LINC Twentieth Anniversary Symposium (1983)

From Viewpoints Intelligent Archive
Jump to: navigation, search
now pass on the touches of the link on
contemporary computer technology mr.
Alan Kay chief scientist at Arak corporate research
well and I got called on the phone told
dr. Waxman that I couldn't think of anything I'd rather do then come to a
celebration which I think of this as rather than a testimonial of the what I
think of as some of the best design done in our field I am a great fan of good
old things and it is very often the case
that the first of a good old thing has a lot more of the essence of what it
should be then subsequent work for a variety of different reasons when I was
a graduate student in the late 60s I
first met Wes and I remember I I was at
that time working as a consultant for a company that was doing the Machine it
was probably the second personal computer ever done and at that time I
being a graduate student of course all things were possible and I knew everything
I remember being rather discomforted by
some of the questions that Wes asked me about this wonderful machine I was
designing and in fact I realized some
years later that if I'd had better answers to his questions that the machine would have turned out better so
when I was thinking about design I was
struck yesterday yesterday afternoon I went over to All Souls Church in the
district and played organ for a couple of hours in fact that's why I missed the banquet I played much longer than I
thought and it was dark and I did not have the instructions of where the banquet was and it was too late to call
anybody and so in fact I came out here last night this instrument I played on
is sort of an interesting example of design it was built about six or seven
years ago the action on the organ is
completely mechanical a reversion of pipe organ design back to a style
originated in the 12th and 13th centuries so one of the things that the organ
designers over the years discovered that after going to electric actions and
pneumatic actions and all that that the old mechanical design was the best
because the organist can feel the pipe bells opening through his fingers it's a
more sensitive musical instrument the other problem you have in an organ is
you have to change lots of stops this instrument had about 60 different stops
and the designers of this organ had instead of using mechanical stop
changing had a complete computer memory system to allow all the stock
combinations to be set up and changed by just the touch of a button so here we
have 20th century technology which does
not add you can't hear the stops being changed except for the change in the
timbre of the instrument and I thought well that's that is tremendous here is
the best of the old and the best of the new then I discovery one horrible thing
when I sat down to play and that is that the bench was too tall yeah I have I
have slightly shorter legs than the most people particularly from the knees down
and I spent the rest of the afternoon being for several hours being acutely
aware of just how high off the pedal
board my feet were instead of enjoying the music and I like I'd like to say
that I think the best thing you can ever say about a designer and this is definitely true about Wes is that Wes
designs the bench well too
and I think that is what design is all
about because design is really the art of providing a package that provides a
service to the user whether the user is
a person who wants to live in a building and the designers an architect or the
user is a computer user who wants to do
something with a computer the I think one of the reasons why the link turned
out so well is that it was not designed by computer people for computer people
or design to just do a computer it actually had an outside purpose that
helped I think a lot of the design decisions along the way so why it's also
often the case that some of the best computer graphics that's done is not
done by people in the field but by people who actually have pictures they
want to look at so they naturally have a much stronger interest in the kinds of
detail that can be put there and the aesthetics there by the the tape drive
on the link reminds me of my favorite
rubber band story which is the one when
I was at Xerox the one of the system's
programmers that worked for me bought through the whole earth catalogue a gadget called a hydraulic ram pump which
apparently has been around there's one company that makes these and they're made out of cast iron and they've been
around I guess for 130 or 140 years and
what it is is a little metal gadget that you stick in a stream and the water
coming down through the stream eventually put some compresses enough
air in a hemispherical dome and this Ram pump that another that a valve opens and
some water is pumped uphill okay so this
is a pump that will you can water your garden with and the instructions are
very interesting on this humph there's one line that says and
under the maintenance every 40 years please change the leather gasket and
when I saw that thing I said holy Christ
today we can't even design a flush toilet without 20,000 transistors here's
this marvelous machine that has one moving part and one thing that you have
to change every 40 years that reminded
me of a person who was mentioned today also is one of my a guy I never knew but
person I'm a big fan of is gray Walter the old and secular Griffo who built
these cybernetic Turtles he wrote a book in 1951 called the living brain and
which among other things he talks about how he designed these little cybernetic
animals that would go around and be attracted to light and go into courting
dances with each other and so forth and the thing that drove the design of these
little turtles was the fact that he only
had enough power to power two vacuum tubes that's all these things contained
so he actually had to sit down and think about what this thing was going to do
when I read that I thought gee that's that's something that we don't very
often do anymore because we just slather on like putting butter on an English
muffin a few more thousand bites whenever we want to do something but we rarely sit
down and think and design so to me what
the link is all about is I think one of
