Dynabook - The Complete Story

From Viewpoints Intelligent Archive
Jump to: navigation, search
again
welcome to this program this is one of our occasional
series of Bay Area Computer History perspectives
I'm Peter Newark say from Sun and also
organizing the program is Jeanne treichel also from some apps
Jeanne could send up a moment yep away we have done
so Jeanne and I appreciate any
no suggestions for future programs the next program in the
series will be October October 24th
Neil's Neil from that s RI talking about shaky
major contribution from the Bay Area
to robotic history so our past
programs have often been in the form of panel presentations
is your Ethernet for example but
on the subject tonight Dynabook it's pretty clear
eally one person to talk about it and that's Alan
Kay and Alan is going to talk to aside
thank you very much Alex
well
one
of the things they're enough people with gray here
in the audience so that I can't lie
as much as I usually do when I tell this story
the other thing is a couple years ago
I wrote a fairly extensive history
of small talk which is published as
a preprint in the
sig plan as
part of the hopeful to proceedings going to be
in a book sometimes we have some in our office
I'm going to leave one with Peter and Jeannie and
it's more detailed well
it's more much more about small talk than
the Dinah books it has that story in
there it's more detailed and I can cover tonight
one of the things I discovered when I wrote this
is it is simply not possible to tell the truth when you're doing
a history
even though there are about 50,000 words in
it I had to leave out I apologize in
the beginning for how many things I left out I also
found myself leaving out enormous
number really juicy and very
funny stories just because the page
limit was getting longer and longer and
longer and it was impossible to
express even in this longer thing
all the different intellectual debts and so forth
and again tonight the
stories that I tell you can't possibly be the
truth and a lot
is going to be emitted and however
some people in the audience might help my
recollection on a couple of cases
that case I'm showing early signs
Alzheimer's on just that although turns out that Alzheimer's
is okay for one of these things because
remember the stuff that happened earlier quite in detail
so I think I
think I do remember that now
what I'm going to do is I was
originally asked to recapitulate the
talk I
gave at hopeful but that
really intended to be supplemented by the by
the paper which many of you have not
read so I've done a different talk and
I'm going to move through it as quickly as
I possibly can if I don't degress too much we
might get done more or less on time if I do degress
then we'll probably
time will run out before I ever get
to the invention of the Dynabook we'll see
so
let's see I'm going to have
a flat screen display here which is for controlling
the video but I don't see how to control the slides okay
alright
I have to go to another pillow okay
this
is a mode full user interface
okay slide projector
on
first one to go forward
get okay goes fast forward I
don't want to go okay so
the context for this the other thing I found out
in preparing this
talk is a lot of my best explanations came
post facto this is partly because
we found it extremely difficult to explain these
ideas to Xerox and over
the years I found better examples
however in order to keep this historical I want
to try and stick pretty much with what
we're thinking at the time and so I've
left out some of some of the better ways of explaining
this stuff but I think there's enough context now this is after
all let's see 27 or
28 years later there's
context so it shouldn't matter but I want to have set
up one piece of context which in fact we did think about
back then although we didn't completely
realize that we were doing it ourselves was
much more when we're thinking about how to help kids be creative
and that is an idea from Arthur kessler's
creation which is written written in the mid-60s and
one of his models of the creative act
is that we tend to think in context
the context here is pink and here's
a train of thoughts working in
through the pink you can think of the pink as a paradigm to
give us a goldfish bowl you can think of as a box whatever
you like but the idea is that
there seem to be constraints in
human mind to keep you thinking about one reality
at a time and so when you have
little deviant thoughts
little blue thoughts they get
immediately you go back to
pink thinking but every once in a while maybe
when you're in a hot tub taking a shower out jogging
something one of these blue thoughts breaks through and all
of a sudden you see that this pink thing that you
thinking about is really a blue thing
and moreover it's much more interesting
as a blue thing than it was as a pink thing so this
is what Kessler calls by association the
colliding of two contexts usually
by analogy or metaphor and
of course 90% of them are really
bad ideas most ideas are
bad but if you
can filter these this is the way a lot
of creative acts get done and Kessler pointed out that
hree basic emotional reactions to doing this if
you're telling a joke it's hahaha
you're doing science it's a ha and
if it's art it's off
because as he points out a joke
is leading somebody down a garden path
and then realizing it was about something else all along
and science has the same
thing we often feel that the ideas are all
can't see them because they're transparent
to the context that we're in and of course art is
again being able to see one thing as something
else in a more powerful way so
the context for this is that I think
in most of these talks
you'll hear people talking about things that they've done it
will they'll fall into these patterns you know we are
looking at it this way and all of a sudden you see it something else and that
gives us a new insight and new energy
and of course one of the things
not that this group needs this but it's
amazing how much being
generally well educated helps
because if you're only thinking about pink if
pink things in your brain is really hard to get blue
ideas one of the problems with Silicon
Valley engineers is they're too good at what they do and
so they spend all of their time optimizing in the
only context that they've learned so
you see super optimization is
not particularly a creative act by Tesla's model
so thinking about other things we
reverse some of the most unusual things
Dave Smith who's sitting in front found himself
I think to his great surprise reading
books by Ernst Gombrich one
called art and illusion which is about
theories of artistic perception doing
a computer science thesis at Stanford
and that wasn't the most unusual book that I read
so
the the
genesis of all this stuff is Memex
this is a picture from this the
Life magazine version
of this article written by Bush fan
of our Bush 50 years ago
there's by the way for those of you were interested on October
12th and 13th at MIT
there's a 50-year celebration for
Bush and a lot of people who are interested
in mimicks
and linking of information angle Bart will be there and so forth are
showing up to commemorate Bush's vision and
as you see in this in this drawing
the idea was he said
not too far in the future people will have in their
homes a desk that has inside of it about
five to ten thousand volumes worth of books held
on an optical storage
that there will be screens
for viewing it there'll be ways for putting in information and
drawing in information ways for searching it and most
importantly there'll be ways for linking the information using
calls called halves he said
there'll be profession called pathfinders and
pathfinders will make interesting connections
between knowledge and cell lows you can add
them to your mimics as you go along now
I was five years old in 1945 so
I missed this particular article
although I did read Life magazine back then but
ankle Bart was in the Navy in the Pacific
and did see
this he was in Guam or something and
it started a train of thought that for him gradually
culminated in setting up a project to actually
I'll return to that in a minute I
found out about it when I was about 14 or 15 because
Robert Heinlein like to use generic
names so
would take something that's capitalized like Memex and
he would uncapitalized it and use it as
a generic term used in the future as
a descriptive term so he would you say something like so-and-so
used his Memex to do blah blah and
Heinlein was very reliable about this
and so I discovered early on if I actually looked up
these uses of these things there would be something
in somebody had done something that
gave him this idea so
he would give brand names to things and if you look up the brand names
at Heinlein use you would discover somebody had had
invented a precursor technology that Heinlein hit
so that's how I found that science fiction is very useful especially
back back in the old days when they actually wrote about science
so
fast forward to around
1962 here's Ivan someone
sitting at the console of a computer tx2
the size of this room and I think this is
believe isn't a genie that Ivan gave his talks are
not going to say too much about sketchpad I
didn't know about sketchpad at the time although I'd
started programming for the air force the year
earlier and I'm going to show
you what you see on
the screen there is Ivan has drawn a bridge in and
sketchpad was peculiar even
by today's standards in that you
could attach meaning to every element of a drawing and sketchpad
would attempt to synthesize all of the meanings of
the parts into one grand
simulation and the simulation was
dynamic if it was under constraint so if
you drew if you drew a linkage in with sketchpad and
the linkage was not completely constrained
to be still the constraint solver
and sketchpad would actually iterate its way around through the degrees
of freedom not unlike the way a Prolog program
will give you all the all of the answers for
a particular retrieval but
the one of the really interesting elements
of sketchpad and we may not talk about
it too much further depending if I run out of time or
not is that sketchpad
had
a peculiar another peculiar notion which
is that you didn't have to be exactly right in
order to be interesting so the
constraint solvers in this were not symbolic constraints
Altis they weren't performing logical
operations but they're actually performing various kinds
of arithmetic operations and
fit operation so here's
a here's an example of the
view go ahead
go over here hit
play and see what happens
okay I'm worried now
so somebody well
I hope it should be warm
of course this thing is playing away and it
doesn't showing maybe
yeah okay so let me run just back
in technology wonderful now it's
blanking on me and
try to rewind a little bit
yeah Oh
another mode
okay
so
as
the Marvel it's moving to watch I'm just going to show you one small
I don't
know what it's doing really
okay he's drawing yeah
drawing a rivet here and
one of things you can do is
point you just told the constraint solver to go on
and to line things up and the center of the circle
oh boy
now he's going to distort the thing and
