dynamic nature of the medium was made clear to the users as they
became acquainted with the toolkit for drawing, editing, viewing, and
communicating. Smalltalk was not just a language, and the Alto system
was not just a one-person computer.
Together, the hardware, the software, and the tools for the users to
learn the software, constituted an environment -- a small
symbolic spaceship that the first-time user learned to control and steer
through a personal universe.
The ability of the users to personalize their representation and use
of information became clear as the children from Jordan Middle School
experimented with changing typefonts for displaying letterforms, and
with changing the bit-maps of the computer to create and animate cartoon
images in mosaics, lines, and halftones. The users not only had the
capability to create and edit in a new way, but once they learned how to
use the medium they gained the ability to make their own choices about
how to view the universe of information at their fingertips.
The editing capabilities of the Dynabook made it possible to display
and change every object in the Smalltalk microworld. Text and graphics
could be manipulated by pointing at icons and lists of choices --
"menus" in software jargon -- and multiple "windows" on the display
screen made it possible to view a document or group of documents in
several different ways at the same time. The filing capabilities made it
possible to store and retrieve dynamic documents that could consist of
any collection of objects that could be displayed and have something to
do with each other. Drawing tools and painting programs made it possible
to input information freehand as well as through the keyboard.
The structure of the Smalltalk language, the tools used by the
first-time user to learn how to get around in the Dynabook, and the
visual or auditory displays were deliberately designed to be mutable and
movable in the same way: "Animation, music, and programming," wrote Kay
and Goldberg, "can be thought of as different sensory views of
dynamic processes. The structural similarities among them are apparent
in Smalltalk, which provides a common framework for expressing those
ideas." A "musical score capture system" called OPUS and a graphic
animation tool called SHAZAM were part of the Smalltalk-Dynabook
toolkit.
In 1977, Scientific American's annual theme edition was
dedicated to the subject of "Microelectronics." Alan Kay's contribution
to the issue, "Microlectronics and the Personal Computer," was the only
article that directly talked about the meaning of this new technology
for people. The magazine's editors summed up the piece in a
two-sentence subtitle: "Rates of progress in microlectronics suggest
that in about a decade many people will possess a notebook-sized
computer with the capacity of a large computer of today. What might such
a system do for them?"
One of the first things Kay pointed out was the connection between
the use of interactive graphic tools and the exercise of a new cognitive
skill -- a skill at selecting new ways to view the world. The metamedium
which Kay still saw to be a decade in the future would only achieve its
full power when people use it enough to see what it is about.
The power that the 1977 prototypes
granted to the human who used such devices was the power to create many
new points of view.
This freedom to change one's view of a microworld, Kay believed, was
one of the most important powers of the new kinds of representational
tools that were being designed and tested in the late 1970s. In
describing the way children learned to use the Smalltalk system, Kay
also described something of the nature of the experience:
Initially the children interact with our computer by
"painting" pictures and drawing straight lines on the display screen
with the pencillike pointer. The children then discover that
programs can create structures more complex than any they can create
by hand. They learn that a picture has several representations, of
which only the most obvious -- the image -- appears on the screen.
The most important representation is the symbolic model of the
picture stored in the memory of the computer. . . .
One of the best ways to teach nonexperts to
communicate with computers is to have them explore the levels of
abstraction at which images can be manipulated.
Kay noted that when he gave the same tool that the children used as
both an amusement and an entrance into Smalltalk programming to an adult
artist, the artist started out creating various designs similar to those
he was accustomed to making on paper. Eventually the artist discovered
that the properties of the new medium, and his increasing facility for
commanding those properties, made it possible for him to explore graphic
universes that he could never have created with his old media: "From the
use of the computer for the impoverished simulation of an already
existing medium," Kay wrote, "he had progressed to the discovery of the
computer's unique properties for human expression."
This freedom of viewpoint was only meant to be explored and
demonstrated in a preliminary way in Smalltalk: It was Kay's hope that
many new metaphors and languages would evolve as time went on, into what
he called "observer languages":
In an observer language, activities are replaced by
"viewpoints" that become attached to one another to form concepts.
For example, a dog can be viewed abstractly (as an animal),
analytically (as being composed of organs, cells, and molecules),
pragmatically (as a vehicle by a child), allegorically (as a human
being in a fairy tale) and contextually (as a bone's way to
fertilize a lawn). Observer languages are just now being formulated.
They and their successors will be the communication vehicles of the
1980s.
Kay set forth his theories about
personal computers as the components of a new medium for human
expression , and compares the recent and future emergence of
personal computers with the slower development cycles of past media.
He also predicted that the changes in
the human social order that were likely to accompany a new computerized
literacy would be much more sweeping than the effects of previous media
revolutions. The creation of a literate population would be the
first reason for such a change. Out of that literate population, perhaps
a few creative individuals would show the rest of us what could be
achieved. He declined to predict the specific shape of these social
changes, noting the failure of previous attempts at such forecasting:
We may expect that the changes resulting from
computer literacy will be as far reaching as those that came from
literacy in reading and writing, but for most people the changes
will be subtle and not necessarily in the direction of their
idealized expectations. For example, we should not predict or expect
that the personal computer will foster a new revolution in education
just because it could. Every new communication medium in this
century -- the telephone, the motion picture, radio and television
-- has elicited similar predictions that did not come to pass.