the three best jobs of design in the early 60s my life these are my three
favorite systems the link Ivan Sutherlands sketchpad and a system
called jaws I won't bother explaining what they are except to say that jaws
was probably the best interactive system pound-for-pound that has ever been done
if you look at total attention to the user
shades of jaws are in all interactive systems today it was a system however
that was remarkably unappreciated for its time since it actually appeared a year before
basic and is nowhere to be seen now because it couldn't be recognized
I think recognition is the hardest thing for people when they're presented with
something new Dec had the most difficult time
recognizing what the link was even with
deference to the remarks of the people from Dec today and so forth the I can
tell you that when I was at a graduate student at the University of Utah we thought about buying some of the Dec
versions of the link and rejected them because they weren't enough of the link
ideas actually in them they in fact threw away the whole integration of the
of the system that was one of the hearts of the link and I think all the link
link users know that one of the great things about it like let's say a
Japanese Honda car is that you didn't
have to decide whether you wanted the hood latch down there by your left hands
the link designers put the hood latch in before you knew that that was something
that you actually wanted all of the things that were important to have as a
total system where delivered to you so what you got was neither a piece of
hardware nor a piece of software but a thing that enacted a service it took me
a long time to realize that oh we have the lights down I'll show you brought
along my little slide the collection
show you a little history
well this is a machine that this is
called the Flex machine and this was a machine that was designed and built in
the period 1967 to 69 and inside the
under the hood there were about 700 ICS
and this machine had about half the computing power of Burroughs be 5000 in
fact it was an attempt to scale down a mainframe computer that directly
executed higher level language code and I learned a lot from that experience
this is actually done as a as a project
for a company the pointing device there
is not an option like the link this machine had no optional features on it
the optional features were that you got to make it do anything you wanted it to but it was an attempt to be a complete
self-contained system for professionals in their field we thought of people like
doctors and lawyers and and so forth
high resolution calligraphic display and
perhaps the first machine to have be ever called a personal computer now the
reason I'm showing this is because this machine actually turned out to be a
failure and when I was doing it I knew
that the link existed and I sort of knew what it was and my my initial hand wave
at the link was to say well hell it's just execute smih sheen code and doesn't
have enough memory and it's to to piddling really small to do what we want
to do what we really need is some power here so we built in a lot of power and
it turned out that the users were repelled by it the fact that had one of
the first window interfaces that was ever done what the users are still repelled by it that we tried out these
things on and it took a few years later
for me to actually understand why and that is that the
what we weren't giving the users was
something that was integrated and user friendly we're simply giving the users something powerful but in which even in
this higher level language they were able to do much less than the lower
level language on the link because the particular way of thinking that they had
to use to use our language was much more foreign than using the link assembly
code so what's the I guess I have a thing here
so now 1968
while we're still working on that other machine I saw at the University of
Illinois the first few little lights on
the first flat panel display called the plasma panel that was done there and I
realized that it wouldn't be long until you could build a computer in which the
only part of it that you saw was the part that did something for you so one
of the ways of thinking about a computer is its maximum size should be the air
the area that subtends the display and you don't want to see anything else so
this idea which is now famous as the Dynabook came to came to life and in
fact when when I went to Xerox this became the project that that I worked on
there with my group so the idea here is a personal personal computer as I
defined it then was something that was so portable that you could carry
something else to 0.5 herniation x' per
block does not count and if you have to
plug it into the wall then you have to think seriously about it because after
all you don't have to plug paper into the wall and you don't have to pull like a pencil into the wall you should be
able to use it out in the grass we thought that the display would be one of
the an eventual liquid crystal display a reflecting display and that if we needed
a keyboard we could display it on the display and use a touch sensitive
surface but that most of the time we wouldn't need a keyboard we just use
pointing operations and so this became one of the three dominant metaphors that
went into the personal computing work done at Xerox PARC and in fact the
this is the machine that we eventually built which incidentally has its 10th
anniversary this year it started working in April 1973 and this machine is a
spiritual D descendant of both the link and some of West Clarke's other ideas
now I have to tell you these these
little technical stories are are kind of perhaps sometimes boring but in between
the time I worked on the Flex machine and the time that we did this machine
one of the things that I saw were West Clarke's documentation and diagrams for
this machine called the tx2 computer and the thing that struck me about that was
not exactly the layout of the machine but that he'd done such a goddamn
aesthetic job on these dot you should see these things you could you could
hang them up put them in a frame most beautiful engineering drawings I've ever
seen of this machine and way of thinking about how it was was done and I realized
that one of the problems with the work that we've been doing is that we weren't
being simple enough we weren't sitting down and thinking long enough and hard enough so in a
couple of years before I went to Xerox I spent a lot more time trying to think
what I what I thought of in those days were linked like thoughts which is
essentially removing features rather than putting them in and so my now in
fact this this machine was kind of a compromise between three strong
personalities one was me and I wanted to
build a machine that looked a lot like in fact one of the so-called portable
computers of today that in fact was going to be a link except we were going
to build it essentially out of a data general Nova but packaged it into a
suitcase and be able to carry it around and do lots of experiments and
it had many of the features that we associate today with micros Butler
Lampson wanted to build a PDP 10 for 500
dollars and Chuck Thacker wanted to build a 10 times faster Nova built out
of eckle and during the summer of 72 I
worked out the human factors of this particular display my main contribution
to this machine actually was the display design how it would look and so forth
most of this machine is attributable to
Chuck Thacker with a little bit of kit would sing by by Butler one side note
just again for those who are technically inclined the major reason why this machine works is that it uses a strategy
of multiple program counters which I believe was first used on the tx2
designed by Wes Clark this one had 16 different program counters and I forgot
how many that the tx2 had has something like 20 I forgot it had some long name
like a multiple selective sequence multiple sequence wes what is it what
was it called multiple sequence
something but the result of it was that
you could switch from one low-level task to another without any overhead at all
and the result of that allowed us to
throw away all of the peripheral controllers in this machine and save a bundle of money which money we put into
memory otherwise almost all of the
feature Hardware features on this machine are exactly the opposite as they were on the Flex machine so when in
doubt do the opposites like with the microphones you don't know what to do do the opposite thing and see what see what
happens and it worked out quite successfully so you see the same kind of
display the shape of the display was there because this the original name of this machine was the interim Dynabook it
was designed to be the functional box to learn about the part of the personal
computer that comprises more than 90% of it and that is the software the reason
we don't have a true personal computer today is because most of the people who
are building them and selling them are simply Hardware people simply software
people are not enough either what you have to have are people who want to
deliver a service machine as a complete package and that is something that we
don't have in a strong fashion yet
here's the this is a pointing device called a mouse that was actually
invented in 1965 by Doug Engelbart and
his group I should mention that most of the ideas that go through this whole
stream and nobody has mentioned it today yet is really are part of what I call
the ARPA dream or the ARPA image of interactive computing which started off
in the very early 60s ARCA was one of the main funders of Lincoln labs and the
work their funders at MIT many of the route 128 companies and some of the
Silicon Valley companies are a direct spin-offs of this particular dream one
of the parts of the dream is that the pointing device is about as optional as
a letter e key one of the things that
you should have on a true personal computer is the display that you can
look at for eight or nine hours without getting eye strain and so a considerable
attention was done this is an example of capturing a printing quality font and I
discovered that by doing a sort of a
Leonardo da Vinci gunnery table version of the modulation transfer function for
human vision that you could get a display to look like that I was taken in 1972 and that was the origin of the
displays that have come become known as the park bitmap displays trying to
understand what you have to put up there so that the human eye will interpret it as information rather than as dots also
one of the things I had always hated about computers is that people never designed on them and I think the reason
is that most computer graphics is very crisp and I felt that design is a kind
of a smudgy process and so you should be able to have smudgy graphics and so part
of the thing that led to this display is to be able to do smudgy kinds of drawings and of course if you have a
display that has about the resolution of a newspaper then you can put up anything
that a newspaper can represent in different fonts and so forth
and finally because display areas are always too small
give me one the site of size of a desk and I want one the size of a wall you
need to have some way of effectively multiplying the ability of the display
to show you information and the solution I came up with which this particular idea was invented around 1971 was
overlapping windows each window actually
is not a static picture but is something that you can think of as the back end of
a telescope that's looking into some simulated world and if you move the
pointer into a window that's partially occluded it pops up to the top instantly
so you can move around and have there's enough sticking out like on the papers
on your desk to remind you what's in each different window so you may have as
many as 10 or 15 different window areas all of different kinds of information in
them and then finally a bunch of us at
Xerox in 1978 did this machine which is called a note-taker which unfortunately had this point 5
herniation is per block feature but it
was the one that directly led actually to the Osborne and other machines Adam
Osborne saw this in an early stage and decided to do his machine I don't
personally think that this particular what it's called form factor and the
gibberish of the trade is right because it's still too big it's it's something
where you can always sit around and make a decision to leave it behind for some good reason and that is not a personal
computer ok lights again please
now I heard this morning that from dr.
Brandt he made a statement that yesterday's breakthrough is today's
graduate seminar is tomorrow's off-the-shelf home entertainment would
that that were true but in fact yesterday's breakthrough is the thing
that is most often forgotten it's not today's graduate seminar and it's
definitely not tomorrow's home entertainment because what usually happens in the the grand tradition of
Hollywood producers sitting around a table looking at a script and saying hey we've got ideas too too often people
it's not a question of not invented here it's a question if I want to invent this
myself and the if you look at the micro
computers that are being sold today you'll see hardly anything that
approximates the kind of total system design that the link had I think we all
can realize that they are not sold as complete packages they're not sold as
Honda's most of the things that you need can't even be plugged into them there
are millions of wires to worry about and so forth but what is that may not be too
surprising because after all that was a garage culture perhaps less forgivable
is the fact that the mini computer industry ignored most of the things in
the link and that's one that I cannot believe since they came out from the
almost the same culture but in fact this idea that you should be able to sell
system components as separable components that you as was jokingly
referred to this morning that the difference between the regular model and the deluxe model was going from four K
to eight K to me is it is a somewhat sad
commentary on the way marketing people think about things because if the user
needs eight K then you should never sell a 4k version
so I also disagreed with him about very
strongly about the goodness of funding and basic research I think that basic
research funding in this in this country except perhaps for the physicists is
pathetic the amount of venture capital
money that's around today is sucking up all the good people from the field and sucking graduate students out of the
schools in fact there's very very little basic research being done in this country
right now the Japanese recognize that
when they put in their 10-year plan for a MIDI of saying now for the first time
we are going to do basic research in computer science what they mean by that
is the age era of check cherry-picking American research is over the reason for
that is that there's very little left to cherry-pick and we're not funding it
ourselves the
back to this design thing I was thinking that as I was going down to the car and
coming back again that elevators have been around a hundred years or more
since Otis invented them but the elevators in this building show a market
disregard for the elemental facts of
elevator design if you notice that it's still possible to get into an elevator
after a hundred years and have it not work properly that really blows my mind
almost blows my mind as much as the overhead rack in certain Jets now you'd
think that if people I'm sure have flown l-1011 s most l-1011 's have an overhead
rack that is about this wide and has a double partition in it so there's almost
nothing that you can put in there now why did they do that how could somebody have done that since
the 747 overhead racks have been around for five or ten years that's exactly the
same kind of obstinate no nothing isn't that we see in our field people are
sitting around saying hey I can design an overhead rack too we don't want that
we don't want innovation for the sake of innovation Tennessee is that much you
know I've noticed in my travels that most people are not geniuses spectacled
early people that call themselves designers and I think that this this
whole area of misplaced improvement of things is one of the most pernicious
things in the field I was just talking to Wes beforehand that it's much much better I think to go back to zero and
start from scratch periodically than to try and prove something that's already
there you may get a different slant on it I think the main thing that you can
get from systems that exist if you're lucky is to try and understand some of
the aesthetics of the of the design this is what I call the siren song of
generality and I think this is this is something you know it's funny I have to
admit my main problem with Wes is that I always I'm getting light hits from him
and I'm it seems to take three or four
years for some of his statements to sink in but he and I were in a little in a
funny little group called educational kits group a long time ago
and he said something that I totally disagreed with back then and I hate to
admit it but I I've changed my mind about it and Wes was going on about what
we should do is this try and design who's talking about designing computers
like appliances I don't know what word he would use but the idea is let's
design a computer that you can't program which to me that's like heresy that's
like telling Christian Scientists about appendicitis you just don't want to hear
words like that and but now I understand
what he meant what he meant is that if you design a computer as though it were
appliance as though you couldn't program it then you get away from the siren song
of generality the the problem with
generality is that because you can do
everything you often try to and the
difference between those two is tremendously profound it's the difference between in art schools when
you go to sculpture class the first time they do not lead you to a lump of clay
and get you to push it around the reason is there are too many things you can do
with that lump of clay instead what they do is they give you a bar of soap and get you to carve it out in irreversible
medium same thing in painting class you get watercolors before you get oils so
you don't get to back up why because you can't do anything really good by
debugging it into existence you have to have an image
now why did it take me so long to learn that much much better to sit down and
pretend that you have only two vacuum tubes like grey Walter and spend that
whole year that you were gonna spend debugging trying to get an image of what
the thing should be and I think this is the this is what the people in the the
search of excellence book mean by keep it simple stupid the hardest thing there
is to any designer because design means power this design is the power to shape
the power to shape is the power to get oneself into deep dark trouble by trying
to do everything that the medium can do instead one should like Bach who was
probably the most had more musical resources under his belt than anybody
that has ever walked this earth could sit and just illuminate one idea for 10
or 15 minutes without straying away from it and be content to let that be the
best illumination of that idea that he could do and I think I'll just close
with that and say that I think Wes Clark is one of the best illuminator of ideas
that I know thank you very much [Applause]