we turns on the solver
again the solver will line it up so it's symmetric
the center of the arch is the
center of the cross piece the
idea of a sketchpad is you put in generally what the thing is
give it some additional rules that characterize
the particular drawing you're trying to make those
ither have to do with the form of the thing but also can
have to do with the meaning of the thing and here's
another interesting thing about sketch pad
this is by the way before
there were real computer display so every dot is being
put up here by brute force by the TX to
the
nice thing about those old calligraphic displays is you got
even real time scale
changes which is still hard to come by on
on current day machine so it's too
bad and
now you can get a few more rivets
and now it's fun is he goes to the master drawing
says I better get rid of those cross pieces so I'll make them transparent
and lo and behold when we go
to the drawing all rivets
o this is actually the first object-oriented software
system and really
strong glory and anything that
you make out of it is in turn a master drawing
so here's he's made
the rivet in the flange a master
and now
he's making instances from those okay
so I have to go put it out of its misery here okay
we can change the change to the next tape
so sketchpad had
three really amazing
planners
there's
Ivan three amazing is what
people look like back then
so once said to Ivan in
one year almost entirely
by yourself you created the first real interactive
graphics program first system with a window if you're noticing
it has a clipping window and the drawings on our much larger
screen the first object-oriented software
system and the first real-time
problem-solving system you can even call it
an AI system I said how did you do
this and he said well you know it was hard he
actually in his faces here is
quite apologetic if anybody how many people here read
this what's amazing
about reading this thing besides how great it is still today is how
apologetic he is it doesn't do more
this is why it's not good
closely into the future you might get discouraged
okay another
thing that was happening right at the same time is the first live
vote for the first personal computer which is the link done
by Wes Clark and Charlie Molnar here
one of the design criteria is that it
shouldn't loom over you so it's designed to
small and actually the first 20 people built their own they
came spent a summer at Lincoln labs
and the summer was actually learning about digital
electronics and the project they did was to build their own
computer which they took back to them back with them to their lab
about a somewhere between a thousand
and two thousand of these machines were built in the in
the 60s mostly for biomedical technicians
these machines also
influenced Dec tremendously for instance
drives at West invented for this machine if you look there looks
like a deck tape but in fact deck tape was
invented was originally called a link tape deck tapes
just held files they were boring but link
tapes actually held pages of the paging virtual
personal computer only
had two K of 12 bit words on
a paging virtual memory on this tape and it was amazing
what you could do with this machine is a beautiful machine designed
to be taken care of by
people in labs for themselves
have a very nice programming system in a beautiful little editor the
debugger on it was quite remarkable in that you instead
of single stepping your program you single step
the machine so the software and the hardware were
such that you could actually slow the entire
machine down as you move
through to see what was actually going on beautiful
machine so
the other thing that happened around 1962
was Advanced Research Projects
Agency phased out of the space
business and had some money left over and
for various reasons all of them justified
they like this guy Jack Lickliter
known as Lick and decided
to give him a bunch of money to do whatever he wanted
and he was a psychologist and had
been involved in some of the human factors design of the sage
system in the 50s he is both a b b and
an at MIT and he decided that the destiny of personal
computing was to be a complementary partner
for human beings complementary
in the sense that not trying to do what a human does
but to do things that humans don't do well in
a very interactive way
so he was one of the people who could
extrapolate from pointing light guns
at digitally produced radar screens
to the idea that this thing might actually help
people think and his phrase
up there is in his one of his earliest papers
in 1960 which is called man computer symbiosis was
to think as no human has ever human
being has thought has ever thought I guess he
was a great guy and he
came up with and they just died
a few years ago and Nicholas Negroponte and
I are trying to raise money for a permanent
Licklider lectures at MIT
so several ideas he had one
you shouldn't spend any time more time in Washington than
you had to nobody
sensible likes that place so he
got the idea that you should only spend two
years there and then he got the idea that
the only people who should
article community should be people who came out of the
Arco community and that that person
would spend one year as a as
a deputy learning the job and then
two years as a director so this is the stream
of people who created the
funding that funded most of the technology that we have today
how did I even get
there well he got drafted not long after he did sketch
pad and Lickliter did not want him to go to Vietnam
so they pulled some strings and as a second lieutenant
he was made head of ARPA at the age of 26
that
was one of the great moves of all time
because I even you could even though each of these
four guys are fantastic you could argue that Ivan was the greatest
of the four of them because he was the youngest he
was the most vitally interested in the input-output
human interface part
of the thing he was the person for instance who first funded Engelbart
number of other things Taylor
will meet him again a little bit later because he was one of the people who
set up Jaques Park and Larry Roberts is
best known for the for the ARPANET
Larry Roberts was Ivan Sutherlands
room late at Lincoln labs
so here's the angel Bart
a little bit
show in a couple of minutes on their show in Engelbart
clip as I saw the system just
not yet this is happened
before I came on the scene then the
n I was
programming at the just trying to see what
it looks like here I was programming at
Air Training Command in the Air Force in the
early 60s and some
unknown programmer came up with a great idea because
back then didn't really have operating systems or
the most places would build their own little
operating system to help them process back batch jobs and
something and Air Training Command had many different
phases a lot of them had burrows
and somebody came up with this great idea for
how to make data
sensible to people and that
is the following each record on one
of the two twenties tapes was a long thing
those tapes didn't move very fast back then could
write long records on them and they're in
parts the first part were set of pointers back
into the second part which were set of procedures
that knew about the third part which
were the set of data records okay
and so the way you use these things so the operating system was minimal
right all
you have to do was to read this thing into the core
memory of the Burroughs 220 and start
jumping in direct vectoring
in direct procedure called indirectly through these guys
were into standard orders that had to do with
reading and writing and finding out how many fields the data had
these procedures knew how to give the answers that these
and the behaviors that these guys did and the
cold mess here knew what to
down there which are never touched directly by
anybody that a great idea don't
you wish you had that on the internet I
thought it was a pretty
good idea never been able
to find out who had this idea was there when I arrived
in 1961 somebody had just done it and
it worked extremely well until
the government decided everybody has the program in
COBOL a few years later and then it
got thrown out so this
is my first I
believe this antedates all
other forms of abstract data types that
I know about there's an early paper by Doug
Ross that has some of these ideas from MIT I
think this is a digital invention by somebody and
it certainly it's stuck in
my mind the other thing
that I learned there which I'm not going to try
and explain is it just happened that the third machine I
ever learned was the Burroughs be 5000 which
is probably the greatest not probably it
was was the greatest machine architecture ever
designed I think
7 out of the 10 best software
thought up are in the hardware of this machine I
will not try and explain with all of them are but it
was basically a crash 'less machine is the first machine
to do what is known as capability addressing built into
the hardware one of the first machines to have a stack the
first machine to have byte codes there's no there are no addresses
allowed in the code all the code is relocatable
and in fact it was kind of an object-oriented
machine although this was before people knew how to handle it that
way it was actually done for Algol 57
Algol 58 which on
this machine was called Balga back then but
hey got around to building the machine Barton guy
who designed to be 5,000 had
designed the first moderate what
consider the first modern compiler which ran on the Bo's to 20
and Knuth was one of the programmers of it
why do I say modern I think everybody
in here knows what I mean when I say pass and a half
compiler which is a compiler that streams once
through the source code puts
the in fix-up pointers as it goes and then as it finds out
information about for references fixes those I mop the first
one of those ran on the bar was 220 when Barton was designing
it he realized that the intermediate
output of the compiler was actually more
interesting than borrows 220 machine code so he thought about why
not just design the machine to run the direct output
of the compiler and forget about all this down
below which had nothing to do with
higher-level languages and I
commend his paper to you I think
papers ever written in computer science called
a new approach to the functional design of a digital computer
Western joint computer conference
1961 it's only four
pages long that gives the complete architecture of this system I should
mention by the way you know hot Java is
a hot thing now the virtual machine
on the B 5000 is much better than Java
and fact
the all the virtual machines
we get at Xerox PARC were based on this particular architecture
anybody who wants to do a good virtual machine shouldn't
bill themselves of that information rather than just trying
to roll their own
here's another
thing that I came across at 1965
which is Gordon Moore's original graph
saying what was going to happen to Silicon
published in data nation 1965
I read it while I was helping debug this
infamous system called the chippewa operating system
for the 6600 in seymour crazed lab
was back when you still work for control data
so this is I'm semi-log
paper and first