Millions of uneducated people in the world have ready access to the
accumulated culture of the centuries in public libraries, but they
do not avail themselves of it. Once
an individual or a society decides that education is essential,
however, the book, and now the personal computer, can become among
the society's main vehicles for the transmission of knowledge.
The difference between a Dynabook of the future and all the libraries
of the past, however, would depend upon the dynamic nature of this
medium. A library is a passive repository of cultural treasures. You
have to go in and dig out your own meanings.
A Dynabook would combine the addictive
allure of a good video game with the cultural resources of a library and
a museum, with the expressive power of an animated fingerpaint set and a
synthesized orchestra. Most importantly, it would actively find
the knowledge appropriate for the task of the moment, communicated in
the form and language best suited to each individual who used it.
The intelligence of such devices -- the reason that software
breakthroughs in artificial intelligence research would someday have to
intersect with the evolution of personal computers -- would influence
their ability to bring resources to the person who needs them.
When the machines grow smart enough to
communicate with eight-year-olds, then the question will shift from how
to build a computer that people can easily use to what we all do with
that kind of power.
What if libraries were able to find
out what most interests you and what you most need to know, and showed
you how to find what you wanted? What if you could say to the
library: "I wonder what it would be like to live in the Baghdad of the
Caliphate?" or "I wonder how it feels to be a whale?" and expect the
library to show you? Do you like Van Gogh? How about a simulation
of the fields outside his house? Would you care to sit in with Louis
Armstrong or Wolfgang Mozart? What would it do to the world if we could
all see how everybody else lived and share in their cultures?
If the first effect of the coming metamedium was likely to be the
creation of a literate population who shared a new freedom to use
symbols and to choose how to view information, then the second effect
lay in the power that would be unique to this medium --
the power of simulation.
Simulation is the power to see what you imagine, to create worlds
that obey your command. The computer can build instant sensory
representations. The user/programmer explores a universe that reacts,
in which the degree of the user's power depends upon and grows with
one's understanding of the way the worlds work.
The power of simulation to empower the imagination and give form to
whatever can be clearly discerned in the mind's eye is what makes this
kind of device a "fantasy amplifier." Although there are several
homilies that are entitled to be called "Kay's First Law," the statement
that he most often calls "Kay's Second
Law" is: "Any time you build a fantasy amplifier, you have a winner."
His reasoning is that game playing and fantasizing are metaphors for the
kind of skill people need to get around in the world.
"We live in a hallucination of our own devising," Kay is fond of
saying. But our illusion is so complex, so much of the world we
experience appears to be beyond our control, and the operating manual is
so difficult to find, that we all tend to get locked into the way our
families, societies, and cultures see the world. "We can't exist without
fantasy, Kay asserts, "because it is part of being a human. A fantasy is
a simpler, more controllable world."
And by practicing how we would control a simpler version of the
world, we often figure out how to operate the world outside the fantasy.
A game is both controllable and challenging. It is entered vicariously,
purposefully, and with an open mind about the outcome. Sports and
science and art all involve vicarious, purposeful fantasies in that
sense. That's why he feels that video games were not a fad but a
precursor to something with much more profound power. And that is the
most likely reason why he joined Atari Corporation.
The power of simulation is not necessarily or exclusively a
beneficial one, as the legends of today's system-crashers, obsesses
programmers, and dark-side hackers attest, and as Kay warned in his
Scientific American paper:
The social impact of simulation -- the central part
of any computing -- must also be considered. First, as with
language, the computer user has a
strong motivation to emphasize the similarity between simulation and
experience and to ignore the great differences that symbols
interpose between models and the real world. Feelings of
power and a narcissistic fascination with the image reflected back
from the machine are common. Additional tendencies are to employ the
computer trivially (simulating what paper, paints, and a file
cabinet can do), as a crutch (using the computer to remember things
that we can perfectly well remember ourselves) or as an excuse
(blaming the computer for human failings).
More serious is the human propensity
to place faith in and assign higher powers to an agency that is not
completely understood. The fact that many organizations actually
base their decisions on -- worse, take their decisions from --
computer models is profoundly disturbing given the current state of
computer art . . . .
The fact of simulation is so seductive to human perception,
and so potentially useful in "real world" applications, that its
widespread use is inevitable, once personal computers grow sophisticated
and inexpensive enough. The ethics of how and for what
purposes simulations should and should not be used are only
beginning to be formulated. The historical events, debates in PTAs and
legislatures, and growth in public concern that will accompany the
introduction of this medium will help determine the shape of the future
ethics of simulation. The best place to look for expert guidance, Kay
suggests, might be to those of us who are the least prejudiced by
precomputer ways of thinking:
