Alan Kay at Pacific Northwest Bell Technology Forum

From Viewpoints Intelligent Archive
Revision as of 19:34, 5 December 2017 by Ohshima (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
as I heard some of the talks this morning I sort of changed a little bit
around what I was what I was going to say I decided that a good theme for this
talk would be from one clip that has had
a lot to do with the designs I've done over the last 25 years
McLuhan in the 60s said I don't know who discovered water but it wasn't a fish
and we can even imagine the fish having philosophical discussions about their
world without ever realizing that they're embedded in the context that
they can't see and I think a lot of what is going on both generally and within
your companies can take some some
inspiration from that there are secondary things that come out of that
one of my favorites fasces is the Olly
I'm as smart as you are fallacy and that's off and most ways that's never
the never the problem and it's I think it's as we talked about in the next few
minutes that the NIH is not a good way of describing what the problem is I
think that is a factor and we see it but the biggest problem is I would call
hyper rationalism in American business and in particularly in American technology we really think that we can
sit down and rationally lay out problems
and solve them by analysis favorite joke
at Xerox PARC where I was for a decade in the 70s was when we called the
Mexican airlines joke and this joke a
friend of mine was had to go from Mexico
City down to the Yucatan Peninsula got to the airport and saw the plane was
incredibly old dc-3 with paint peeling
off and oil dripping out of the engines and with some trepidation he walked
inside the plane looked in the cockpit and saw that half the instruments
weren't even in the in the instrument panel like they're just wires hanging out and the ones that
were there twisted about at crazy angles he thought to himself oh no I'm gonna
die on this trip but he had to go so he
got on the plane and they had a scary takeoff and dodging mountains lumbering
along down to the Yucatan Peninsula he decided he would poke his head back in
the cockpit and see how the pilot was doing so he stuck his head in there and
then noticed that on the instrument panel that every needle was pointing
straight up in other words this pilot
had solved the user-interface problem that was the first one you ever want to solve and that is how can I tell when
everything is all right
so any instrument that hadn't didn't conformed to this had been simply ripped
out the ones that had needles have been
twisted around so that up meant okay now
our society is incredibly interested in designing the perfect instrument to fit
in the instrument panel we will tell you exactly how low your oil pressure is but
in our design of the cockpit we will not give you any good way of knowing that
your oil pressure is low at all if you happen to look over there you will see
that it is incredibly low but out of the entire field of view we won't provide
any way to let you know that something is going on and that is the difference
between our hyper-rational left-brain culture and a much less rational but
much more field oriented visual culture such as we find in Latin America and
we've gotten to the point now where we have to consider much more than the sort
of Calvinist traditions of our society to the next 20 years Whitehead
philosopher Alfred North Whitehead pointed out that the great invention of
the 19th century was invention itself and he backed that up by showing that
the going to the British Patent Office and showing that the incredible rise in
the number of patent submissions factors
of 10 many factors of 10 more in the 19th century than in the 18th century
and what was even more important to whitehead were two other things one is
that the number of crackpot inventions that were under consideration for
patents or had gotten them was enormous they might think well gee why would he
care about that well that to him meant that almost everybody was inventing
didn't matter where they're a good inventor a bad inventor invention was something that you did in the 19th
century know something you didn't do in the centuries before if you urge a
random the other thing was is that even though
the number of crackpot inventions was enormous the number of good inventions
was also much higher before in history
happen was something very much like a something that I my analogy to that is I
had been interested in Renaissance painting for many years and looked
forward to this my first trip to Italy which I took some 25 years ago or so and
of course I went immediately to Florence and I went immediately to the Uffizi
gallery in Florence and I started going through the Uffizi gallery and I saw a
lot of things I was was I saw an
incredible number of meat paintings came
out of that tour with the realization and everybody physically have been the
part of what the Renaissance was about is all of a sudden they needed to had an
enormous amount of wall space and a new
wall space in it who could actually physically manipulate a brush had been
taught the techniques of under painting and drape painting and all of those
things and in fact the percentage of really great stuff in the Uffizi gallery
was much more like we'd expect from looking at something today but not what
we were led to believe from thinking reading about Renaissance art because they never show the media mediocre stuff
in the books the only show of the great masters and you get very little sense of
what was actually going on back there well I think that we're in a very
similar situation and I think that a lot
of what programming today is is very similar to that I think that if you by
and large look at programming as it has gone over the last 25 years you find
something very much like the new fitzy gallery is that anybody who can type in
a key punch over the years has can't has
and has learned how to program and has written programs what we have is this
enormous variety of ways of doing things some of them
existing genre but some of them in just
random ways or I'm as smart as you are ways they're not going to help and I
think that one way of looking at this is to go back to McLuhan statement and try
and understand what it actually means what it means is that any rational plan
that we make today is probably wrong
because the history the 20th century I
told you about the 19th century out of 20th century McLuhan I think came up
with a great sentence he said the 20th century is a century in which change
changed so what our whole notion of what
it actually means and what what that actually implies I think is critical in
doing software design basically I'll just make the blunt statement I don't believe that any software any
architecture that responds to current
day market needs is going to be a success I think it's going to be an unmitigated disaster because it is
actually tied to assumptions that we're making that says the future is going to
be like the past and that simply isn't true may have been true before all of
our schools in the United States are based on that idea it takes at least ten
ideas ten years to cycle any idea through a teacher's college that
assumption is is that that I'd all the ideas that are worth while teaching kids
are at least ten years old and nothing new has happened in the last ten years that's a fallacy so we're very much like
a situation in the beginning of this entry when the motion picture camera was
invented this is another McLuhan observation and it took 25 years before
DW Griffith figured out that you could move the camera
as all the people who made movies back then we're stage directors that thought
of the camera as being a person sitting in the audience in fact not a not just
in the audience but in the Royal position in the audience which is what
all scenery in the you know wonder how scenery is designed on stage it is
designed so that the King can have see everything in perfect perspective it
doesn't matter whether anybody else can or not so this idea of having the camera
there and having people slavishly imitate stage practices is something
that we do today so we're very much like people in the cathedral business during
the Romanesque period who suddenly had come up with an idea gee you know wouldn't it be great if we let some
light into these huge buildings what a
great idea and what we should what we need to do is figure out some way of
getting Windows without having the building fall down because the forint
unfortunately the roof is held up by the walls if we Pierce the walls too much
they can fall down to the wind can blow the thing over and what we're not doing
is saying our new vision of what change
is absolutely requires a different way
of doing our structuring so for instance
my big reaction this morning I'll tell you is when I heard people talking about new architectures and then using in the
same breath the word data it can't work
it just simply can't work data is too fragile to be put on a network the kind
of network that you're talking about so any kind of system that involves simply
integration and adaptation that retains the old concepts of data and procedure
without new ways of thinking about protection and new ways of thinking
about what those modules are simply won't work it simply won't and so what
what I heard this morning I'm going to sound the first order Theory's sounded
great and I think it's something that the telephone industry should have done
a long time ago because you had the leverage and the ability to do a long
time ago but simply trying to say to integrate things through network and
retaining all of the old things that you're going to try and integrate is simply painting over rust that will
eventually bring down the girders of the structure you're trying to build you
have to find a different way and what we're in the middle of one of these things that Coon and popper called
paradigm shifts and so what's and I
think the other thing you can't say is coat the two words that are bad here
once you say network is our data and code you have to go to something more
like saying gee if we're really going to have the kind of transparent virtual
system that we're talking about we have to have something that is much more like
a biological system than the kinds of fragile systems that we can build out of
balsa wood if you look at biological systems they work because each little
site module a tiniest module is its own miniature ocean every cell in our body
and we have a trillion cells is a miniature ocean it maintains and it
spends 90% of all of its energy simply maintaining its internal ocean against
all factors in the environment it only lets certain hormones in the cell what
the cell is therefore like being a liver cell or a kidney cell or something like
that is just the smallest part of what the cell actually does the big change here
in thinking about this in software is we have thought in the past of having a
software module spend 90% of its code or
maybe a hundred percent of its code executing the function for which it was
designed that's no longer be gonna be the case maybe 10% of the code is going
to do the function 90% of the code that's in that module is going to be
there to protect it from all the other modules and
and to be able to explain itself to you you're not going to find these modules
through a dictionary the modules are going to find you because what you're
going to do is say is you're going to say I need things that will do this if
you think of what it means to to do information retrieval you can find things by key word retrieval you can
find things by browsing but the easiest way of finding things is in a crowd here
I could say something that uses no keyword that people are familiar with
but if I gave a description and you understood the description then you could say I'm a match to that
description so these modules are going to respond to calls for help and just
like antibodies they're gonna come B opponents of the
things that you're putting together in other words these modules are going to act right part of what I want to talk
today is is some of the implications of that and I think that NIH not invented
here is not going to is gonna may be a problem for for modules but I think the
biggest problem is not with our code but with there is somebody else's ideas
because our industry has been marked or my industry the computer science
industry has been marked for the last 25 years by an enormous least stubborn
process of people not being able to move on from ideas that were thought up in
the 50s basically we are still writing code the way that it was done in the 50s we've gotten rid of the key punch in
most places and still find them in a few other places but in fact the way we
think about algorithms the way we think about machines and so forth hasn't
changed nothing so I'd make a blanket statement and say that good ideas won't
cut it you can't get by on good ideas you have to have some great ideas and
the reason is is that a good idea like a rational plan may be obsolete by the
time you implement it remember it takes more to implement systems and so if you
only have a 3-year idea you have just killed yourself you have to have 15-year
ideas and because you can't predict out to 15 years ahead you have to yeah that
can change it changed to model whatever
15 years from now now we're systems had better not have
anything to do with the current needs of your marketplace they better have
something to do with the idea that you're going to have to build complex structures that work over networks and
you better not have one single line in any of your descriptions that has
anything to say about what these things are going to how these things respond to
a particular market need today that's a way of polluting the whole design so
you're out after something more like biology you're saying I better not worry
about the organization the organisms I'm going to do I need something like DNA I
want to be able to build several million species out of the same stuff and I want
it I don't want to have to think ahead about what all of those species are going to be so I think the most
important thing most important analogies in this process are coming up with the
notion of what DNA is in a notion of what embryology is those are the two
most stable things in in the living
living systems now her oddest Herodotus knew this he
had this wonderful statement 2500 years ago that the only constant is change
itself the Greeks understood a lot of
this because they their change their world was undergoing the same kind of
upheavals probably the most traumatic event that ever happened to the Greeks
intellectually was the change from the
Pythagorean way of thinking to the thinking of Plato Plato and Aristotle it
wasn't all a good change for them but they were forced into it because the way
Pythagoras thought about things was that the world was completely rational and
could be rationalized you know we call
in mathematics things that are fractions rational numbers and that's because
rationalization means that you can measure them and you can form ratios of
them you can express things as ratios of other things and that was very important
to the Greeks for the kinds of things that they did and Pythagoras found out two very horrible things about the
universe which even the state of
Tennessee has not owned up to yet
one of the things he found out was that the certain triangles analyzed they just
couldn't believe that because you can draw the damn diagonal but if you take a
triangle it has to a side length of one
draw the diagonal then the diet length of the diagonal is a square root of two
and it's very easy it was very easy for the Greeks to show that the square root
of two is not a number that can be formed by any fraction there is no ratio
that expresses that that was like finding a crack in the universe to the
Greeks and in fact they never there mathematics could not recover from that
they were not able to go beyond and deal with what that actually meant so was
their worldview being able to measure things and Pythagoras did it again with
a more subtle thing which is the the
awful thing that if you if you hit play
the note C and you take its its second
harmonic which is a G make a note out of that and take its second harmonic which
is a D and take its second harmonic which is an a you can go spinning your
way through the harmonics of each of these notes as you go along getting all
the notes that fill up our scale and what they found out in 500 BC is that
that circle does not come back on itself now how could God have done that
and today every musician that plays a piano is forced to play in notes that
are purposely off-key in order to have 12 notes in an octave game string
players hate to play with piano players for that very reason because we are
inescapably forced into playing off-key because God did not see fit to make this
the 12th fret of to something that was Co measurable so the universe is not set
up the way we really would like it to be and every time we run into one of these
things there is a shock and every coil
now that brings me to thinking about how
do we invent things we predict the
future I think there are three main methods of inventing things the first
one is the brainstorming method that's where you get a bunch of people in the room and you you basically shove
ingredients together Jam ingredients together and you try and give names to
the resulting messes some of the messes you can paint over and then go out and
sell that's how the boombox was invented you take every obnoxious thing in a 12
year olds room watt it together and you
have something you can sell as a product and most businesses are much more
rational in that they have a goal-oriented strategy which has said find a need and fill it that's what I
was hearing this morning there are needs and we're out trying to fill them that's what most people do that's what
marketing people do when they try and predict things and the problem with that
is that has failed to predict or to deal with any of the great technological
inventions of the 20th century so the great technological inventions of the
20th century had nothing to do with either of these two processes so the
third way which I think is the powerful way is invent something that creates a
need only it can fill
okay lots of famous examples of this I have I have a morgue file that's about a
foot thick I collected these because the companies hate to believe in these ideas
because the the process of in that kind of invention is non rational but a good
example is was folklore and written in a
book by a former Xerox executive by the name of John guess our who's with the company from almost the very beginning
in his book was called my years with Xerox the billions nobody wanted the
central story is that after they had perfected the 914 copier which some of
you may remember the first plain paper copier in 1956 they took it to IBM
because they didn't have enough money to build factories to make this thing so
they said to IBM here take this build factories go out and sell it we just
want a small royalty we'd like to see this out in the world and IBM did what
most companies do when they don't know what to do and that as they went out and hired a consulting firm now Mike Hammer
has a wonderful definition of a consultant he says a consultant is a person who knows a hundred ways to make
love but doesn't have a girlfriend
like like insulting jokes there's another one is the difference between
the consulting firm and a Boy Scout Troop is that the Boy Scout Troop has adult leadership anyway
the consulting firm they hired was Arthur D little and Arthur T little spent 18 months coming back with a
report this that conclusively proving that there was no market for plain paper copying it was based on two major things
they went out and there there was no market demand for it people didn't do
much copying and it was more than ten
times as expensive as Mamiya graphing both of those things they thought just
how could it possibly work nixed at IBM came back and said no this enraged the
Haloid corporation people to such an extent that they went out and raised the
money for their first factory using their life insurance as security and that was about two hundred billion
dollars ago that story and there are many of them the airlines after will I
mean the railroads after World War one conclusively proved that railroad
transportation would always be cheaper than airline transportation you know
what they were exactly right exactly right the problem was is that they are
now almost non-existent they were so right they aren't around anymore
now's the problem with being right because rightness as we'll see as I go
go into the sort of the main body of this talk has nothing to do with the
perfection of your logic people were incredibly right when they're working on
alchemy in the Middle Ages but they were in a context where they didn't realize
that you had to put energy into any transaction with matter and so they
could not come out they were quite right and believing that you could transmute lead into gold you can absolutely do it
it's been done but you can't do it the way they were doing it because they were
in a water that they couldn't see and the water that prevented them from
seeing what the true pathways were so I believe that the you know the message of
this talk is that logic is a weak method and that humanity has made its way
from hundreds of thousands of years ago primarily by choosing powerful context
in which to think we always think weekly we usually think logically even
primitive people's think logically but their contexts are such that their
results don't help them very much and so the big thing is not whether you
can substantiate something with the chain of logic but whether you've got a context that's going to work for you now
another example of a great opportunity this is my vote for the first personal
computer this slide was taken in 1962 so
this slide is 25 years old this is a machine built at Lincoln labs in
Laboratory of MIT and is built for
biomedical technicians the had everything that we associate with
personal computers today and for those of you who are familiar with Digital
Equipment Corporation machines those things that look like deck tapes were
actually originally invented for this machine and they held not files but the
pages of the virtual memory this first personal computer had so this is a
machine and about 2,000 of these were built in the 60s mostly used by
biomedical technicians who couldn't afford programmers didn't want to wait for the mainframe and remarkably almost
900 of these are still does the
longest-standing computer architecture the reason is it was designed for
hooking up to things like nerve cells
and nerve cells haven't sped up in the last 25 years this machine was
programmable by its and users the end users like the machine and they just saw
no reason for throwing it away so there are quite a few of them still around and
praise them lots of places when I give talks I show this slide and people come
up to me after and say yep I got one - really great now the lesson from this is
that Dec Digital Equipment Corporation also came out of Lincoln labs in the
early 60s that's where they came from when they saw this machine they said to
themselves oh boy what a great way of building a small mainframe computer but
people aren't used to display so we'll take that off and make it an option and
people don't know what virtual memories are so we'll we'll take those pages off and I've got it we'll put files on those
tapes and we'll call them Dec tapes so a deck was able to invent from saying this
machine in 1962 was the mini computer a great step backwards dec missed what
what the actual future of this was but West Clark who invented this
machine saw it all he understood what
what what it meant to have this and in fact part of the design in this machine
was it was supposed to be low enough so you could see over it the idea of the
computer is an intimidating box that loomed over you he purposely designed
the case low like this so it would seem friendly also at Lincoln labs ivan
sutherland invented the first interactive graphics system in 1962 and
a my my deal in showing these old old
tapes is promised never to show you something that you can go out and buy so
i'm going to show you this graphics system from 1962 and see what you think
about sketchpad
it's done on a computer quite a bit larger than this room in fact the last
computer in the US that had its own roof
this is 1960 of 1962 this is the first movie the iben was the one who invented
the rubber band technique so you have seen that before but now look he points
at the edges here and says the sketchpad I want you to make those all mutually
perpendicular sketchpad just figured out how to do that
get pads the first system to ever have a window so in this first graphic system
he's drawing on a virtual sheet of paper 1/3 of a mile on a side again why is it
called sketch pad well you can just draw in and then give it a couple of rules and sketch pad figures out there the
rule was make it parallel and perpendicular here the rule is collinearity so the little dashes he's
laid down are being positioned exactly on the guidelines below so sketch pad is
not just the first interactive graphics program it is also the first non
procedural programming system and its ability at problem-solving goes far
beyond any schedule today can handle complex nonlinear problems to draw in a
in a bridge on it without it ever knowing about bridges it could automatically calculate the stresses and
strains you could draw an electric circuit and simply by giving the rules for omoi and what a battery was again
here you see he wants to make a rivet he just catches in uses that as the Center
for the arc and then to finalize the
drawing he again points at it again he's going to use this rule of making these
guys mutually perpendicular that will drag the center guy which will drag the
top guy which will give you a perfect rivet he could have constrained those to
be the side lengths he could have constrained those to be dependent on
something else
so we have to wonder when we look at this thinking of what was in Ivan
Sutherlands mind when he what he did this
and here's another idea which still according to one of the talks this
morning still hasn't made it in the door of this company here's an instance of
that rivet he was working on a master
drawing and see each of the instances can be dynamically scale rotate
it's gonna last well idea what this is this first system which it was discovered that the light pen was a very
bad input device the book runs out of your hand in about 20 seconds in spite
of that it's been the most reinvented bad piece of technology in the last 25
years it's been reinvented about 90 times so here are other instances now
you might wonder why is the display jumping around well he didn't really have been in the cute air defense
computer I hear what he's doing is saying I don't like those cross pieces so he goes back to the master drawing
makes them transparent and lo and behold we see into felt that change dynamically
let's get it also the first
object-oriented software system 25 years ago it had dynamic inheritance and all
the things that people are still wondering whether it's a good idea today
and in fact it's one of the best ideas anybody ever thought up and it certainly
is the best idea that most people have never adopt it is the way you build a
Gothic cathedral and here he's showing that anything that you make any
conglomerate conglomerate that you may be a master
well that was sketchpad I'd like to tell you where you can buy a system this good
and once they asked Ivan Sutherland you
know 10 or 15 years ago I said Ivan how could you possibly have done the world's