thing Moore did was just plot what
happening with integrated circuits from 1959
up to 65 and notice that it was an
exponential power of two every year and
started asking questions about what
would it be like if you could continue that and a few years later
he did a more careful analysis and said well it's not going to be worse
than a factor of two every two years that's this red line
and here's the line for
memory dynamic Ram memory
as it started in about 1971 so
Gordon Moore on the basis of physical principles and
because he picked the right kind of silicon MOS silicon
to look at this
is what led many people astray because back in those days it
was so slow that nobody would ever think of building a
computer about it because it basically has it
works like blowing up a balloon with buckets
basically a capacitive based way
of doing switching and so you have to put a lot of charge into
an MOS transistor in order
to get it to switch it took a long time but
gordon moore looked at it in part by
talking with him about it once in part just
because it was so much simpler to make that than bipolar that he could actually
do some math on it bipolars has like 14
layers and it's very complicated but the mos
he realized that the smaller you make an
transistor the faster is going to be able to switch and
so as a main question was how
small can we make these guys and still hasn't be a transistor and
answer back then was about 5000 electrons we're now
getting close to so
5,000 electrons took his
graph out to about 1995
so this is one of the great thirty-year predictions of all
time and of course everybody here knows that
did the other thing you need to do the to predict
the future which is he helped invent it right
the left fairchild is one of the founders
of Intel and is worth a few billion that worth as much as Bill
Gates but he's a few billion is is nothing
to sneeze at okay
so I have to tell you that when I
first saw this it meant nothing to me because
I was be bucking a machine that ran at
10 myths in 1965 and it required
liquid freon cooling
so there's no particular way
extrapolating
at that time I found
myself at Utah and
that Utah the first thing
when you walk in a Dave Evans is office this is what he looked like is a
real friendly guy look before you even got a desk he
had a pile of these things on it about this high
so the first thing he would do is give
take this and we eat it so that
and then you were supposed to come back and if you could
talk sensibly about it then he would give you a desk
and
it happened on my desk was a pile
of tapes and listings that was
supposed to be the algal compiler for the UNIVAC
1108 at the university of utah had but
in fact it turned out to be the first Simula
system here's another guy I ran into now
I made this really murky this is fact is
not even a picture of Bob Barton because I
think we have determined that cameras
cannot take pictures of him
whose calls goes to walks because he was often
seen in Merrell engineering building but he's never seen entering
or leaving it
in fact he he kind of hated the
whole idea of school and he kind of hated graduate
students and really hated burrows and
but Dave Evans had convinced him
to to leave burrows for a year and be
a professor of sorts at the University
of Utah I'll just tell you one barton
story which is I took this course
called advanced systems design and
he handed out this sheet the first day and
he said well there's something known about this is in 1966
there's something known about advanced system
design and it's all written down here and I expect you all to learn
this and understand it he
said but my job this year
is to firmly disabuse you of any fondly held notions
you might have brought into this classroom in
entire classroom simply garbage collecting
our brains one of the greatest educational experiences
I've ever had he never referred to advanced
systems programming again so the effective
that we would read all of these things his
what he was concerned was was that
none of us could think and there
were many dimensions to his complaints about how
we weren't able to think and during
this year he explored each and every one of them
the result of this and of
course he he hated pretty much everything
that had been done all up
to then in the 60s particularly by IBM and
pretty
anything that we thought might be a good idea he had fifty
five reasons why it was a terrible idea
now you think gee whiz this was
handy but down or taking a course
but you have to realize this guy had the eloquence of
William F Buckley although he was to the left in his political
leanings he was one of the most fascinating people you
can ever imagine running into and he was
extremely well-read excuse his degree
was in mathematics and he was extremely well-read
in philosophy and it was a wonderful experience
now the result of that class was that the students
who were left at the end of it were free
and what I mean by free is he
had completely removed all traces of
religion about every
phase of computer science that
meant there was nothing that we could not there was no thought that we could
not have so the best
professor can ever do instead of putting knowledge into
our mind he removed it
it was the most valuable experience
I had in graduate school because any also
he wouldn't talk about the B 5000 I had to
resort to step on strategies
like talking asking
his wife what it was like when he invent it turned out he invented
the whole thing over one weekend had all
of the ideas he woke up in the middle of the night with all of the ideas for
this machine that just came down it actually fighting
him we found
out he could drink beer problem was it was about six for
a huge guy and he could drink any of us he
could drink australians under the table so
is that we actually approached him and realized one of us
o take about four graduate students
in a row to get him to drink
because you thought was a three to peer state
but eventually
inside so this is a this is one of the great geniuses that
has lived in our in our field completely
unappreciated even today if
you look at what he did in 1961 and 1962
Motorola sells today
I mean it just is pathetic it turns your stomach
turns mine anyway
well there's this pile of stuff on
my desk that turned
out not to be alcohol at all Simula he's an alcohol
that had been doctored to make a Simula and
in the process of finding
we realized it was a programming language that did the
same kinds of things with sequential code as
sketchpad the data structures
were basically very similar and this
has a disadvantage it wasn't as beautiful as the constraint
solving methods of sketchpad but on the other hand you
could do many more things with it because
you weren't limited to what a constraint solver
could do this is a huge
revelation now here's here's one of these opportunities
when you wind up with something new
so data structures and
procedures with a way to program and
if you held on to your pink thoughts
and looked at Simula
as most people did you would wind up inventing
an abstract data types which many people did happen
in the 70s now this is
a really weak thing because an abstract data type is kind of a
slow way of doing what data structures do why
do I say that because an abstract data type
still thinks that you're supposed to use a
generalized assignment statement on its
simulated fields okay and
sign assignment statements programming
is not a very strong way of doing things and so most of the
abstract data type languages have wound up
requiring about the same code to
program things as data structure languages before them
they didn't show this factor of 10 difference
that we found from OOP now there's a different
way of looking at this and it came about
because one of my undergraduate degrees was in molecular biology
and I had read this book a few years before
published in 65 called
the molecular biology of the gene and in
it it had an assay for the first time ever
of a bacterium
ecoli bacterium such as we have
millions of in our stomach this assay is kind of interesting
it's about 70% water but it has
about 120 million organic components
that is things that interact
with each other informational E and if
represent just those forgetting about the water it's about
100 gigabytes as a parallel information
processing system and when I made this slide it
was 200,000
well how about 50,000
to megabyte desktops in
one of these eco lies and it was fast if
you take an atom as being the size
of a thing toys of a tennis ball then
one of these white proteins here has about 5,000
atoms so it's about the size of a Volkswagen and
these guys can move
their own length in about two nanoseconds
now if you scale that up that's
faster than the speed of light
right anybody ever wonder why chemistry
works this is it's really
amazing to me that they never tell
you in in high school so it's actually the
smarter kids that wind up rejecting tan rejecting
chemistry because how can the molecules find each
other it just really doesn't make sense if you actually think about it
the only way it makes sense is
to realize just how much thermal agitation there is as
you go down there for instance
the deceleration on a bacterium in
water is about a million GS
water is like asphalt to
these things they are incredibly strong when
they stop lashing their tail they stop there's
no inertia of any kind here
and the thermal agitation as I say is
fit were scaled up it would be things moving the
side moving eight or ten feet in a couple of
nanoseconds so you can imagine what that is like you can
imagine why biochemistry actually works
that's a pretty complicated thing and furthermore
it's only one five hundredth
e size of the cells in our body and we have between
10 and 100 trillion of them so
you start thinking these cells have
60 billion organic components as opposed
to 120 million and yet
50 cell divisions will make an entire baby
now if you compare this against computing
even today even against the internet with
all the computers on it especially if you think about
it compared to what computing was like back in the 60s you'll
see that this constitutes a different class of organizing
complex dynamic structures they just
aren't like computing machinery at all and
this is what I got from Simula
so
Simula is not a better old thing but almost a
new thing and got this
hit that the one of the great
ideas you could ever have in
doing programming was to organize
that the way biological cells are organized
that is progressed
along so you have this nice idea
of encapsulation which is the main strength
of it you
have an even more weird thing but it's partially from
Simula that you can simulate anything including
structures if you want to go back to them but
in fact you get to deal with many
many more things on a much more natural way
using an architecture that we know from biology scales
incredibly and
you have something that biology doesn't do which is the DNA
is not in the cells here but it's extract
get out and so there's only one copy of the DNA which if
you twitch it everybody is going to feel
that so that was a huge that was the last that
happened to be in November 66 it was the last time I ever thought
about doing procedural and data
structure programming I just realized that that was
a that was just an artifact of the past it meant nothing had
nothing to do with the Destiny's computers in any way
so
then I saw let's see if this works
now this
thing is gone then angle Bart came
to Utah I
mean those AZ
travel was a projector you could stop and start because people weren't used
to seeing cursors so stop
the Sony say I'd like to go to produce but
I'd like good
produce a good thing I'd like to say one branch only
and you know just that little
oh I can say
I'd like to see one line or I
could see but there's a little thing I can
move does remember place that I found here
so here I'm afraid I'll need a different picture
than you so here's I do with
picture drawing capability here so slight map if I start
from work here's the route I seem to have to go to to pick
up all the materials and that's my plan for getting home tonight
but if I want to I can say the library
what am I supposed to take up there I can just point to
that and all the overdue books
in all while there was a statement there with that name go
back what if I want my supposed to pick up the drugstore
I see you understand all right
target can do
things if I wanted to say I'd like to interchange partisan
cans materials jingle
they're all numbered right if I care to know
they're changing them very quickly can subpoena
interchange the products
they do it in August we numbered
so it's pretty pretty
sickening to realize that this was done on 192k
machine that was time shared you notice the response
ever wish your pc or macintosh
would have that response well
you can't they're too fast
because they're too fast nobody bothers
program them in a way that humans could use them but the
Engelbart people really understood that this was
all about close interaction and so they really paid attention to
a lot of these things this is a beautiful system in so many
ways and I don't know their
angle Bart has been here to talk but he
really deserves to say more about all
they did compete they did conferencing they had programmers
up in Oregon that worked with the program was done in the Monroe Park
things that people are still trying to do today he
did very successfully in the late 60s
well I made a huge impression on me and
around that time I got involved with a
engineer who
wanted to do a little machine and together we did this thing
called the Flex machine this is a picture of it on its own display had
the first object-oriented
language I designed ran on this machine directly in the hardware
so he made many mistakes on this
machine one of them was just trying to scale the B 5,000
down in history we learned from that experience that
as Barton used to say good ideas don't often
scale he
had one of the earliest clipping dividing
window system and multiple
windows and a bunch of things so windows
current idea back
then so
another thing I saw early
on was this system at RAND
Corporation the same time
Lee look they match
go back
Iran had invented the tablet
the same year the Mouse's invented 1964 they
built this system whoops
first we erase it go arrow then
connector out of the way so that we may draw a box
in its place seesee wants a box and
makes one form now it's recognizing his printing
the printing in the box is being used as commentary
only in this case the box is
slightly too large so we may change its size that's
where modern-day window control came from literally Android
flow from the collector to the Box
attached a
decision element to the box and draw a flow from
it to scan we then
erase the flow arrows attached to the process post new
area and move the box to a new position
this allows us to
draw a new box
okay
nice huh so
one of these days I have the whole movie and Dave remembers
well I used to show it to the group every few months of drugs
part because what we wanted was something that nice
I showed that movie
at Apple when they were doing the newton and said you
know people have already done a really great hand
character recognizer why don't you use it no no
we're going to recognize handwriting I said well you can't recognize
handwriting because people can't even recognize their own handwriting
half the time they
said well go corporation is doing it I said they can't be doing it
it's not possible once you do this this
a great recognizer done by Gabe groaner
written up in 1966 and this system
was around 1968 or I was 68 when I saw
it today
I can't talk about it more it's a beautiful first modeless
system truly beautiful in every way
and right around the same that
summer I had seen a little square inch of glass
with a few blinking
lights on it little neon lights it was the first flat
display at the University of Illinois then
in the fall I visited Seymour Papert
for the first time and got an enormous
enormous life-changing
insight and the insight
was that the computer is not a tool the
computer is not a vehicle that's the way we were thinking about it angle
Bart was sort of thinking about it as a Model T compared
to IBM's railroads it's not a bad metaphor personal
computing but what Seymour was doing
with children had no interesting analogy
through things like tools or it's
a vehicle the only analogy it has it could work
at all was an analogy to the printing press
because what it was was a vehicle for
representing powerful ideas the powerful
ideas of our civilization in a concrete form
the children could actually learn not
in an abstract form or the abstract form
was actually a concrete way of
dealing with these things and on the I'll
never forget that day and on the
plane coming back from MIT
I designed this machine and made
a cardboard model of it when I got got back
and the lady became called the Dynabook because the kids
computer can't work on a desk kids
are mobile kids have to go places
and I got there many things we did this
was hollow so you could fill it up with lead pellets and see how heavy it
was would get before anybody didn't
want to carry it anymore it turns out back in those days
before people had
to buy these things because they work for companies
the limit was about two it was about two pounds
about one kilogram beyond
doubt is too heavy I think that's still a good figure today
two pounds and the
other thing we discovered of all the different ways of making one
of these things the main thing you wanted to do is
to have it be really slim and at least one dimension the reason
was you always have to carry other things too
the portability became defined
as meaning you could carry other things
too we realized that this
was a different kind of thing than the desktop computer
we thought about how good would you have to make it
to do mundane things on it like we do on
paper paper you can write the Declaration of Independence on it
but you also put grocery lists on so
f the imagines things we actually did was to carry this
in to a supermarket and in Salt
Lake City and imagine what it would be like to
have your grocery list on this thing and you have to walk out
with two bags of groceries Plus this thing
so those are the kinds of thoughts in the
late 60s about this
machine which is called a kiddie computer back then
and because I had a philosophical turn
in mind as well thinking
stuff while I was trying to write my thesis I
realized that time
sharing systems desktop computers and whatever
this thing was actually constituted
three different kinds of information systems and
it reminded me of something I've seen before
which is in the history of the book
so these manuscripts cost
about two million dollars typical
ibrary would be about four times the size of this room and
each book would have its own desk the
books were chained to the desk they often have precious jewels
on them because the jewels were nothing compared
to the cost of one of these books often took as long as ten to fifteen
years to make one of these books
give you an example there were only 371
books in the Vatican Library in the Year 1400 it's
one of the largest libraries in Europe the
richest man in France who is the brother of
the King John Newbery
had 154 books when he died
and he was a bibliophile he had these books made for him and
one of the best twenty-five dollars you can ever spend is to
get a facsimile of one of the books he had made
for himself which is called the Book of Hours of John
gusta berry it's marvelous and
it gives you a bit of a sense of what these books
were like and also what is lost by simply trying to photograph
them they're renderings
and capturing of the charisma
of an oral basically oral society
the Gutenberg Bibles
were intended to look like these guys
that was sort of the connection I made was oh yeah nobody
computer should look like so the best
computing back then is to make it look like a time-sharing
terminal without the central mainframe
but and it's
just what a Gutenberg Bible is is the same size Gutenberg
had 253 characters in his font
just so he could imitate every ligature
and abbreviation the the medieval scribes use
they had people come in and illuminate these books
and the Nuremberg Book Fair in
1454 20 of these books were
shown for the first time and they marveled
of course of how alike these books were
it was a new idea that the
errors in them were the same and
a place that was right was going to be right in all of
the books that was a displacing kind of idea
and they were cheap not cheap by our standards
but compared to these things there are only
three years of a clerks wages that's
about sixty thousand dollars so this is a kind of a workstation
and if you think about that it
fit well into what these machines friends
was the link the handmade link was about a twenty thousand
dollar machine and then
about fifty years later came the great
printing was a guy by the name of Elvis whose last
name was not pagemaker
that's
where the company got their ideas that last name was Manutius or
if you like colloquialisms which most
people hate his real name was Aldo minutiae
oh because thought of his eldest
today and he made books through the
size books that we use today
books are like this size are a little bit
bigger and he made them that size because
the streets of Venice and measured people's saddlebags
because the first offering from his
press was called the portable library and what
he printed was not Bibles but everything the Greeks and
the Romans ever wrote he and his sons went
everything that had ever been written they
published forty thousand different titles in the thirty years
press existed forty thousand
different titles each press run was about
five thousand books or so now
even though I'm running out of time it's
it's worthwhile just pausing on one thing to ask the
group here if anybody knows what the actual inventing
invention of printing was
anybody know
what was the f no movable-type was invented by the Chinese
in the 1st or 2nd century AD
yes pardon no
Chinese invented paper
- yes Paul yes
ok
here's the here's the way it worked the Chinese
invented all his stuff centuries centuries
before thousands of years before
they had the whole rigmarole they knew how to
- they had presses that would
press these things together they had movable typing
everything the problem was it took forever to get
a font up and it wasn't just because of the Chinese
characters so you need to 4,000 of them
think about how many you need how many little things you
need to print a book so the secret
here is that Gutenberg was a goldsmith and if you've
ever worked the gold it's a soft metal you can work it with
other metals and so
Gutenberg winks through the following thought process at some point
when he realized that he could make
a steel die that
was the character as he wanted that character
that steel die was tough enough
to just strike negative impressions
of itself in copper or brass
that copper or brass could
then be poured with metal softer
than it like lead and so a standard
type maker in those days could make between 2,000
and 4,000 pieces of type a day
this is why the printing press spread and you
should think about the connection of this to Moore's
law right the printing press never
would have happened the way it did if it hadn't been for
the proliferation of what served as the bit you
had to be able to have these bits spread out exponentially
so that anybody could make these things fairly easily
and in great quantity well
that led to a bunch
of different thoughts because the belief
system for each one of these areas are completely
different and in fact in history what
happened between here and here didn't really count
that happens to
be where we are in computing today nothing
that is happening between personal computing in
this next phase really camps it's all about
this stuff almost everybody
- today 100 million
personal computers in the world almost everybody today is not
using hyperlinks almost
everybody today is not simulating those are the two
things that computers are doubt but what almost everybody today
are doing is simulating the older culture
the paper culture and this is a recapitulation
of what happened back then
so there many interesting things so
one way of transforming this into a
thinking way of thinking is
to think of each one of these as a belief system I
call them the institutional belief system the personal belief system
in the intimate belief system and you
can play games writing
different properties of these
just the main ones we were thinking about back
then where this one is going to be driven by the integrated circuit
and this one would be driven by Wireless many
people do not realize but I hope you get Larry Roberts
here some time to realize that his plan for
doing the ARPANET was not just to wind up with something like
the internet but to do most of it Wireless back
in the late 60s in fact when
Bay Area one of the first things I did was to ride around in
a old Ford van that Stanford Research
Institute had SR I had they
had a model 33 teletype in the back of the thing
and this very very expensive
transponder that would
baud but it would do it with Wireless you could drive anywhere
and be logged into angle Bart's STS
940 and Larry Roberts
had a very elaborate plan that
he tried to implement before art Bo got shut down too
give everybody a handheld terminal and do
everything via the power of the network and
if you think about that as a model for third world countries especially
who can't afford desktop machines it's a really good one
he wrote many papers about
this and it's a very interesting side story here
well I have
to be careful with this one because this has been improved
over the years we've gone back to this many many
times Larry Tesler did a really great version
of this at Apple a few years ago everybody
has their own version most important ideas just
thinking of these as being really different and trying
to understand what is different about them
the
other thing that was going on around 1969 was
Ivan Sutherland had come to Utah where
I still was and he brought with
him the first virtual reality helmet
here's an example of somebody grabbing you
could grab things and move them around and
we realized that it wasn't going to be too long
before this would be an alternative way of thinking
about a Dynabook and in fact it
lot of sense when you're riding on an airplane to
be able to have a virtual display
rather than or rather than a real physical one
another
person I met back
then I have two slides one that this is one of his favorite comments
life is complex
for this occasion the actual should
have been I have a plane to catch
he was out visiting Utah and there's
long I think it was it had something to do with the I
was on several ARPANET committee so I think it was an architect
ARPANET committee saying he was that he had to give
a talk and a drag on dragged on and finally he
had prepared this our talk and he only had 15 minutes before
he had to go to the plane so it gave the entire talk in
15 minutes and it was just the best
talk I've ever heard in my life you know he's
done it for it now maybe 1.1
or 1.2 Lamson is a little faster than usually
talks and it could not have been clearer
and it's all about capability operating
systems and
one
the things I gleaned from this and this is this is my diagram not
Butler's but
they did they weren't it wasn't
really an object-oriented a capability system is darn
close to an object system so if you're thinking about
different kinds of servers
that each object is a kind of a server then
one of the things you realize is a lot
of the services they're going to give have to be generic and
that was in this system and that
led to this realization
that you have a kind of
a matrix between generic behaviors
are your basically your powerful ideas your concepts and
then you have lots of different variations of types or
object classes and
that made me realize that OOP could
actually serve four you could actually build an operating system
out of OOP impact back you probably should because
you wouldn't have to have a lot of artificial distinctions that
are in there and these capability mechanisms
would actually give you protection and limit access and
and so forth the biggest problem with them was
that the typical operating system process back then
had an overhead of several K bytes to
make one and so you couldn't make things little things like numbers
out of this technique so one
of the things that stuck in my mind is we should find a way of making objects
any size without incurring an
overhead
ok another great book by Martin
Minsky my
other degree is in mathematics so I happen
to enjoy this and this is the way I found out about Lisp
because Marvin showed
how gurdial numbers work and in order to illustrate
it in here he actually built a lisp out of girdle
numbers which essentially what GERD L did if you know that
thing of and it was an interesting idea oh yeah
you can use it sir you can actually
use integers to represent data structures because
you can factor them back out and since Marvin and McCarthy
had formed the AI project together and
Marvin d-list they actually did a lisp like language in
there so that's really interesting maybe I should look at Lisp
it was kind
of looked horrible and I discovered this wonderful book
which I have about 20 copies of discovered
over the years that when I need an idea I
usually want it to come out this nice
so I go out and buy a fresh copy of this book and
I when MIT press quit printing
it I'll have to retire they
seem to use up their power
so I have to buy new ones periodically
so however
this let me commend this to you because there's nothing
more beautiful on this earth than page the bottom
of page 13 okay
and there's nothing nicer than he only
takes 13 pages to get to one of
ideas of all time which I think of as Maxwell this
is something like Maxwell's equation this is like everything
this is everything now
of course Turing machines are everything but what's great about this
is it's a Turing machine with slope you
don't have to write tons and tons of code to get it to meant anything so
Universal machine that's small and
grow up like this when I finally went everybody
here in this room remembers who
went through this thing remembers the day they trace this thing putting
their fingers down and at some point you realize
holy this is everything this
is it if
you know if you don't get that experience you've missed
one of the great things in computer science because in it to
me it divides the world up into two different classes of people
the people who think that you
have to write endless amounts of grubby code to do anything and
the people who realize that what you should be spending
most of your time on is the meta system and let the meta system
create the rest of the world for you
so that had a huge
effect on me I think what I'm
going to do is skip the next tape in the instant
of a very funny tape called LSD crocotta
chip and the magical
mystery time sharing machine which was made at
the AI project which is where I went next
and will will refer it to another time it's basically a
nice example of what people were doing that
shows one of the first modeless text editors done on the PDP
one day I'll just pass
on that so I
started I wanted to make one of these machines and of course the
screen display was a way off so I thought
I'll get myself a Sony tummy china chime and
I'll just wedge in there what a link has
and I'll get myself a
modern version of the link for for kids so this
is the thing called a kiddy comp that I
started designing then I wound up
at Xerox PARC and the
chief scientist of
a research center and they were smart enough to hire Bob
Taylor to set it up and Taylor knew
every good PhD in the country because he had paid
for their PhD and he started his
siren song of gathering
together a really fantastic group of people which
included Butler Lampson occluded
number of the people in this in this room and
that five years is the most
fun I've had in a five year
period so I went
saying okay what we really need is something that looks
like this this was 1971 and this
was called Vinnie calm and
we need I said we
need to make about 25 or 30 of these things so we can start working
with with children we
did Dick Schaap and I did some of the display hardware
for this machine and then they got turned
down by a guy who should
have known better he was known in
the in the papers executive X
meanwhile other things were going on we realized that
because these kinds of machines weren't going to be very
portable that we need laser printing and Gary Starkweather
was there to do it he is already
working on a laser printer when I went to Xerox in 1971
and got it working late that
year and Butler
Lampson had this meeting and
said you know we've done so many bubblegum things
in the 60s what
we should do at Xerox PARC is take an oath that will never do anything
that isn't engineered for 100 users so
this was a novel thought his
Butler's idea it was probably the greatest reason
that Park was a success was not that
it was so far out but because it was actually incredibly conservative
because of that we actually wound up doing things
that we'd already done in a half-assed way in
such a way that we could build a hundred of them as there were personal
computers or run a hundred users if it was a time sharing
system there was a huge huge thing
to this idea which today is called living labs
so the idea is you
live in your own creations don't just
demo it make enough for everybody but the programming
language it has to be documented and solid enough
for 100 people to use it so this became a pretty
general principle for most everything that was was
done in the first five years of Xerox PARC and the
most amazing thing about it astounding
thing about it was that with rare exceptions
it actually made things happen faster
now everybody knows why
right because that extra thought
that you did when you were thinking oh yeah I have to engineer
this 400 users prevented those bugs
that add years onto the quick and
dirty thing so in fact the stuffs
worked just it was unbelievable and I
won't go through the because it a
very interesting talk is about Xerox
PARC in those days I'm not going to do
that but it was a fascinating process so
around then there are the ideas we
have to do a Dinah book notebook kind of thing first
and then eventually we
realized that the flat panel displays it actually be easier to make if
they were small because of the contamination problems
and that there would be eventually something
like this that you'd wear on your your head one
the reasons we didn't go much further with that is that the the
lightest head sensing device in those
days weighed about four and a half pounds and
Nicholas Negroponte II had been talking about
simply wearing your computer so
they have something like a wristwatch that knew where your hand was
and when you walk from one room to another it would know that
one room to another this today
is called ubiquitous computing and Nikolas has never gotten as far as I
can tell a shred of credit but in fact this was his
idea and it was an early idea of his
so oh yeah
so let me do we have
yeah tape tape
for what we want
now again
this in the spirit of his being historical just
by coincidence I gave a talk in November
72 I
think this is the right order
to do this I hope
November 72
to a conference which happen to be videotaped
Xerox PARC that
tape got lost was
not in any of the holdings that we had
I saw an excerpt from that tape at
a thing actually Paul Sappho and I
an angle Bart did in Japan and I
asked the Japanese how did you get that he said
we've had this for years so I asked well can I have a copy
of it please and so the only known copy of
how they got a hold of this damn thing I have no idea but here's
here's kind of the way I talked about this
I'm sorry there's
kind of way I
talked about this in
1972
and perhaps lower so let's
sort of see what it is you've already got that on we have
the lights down and I'll be able to cook this I don't
drink so okay
well this is
an existing device something that you a queue
a Packard calls a pocket electronic slide rule
and it
urns out that the way they decided to make it is so
that would fit into Bill Hewlett shirt pocket
that was the way it was was designed
it had to do everything the slide rule would do except to
ten places they actually spent
almost the next three year
on this thing to keep it to go to size specifications
it has batteries and it can indeed should
just positive for a second I found out later that
they actually retailored
Hewlett shirt a number of times
this is
really and the guy who can tell that story is
Tom Osborne who did the HP 35 if you remember
that I knew it was it I got one of the very first one because
knew Tom Osborne and I had it for two weeks and
has a wonderful
experience with it and then it got stolen and that
was when I realized that the new era had happened
you could actually steal a computer now
he used on the gun on the grass
the power that it has and the size
portability and everything else make
completely different so different that Hewlett Packard expects
million of these in the next five years people
who would never buy a calculating machine of any kind buy
these things for $400
okay so quantitative changes if you make the margin
of our qualitative changes
here's our conception of this gadget
called Dynabook also designed to be used in the graph
we don't have to worry about what's
in it it suffice it to say that we designed
the outside package first we
want it to be no larger than a standard
notebook it's about nine by twelve and
we try to make
these specifications for it stylistic once once
it had to do with the kind of quality that we wanted with
the idea that we would try to beat the
technology needs to
be beaten totally in the submission in order to fulfill
the inside of the gadget and that in fact is what we
are doing so the idea here is
the only idea here is that
it's supposed to
be something like active paper okay
and it's not supposed to be worse than paper
in any reasonable way one of the problems
with modern technology they always come up with something worse
we wanted to be able to handle
things dynamically rather than statically
the way paper did but it had no at no
real loss and quality so basically
display in which you can see things you can think of as being like
a television display means for
entering things that has a you can barely see it away
for entering drawings and it has a removable hunk
of storage that will store one 500
page book okay or million characters runs
on batteries and you can carry it with its portable by my definition
of portability okay which is that
you can carry something else too
well tell you
what you can do with it in order
to check out a few things we decided to simulate it using
current computer technology so everything
oing to see from now on and everything all the pictures in the paper are
actual photographs taken
at our lab okay the simulation is in
some sense a real simulation
okay here's an example of the
kind of display for text now
because this is an active medium we can have any font
that we want in text files
for instance we have almost all of this book
typed in we can look at this
piece of text using
any font the we that we wish and Fox of
course have a great cloth coat so it goes goes on
and on about
why you should be able to read from the
from it so here's
the the old character generator
as it was known and
just for reference here is
making a capital to the 20
point Lydian cursive font and
the page of text and the Lydian cursive back then
look like this so it actually compares
quite favorably with anything today
we overdid practically everything i scary
Stark withers first laser printer ever ran
at a page a second at 500 pixels to
the inch as a wonderful
story as he tells it he
needed to get some sort of laser printer from Xerox and
the only ones that they weren't selling back then happened
to be the page a second one and the racks
3600 and he tried to slow it down but mechanically
and inertia wise it was just set up to run at one speed
so he had to speed up the laser so
he could actually so he actually the
first thing ever had a 24,000 rpm
rotating mirror and laser
is beautiful this this machine is
beautiful
okay another thing we're thinking about I'm
gonna have to go a little bit faster now the other thing we're thinking about
is psychological models again
this deeply because I've talked about before Piaget
has a stage theory you go through a kinesthetic
stage you have an iconic stage where
there's more water in the tall glass and later
on around the age of
13 or 14 you enter a symbolic phase where
ason or at least in swiss-french children of
that age no
evidence that American teenagers ever achieve
this stage
still perhaps the greatest single
book ever written on this stuff having
everything to do with the way people learn is
Jerome Brunner's towards a theory of instruction
on October
13th I'm
going to be attending drum burners 80th birthday party in
New York City
this is one of the this is just a
book and he's a great writer as
many wonderful things most of the deep
insights we got about designing user interfaces
came one way or another from this book
so his experiment was to take the kid who said
there was more water in the tall glass and
to put a shield and
discovered to his surprise that the kid changed his mind
said as soon as you prevented the kid from looking at
say oh another there can't be more waters
and what he would do is then
take it away and the kid says no there's more
water put it in there the kid says no there
can't be more water so if you have any six-year-olds you want to torment
so
burners one of burners many conclusions
was that the Piaget way of thinking
about things did indeed have some stage dependence
to it but much more importantly that these mentalities
our separate ways of thinking it's not a
metamorphosis like a caterpillar into a butterfly it's
multiple ways of knowing the world they're
operating at the same time and the development
of children tends to turn on the
dominance in this order but in fact by various
can show that the other ones are there you can make use of them and
that if you think about it for user interface
of the most powerful ideas to think about because the stage
theory doesn't help you much it kind of says you should wind
up with ms-dos if they're over 14 but
in fact if you
look at Brenner's idea it says wow you
should try and get synergy between these different ways of
knowing the world people know the world simultaneously in more
than one way they can remember and so forth in
more than one way so for instance everybody here
has had the experience of being able to remember a route better
if they were the one just driving the car rather than if they're a passenger
everybody here has has the experience of clicking
from one channel to another on cable
TV into a movie they haven't seen for 20 years and
being able to recognize that movie in just a couple of seconds right
ok these are properties of
these different mentalities and they're quite
incredible and can be exploited so
we were thinking about this stuff back then
iconic thinking burners
middle stage so here's
an easy one that the area of the triangle is 1/2
the base times the height but if
you combine it with a little bit of logic you
it's general for any triangle because you can divide things
up this is the power of Greek thought
he Greeks realized you could take things apart and
put them back together again and you could by simply
the triangle in half you could get two cases and therefore
the rule had to be general
fact
many of these Dave Dave Smith's
remembers all of these things we used to
look at so the iconic ways
thinking about addition and multiplication or
a plus B times a plus D
iconic Lee is quite beautiful as well
here's
a different proof of
Pythagoras theorem which is even nicer than the one that we
used to think about back then because it
says here's the triangle here's the C squared
and look we can put four more three
more triangles in there to make this larger thing and
it looks like there's just enough room for the a squared
and the B squared plus the four triangles bingo
that's what iconic proofs are like and they're wonderful
notice that one of the problems with the iconic proof
is it doesn't tell you why this whole just shows
you that it does that's actually an
incredible much more subtle thing to
try and understand why trying right triangles have to have
property besides just understanding that they do
subtle difference between iconic and
symbolic ways of thinking so early
on we started thinking
about we have to retreat from
symbolic language down to the
iconic down to the kinesthetic because there's a lot of good
stuff down here it's our first
essays at user interface design we're
saying let's let's explore this down
here more the more right-brain or more Bruner
type mentality way of thinking
here's a great book jacques
hadamard psychology of invention in the mathematical field
showed that most of the great mathematicians and
physicists of the world did most of their thinking in terms of
visualizations and
about 30% of them you actually use kinesthetic
sensations Fineman wasn't included in this
by an oversight but he
were two of the most interesting ones that actually used
their body in very strong ways for thinking
about things
we looked
at a programming language called ambit G this
is a bubble sort done as
kind of pattern matching diagrams against data
structures I didn't know Dave Smith was going to be here
but here's what he did if Dave remembers it
of an iconic bubble sort
so the idea
if what he did here is to combine the four and the after
into one diagram where the
before links are
ones and the after ones are the heavy ones and
there's a single comparison in here so the idea is that this holds remember
this so
the whole diet if this whole diagram is is true
then you do what the what's
the the fat
links tell you to tell you to do and that we did actually
quite a number of programs doing
this stuff I'm just
going to skip this the realization that there's more than
one kind of programming language and here's a another metaphor that
you can do beautiful things
in each architecture but the scale
that you can work in depends a lot on what the actual premises
of the architecture are so the Greeks built beautiful buildings because
post and lintels but you could not make them very big Gothic
cathedrals have very similar amount
of material to the Parthenon they're basically
made out of air but you can
make them huge because you have this notion of the vaulted arch and the
flying buttress and of course a geodesic dome you
can cover all the cathedrals made on this earth with
the same amount of steel because steel is more
powerful under tension than it is under compression
Chuck
Thacker
so this came about remember
I told you that the our little mini con thing got turned down by
executive X and turned out that
and Thacker wanted to do it for me anyway so they snuck over
in September 72
and this is what Thacker
said do you have any money and I said well yeah I have about
two hundred and thirty thousand dollars that
o use to make it a couple of these things out of these
and they said well how would you like us to make your little machine for
you and he said I'd like it and
I said what is it and he said
well you want a kiddie comp
that works like a Dynabook I want to build a ten
times faster data general Nova 800
want one that runs at 80 nanoseconds and Butler wants
a $500 pdp-10 so
let's let's do a first shot at that so
he gave them we gave them all the
examples that we had
done and another
thing that here's Dan Ingalls another
thing that happens is we had an argument
in the hall about how powerful you could make programming
languages and I said Oh half
a page half a page you can do the most powerful programming
language in the world because I knew about Lisp and they
said put up or shut up so of course I didn't show them
Lisp I sat down it's been a couple of weeks working out
the actual first evaluator in
itself for small talk and they came out to about a half a
half a page so the
small talk that we got was actually not anything like Dave will remember
was not anything like what we were planning on doing it was just
accident it just happened as
the result of this debt and everything
would have been alright except dan implemented it
in fact when I got back
from that trip where I showed the the video he
already had a version of the thing work to the course he only had
to take a half page description and
he wrote about 700 lines of code on
one of the many computers there and all of a sudden we
had a working small talk and once you have a working language you
just really want to write code in it and it
derailed things although it was the language that they wrote
his thesis in eventually but it actually kind of derailed
what the research was to just have this thing around
on the other hand is probably because we don't
whether the research would have been successful if Dan haven't done this
so maybe I wouldn't be here talking to
you if Dan hadn't decided to implement this
thing because small talk might have been the most interesting thing that we wound
up doing we will never know meanwhile
Thacker had a bet he
bet a case of champagne with bill vidiq it
was a Xerox executive that you could do a real system
in just three months so he started the alto
on November 22nd 1972 and
around April Fool's Day and
1973 the alto started working
one person two
technicians and Ed macwrite did the disc
drivers on the thing so that
genius he could just throw stuff at the wall and it would
fall down machine
so the first picture ever shown on the alto
was this clicking monster which
we done as part of our drawing
experiments and now I
should show what the only just play the next
tape here
ight
of a fabless didn't get Tron
kept changing that change the next one please
just try and show
you some of the things we were doing back
then so while he's changing the thing
one of the summary things I could say here
about this project is the thing
that was really great about this project was that we
didn't have a huge plan but
we did have a huge vision what
I mean by vision is like the how
do you do a plan the best way you write down all the things that
should be and then you erase them and try and smell the perfume
that's left because if you write down all
that should be and then try to do those things you
lose why because you don't haven't learned anything yet
this is why people who do things the plans
usually don't do very interesting things much
better way to do it is to set do
all the writing but set
the direction and you something
much better will happen and I think during
these five years just one thing
after another that got done just turned out beautifully because
there wasn't any we never had to justify the
stuff we weren't spending an enormous amount of
money but on the other hand we didn't have to explain stuff
we couldn't have explained we
couldn't exactly because it was all aesthetic there's some
I mean to the part of the reason for
doing a small talk was just how beautiful it was to
do something that was completely this way let's see what this
yeah so here's the alto
playroom this is the
alcove synthesizing music
the alto could do 12 voices in real time
like a podium all by itself
bottle 33 dispatch
here's the alto animation
I think about that in reference
to what you've seen on machines today
this
is a system done by a child here fifteen
year old the CAD
system I mean because did
you
[Music]
okay stop
okay change the tape please
so
we built a zillion Altos
we connected them with Ethernet
we built five or six pages
econd laser printers and put them on as servers we use the
Altos to construct a file system
that's now today called client-server although
I think we should realize the client-server is actually
a bad model Pierre Pierre is what you really want you
want to have equally powered machines the altar was so
powerful it's the inter inter engine back then was about
a six smith machine you have to realize in
1972 so it was about a factor of 50 more
than you got as somebody logging into a time
sharing system that was me in such an enormous
difference i week today
when I see people trying to develop code that's supposed
to run in the future on today's machines you have to do
so much optimization but we made it because
Alto we didn't have to optimize it all for the first few
years we just wrote the simplest code to try out the
and we're able to iterate and iterate and iterate
and iterate how many people here know of
Paul MacCready and the gossamer Condor
okay here's an interesting story
MacCready never intended to win the Kramer
prize but it happened his brother-in-law had
gotten into $100,000 in debt which McCready
decided to assume and
so all of a sudden McCready was a hundred thousand in the hole he's
driving through Arizona watching a hawk circling
in the sky he suddenly realized that the current rate of exchange
the Kramer the forty
thousand pound Kramer prize for man-powered flight was
exactly 100 thousand dollars
and so as he was driving
along he started thinking about gee I could pay off
his debt if I could win this right now it happened that there's probably
nobody had won this prize for 45 years many
groups large industrial groups had tried over
the years McCready won it within six months of
when he from that realization in the desert
how because he realized
that the biggest problem that the way everybody
else had done it is that they would build an aircraft have
a crash and then spend another six or eight
months rebuilding the aircraft these aircraft are incredibly
elaborate he said what we need is an aircraft that
we can have ten crashes a day now
built one that you could actually fix with scotch tape
it was just made out of mylar in a couple of aluminum
spars and the things started working within
first couple of weeks and within six months he'd won the prize and
we should something we should all think about that beyond
a pertinent beyond a certain level we simply
cannot plan our systems they are smarter
than we are but we have to negotiate with them so
build a system we can't negotiate with and it
gets beyond a certain size we debt here's
a couple more things
from back then so here's the way
kids programming we got them to make a couple of
boxes each of them
instance of class box one is called Joe and one is called Jill
and they're both animating separately and
it gives the kids the idea that you can
have multiple enemy and entities of the same type
here's
what the first small talk class is Marian
Goldeen now one of things that that
is kind of of interest that people here's the
first application she did which was actually
a painting program you'll see she built this
after about I'd
say about a month she had seen some of our
painting program so she decided she would do one of her own so she'd see
she picked up a square brush from the brush
menu up there that she made this
whole thing she has a
different shape there so this is the first known that
we know a person own example of a tool being
done by a child what we call an application today
was done by this 12 year old girl I
should mention for those of you are interested in that in
the 70s because there wasn't anything
about computing the girls were just as interested any of
this boy
this
is next year of completing the class
now 13
and here's
what one of her students did which is something
more like Mack draw now instead of a painting system it's actually
an object-oriented illustration system
so there's a menu down
the menus saying Grosso
is that she's going to make it grow smaller going
move the thing so each one of these things has
its own object identity and each one of them is
made for doing polygons so there's a selection handle
twelve-year-old girl it was a girl by name of Susan Hammett
did this and now
she's going to make it real small and give it a lot of sides so
she'll get a circle
change
the color now she's going to
make a bunch of these to make a truck
so this is a very imposing Aalto that
dick shout did around 1975
so here's more of but
the alto animation stuff was done by
Steve Purcell and it could animate about
a hundred and twenty square inches of graphics at ten frames a
second which is very respectable even today was done
much nicer this machine had double buffering
built into it which is what you know anything about animation
trying to do it on today's machines without animation is
kind of silly here's a here's a system
that some professional animators plus ron becker
did this is about five pages
of small talk so
the cell window is over there the animation window is
here and what he's going to do is now give this
ball that he just drew a path to follow
so it's
following the path now he gets goes
into the iconic menu here and picks up the
step now he's single steps it what
he wants to do is to get to the bottom because
the way if you know anything about animation
the the basic technique is called stretch and squash
so we want to do is to make a squashed
form of the ball down here and
furthermore he wants to do it while the animation is running
something that's quite disappointing about the animation
system today is you have to do this stuff and then run
but notice what he's doing he has another transparent cell
overlapped on that one so he's not
picture he's drawing a separate picture but he's using it as a reference
and notice it's being inserted
in there and now he has to put in the
specular reflection in order to get the
animation effect to work
so
this system was done over a single summer is about a five-page
here's
a movie name called gallic here's
a twelve-year-old girl yet another one it's a horse
galloping so she did a galloping horse in
this animation write a program that has the
horse scalp and the cross escape the
n
she added a feature to this adult program
which is to be able to take any
two things and combine them in
your single camera jockey I
can simply pose the jockey
on horse
to make a racial
then
it becomes a single object to be animated
now I can have the
jockey and the horse run across
the screen screaming by making another program
okay so we're getting the
making progress here so just a couple of other quick
things here's a wysiwyg
retrieval system you
fill in the blanks and it
produces what the early window system looks like
here's a an
early slightly
later version of the window system I think around 1975
multiple
fonts here's an early version of the
browser
browser for debugging this is probably
around 1977 or 78 this is a small
talk 76 thing this was designed by Larry Tesler
here's an
early desktop publishing system called the galley editor
and one of the things you might notice this
doing today this is these are embedded objects
when you put the mouse into one of them it
pops up a halo of its iconic menu
so each one of these things is a different
this is a component based system you can put
any kind of thing in there and it simply it's
like open dock it's like open dock
so that was a nice really nice system
oh no
what happened I need
the other slide tray I'm almost done I
appreciate your patience because when I laid
thing out I found it extremely difficult to
tell a story that was somewhat
like what happened to give you a flavor what
it was like to do this stuff and realizing
that there's a limit to people's patience of
listening to this stuff and looking
at it maybe it's while he's changing
well maybe he's already done it
let's try
oh I know what he's having problems with
now how these slide things work you have to push down the
thing to get to zero or
they once did
yeah so
here's a here's a multiple view system
done by trig V rain Scout this is a planner
I think it's the first one ever done really that
could show you so this is a pert chart
at the top I think a Gantt chart
here a
sequence list here and a
interested this is done by a guy who was one
of the designers of the Norwegian ship building
system and so this is I think the
first sophisticated multiple and you see that's the problem
with Windows the way they're used today is they almost
never use to give alternate views of things simultaneously but
that's the whole goddamn idea is to be able
to look at the same thing in different ways so you can understand it
there's one of the really disappointing things in the
commercialization of this stuff that people were so paper
centric when they went at it
so the slogan for doing the user interface
is this is Bruner's progression kinesthetic
iconic symbolic doing with images make
symbols that everything is rooted in
doing and
this general theory
of user interface which is kind of wide
widely spreads as well you want to do things
with your body in various ways and having
a kinesthetic feedback on the mouse of even better
icons for the reason we're talking about
and you need to have some sort of scripting language underneath
for symbolic underpinning and we
realized that the retreat
back to the iconic and the kinesthetic was
only half the story because the other half is
you want to go this direction as well you want
to recapitulate childhood you want to learn by touching things
eeing but in you can't escape
the need for abstract renderings
of your ideas is very very hard for
instance to make
the argument in Tom Paine common sense
using stained glass windows think
about it it just doesn't work television
is no good for this either so
some very very important arguments need symbolic
renderings in order to in order to actually
work and so here's the whole user
interface idea and so in the dust
settled around 1974
7576 we have this
complete system which of course Xerox turned down
as a as a product idea
but that's a different story it's not the story of
and of course we had
to take the Altos out to schools and it turned out
we had to steal the first three Altos out of Xerox PARC
because in by the time it came time to take
the Altos out into the schools Xerox didn't want us to take the
Altos out of the building anymore so Steve
wire and I loaded several of them a
guard had never seen anybody steal a computer
before so instead
underhanded about we simply put the things on trucks and
the old station where I can put them on there it's very easy
then we built this
machine this is 1978 now this
a note-taker this machine has three
8086 ins in it and a small talk
that now runs in under
basically had 512 K memory and
several people dug Fairbairn
here who did it Larry Tesler and
I and I think a couple about it we build about
ten of these machines had the experience of using them at
least once in an airport
they weighed a fair amount and
so these were the machines
that we actually hoped that Xerox would put
out machine
kind of look kind of like an Osborn don't they except they're about five
years earlier a very nice
machine actually so
I think it's a good
place to end with question about the Dynabook is
there still a use a reason
for the Dynabook now that we've had commercialization
personal computing has been a howling success
many people have made lots of money on it
hundreds of billions of people are using it and
I would say yeah the the worst
thing we could possibly do
here is to
continue the trend of just trying to stuff
desktop computers into these things we know
these are going to get down to a couple of pounds very
very shortly but the problem is they're never going to be a Dynabook
unless their major reason
for being is to help people to think
they're right so having a printing press is not
fulfilled if all you print is comic books all
your print is pornography if all your print is money
we're in a printing press was fulfilled by
printing things like
this and things like this
is my favorite meta program the Constitution
of the United States think about it
you can put it on a single sheet of paper here
it's just a few tiny little papers
and it has been the controlling force for
a an organism made up of millions
and millions of mutually conflicting parts
and is not broken
for more than 200 years nobody
has ever built an organization that
complex and this and the only reason it
works is because this is not a set of laws but
a set of principles the only way
we can get people to understand this
is I believe is to have people
understand the difference between laws which
makers obviously don't because they keep on making them
this is case based reasoning of
the worst traditional society way of
doing things see the read why do I say that because you
know King Solomon was the wisest man in
the Bible and it even says why he knew more
than 3,000 proverbs the
way proverbs work if you come home from a trip and your
spouse is glad to see you then
the reason is absence made the heart grow fonder
but if you come home from a trip and your spouse
particularly glad to see you then what is the reason out
of sight out of mind so
oral societies obviously do not list their
Proverbs because they're almost societies
and proverbs are basically stories that
are evaluated the way stories are you never worry about that the movie
you see today contradicts the movie you saw last week that's
they're about they're about how good they are now
so this is not
a story and this
is not a story
now the problem is unless we learn these
other ways of thinking the ways of logic
and the ways of systems thinking our natural
biological tendencies are to grow up and tuck
in entirely in terms of a narrative television
reinforces growing up in terms of
a narrative and this isn't probably the most
disastrous thing that's happened because we are not only getting
lead meta evil eyes we are actually the
world is now descending back
to traditional societies and ways of thinking about
things which is if a story is good
that's all I need to be convinced now
of course we don't want to get rid of this because
that would be going to a theater and never being able to see the
play right you just see people you'd never give
yourself over to this thing but think of the people who go
to the exact same format as
the theater called a political
rally music pageantry
fine words and give themselves over
to it that can't be right so
these ways of thinking about the world are incredibly
important to learn and to learn when we're
going to use one form of thinking or another and
I believe that the Seymour
Papert was right in the beginning that
the strongest ways of learning these new ways to think
these inventions of the Greeks and of
the 17th century people in the 19th
century people of 20th century people are to actually be
able to build things using each of these ways
knowing the world so from that standpoint I
think the world is more desperately in need of a dining
than ever and one
of things that be fun to see in the next few years is
hardware of the Dynabook because that was a lead-pipe
cinch from Moore's law 30 years ago
but to actually see the software the
Dynabook actually needs thank
[Applause]