first interactive graphic system the first non procedural programming system
and the first object-oriented software system all in the same year he said well
I didn't know it was hard and in fact
one of the things that helped him out was that the display wasn't very good he
said when he looked at the display he said boy this thing isn't it is not going to be as good good as paper what
else can it do this is the big
difference between Ivan and almost everybody else is that I even asked that
question before 25 years around DW Griffith and asked that question what
can the cam what else can the camera do 25 years after movie started but Ivan
asked this question right at the beginning and that is the major question
you have to ask what else can it do whenever you're going into a paradigm change because if all you want to do is
your old stuff on it the architecture you're going to come up with it's not going to handle the stuff when you
finally figure out what else that can do you got to ask what else can the to the early on and get some sort of
philosophical way of thinking about it here's another wild man Douglas
Engelbart in the mid-60s and Doug was
the guy who invented the mouse and the black-and-white display and was the
first person to implement hypertext hypertext is something where people are
talking about today but in fact it was thought up in 1945 wonderful article in
Atlantic Monthly by President Roosevelt science advisor Vannevar Bush talked
about a machine called the Memex which would be the size of a desk would
have the optical storage of the contents of a small town library would have
multiple screens pointing devices ways of laying your own information trails
through it and he thought that would be possible over the next
twenty-five years and a number of people read that in the 50s including Engelbart
and decided to try an angle bar did
something else sorry these slides are backwards my fault
Engelbart did something else that was so
almost obscene he actually went into the offices and looked at how people did
things in their offices he studied his users something that we don't do very
much of today one of the things he discovered was that people wouldn't sit
hunched over their desks like they've been taught through 16 years of school
right because that's one of the major processes in American education is to
teach kids how to hunched over their desks and be quiet and he discovered
that even after 16 years all right
dad the people hadn't learned it very well instead they go scooting around in
their office they put their feet up in the desk they zoom their chair over to the blackboard and make marks on it so
Engelbart said well why the heck should we make people hunched over their desk to use their computer isn't it more
important than that so one of his earliest systems had a
laughs board that hooked onto your chair so that whatever you were doing in your
office you always had your computer navigation stuff with you as though it
were important and in fact so let's take
a look at this is a demonstration of
angle bar system 20 years ago
one of the reasons it's fun to show all this old stuff is that these guys in the
60s had incredibly pure vision [Music]
[Music]
[Music]
[Music]
I don't want his terminals right now so I have to ask for something to say about
it so we put a marker tracking spot that he controls so he's sitting in the park
looking at this text and he can point to it that we've carefully reserved for me
the right to control an operator on this so my bug is not powerful in yours
[Music]
in case you having something happen
collaboration here so we can go on into
information retrieval flying in looking at the same display
and that'd be very handy to work we could talk to each other point and maybe later I could hand you the chalk on the
stock for that saying here you control it and let's say this mode now and had
another feature that hardware-wise is available for display we have I'd like
to see you I'm working on it so before I can do that
I see it now computer do the automatic
switching they bring in a camera yes
this feeling was immensely intuitive beyond belief because he not only
anticipated most of what we think of as the personal computer revolution today
he went beyond it in that he realized that people are not natural born Hermits
that we're at communications junkies and that whatever computing meant to oneself
was something that was going to have to be shared intimately and so in their
first design is another one of these things incremental development balderdash it's not the way the good
stuff is done good stuff is done in huge cosmic romantic images and there's
nothing incremental about it at all Engelbart said oh yes of course I have
to work well I should be able to point
if they're going to talk points and move
things around conference twenty five or
six people in all different locations in the country each person's little cursor
would have their initials on it so as they we're talking and moving around you
could say oh this is this person pointing over here and moving this stuff so they understood it as a communal
blackboard a way of sharing a database intimately and of course you want to be
able to see the person now I don't want to make you feel bad but angle Bart
broke his heart in the early 70s going around to
telephone companies trying to get them to buy off on this vision couldn't see
it at all but this is what you want isn't it
that's what you're talking about that because more than what you're talking
about because it says yes you have to have face-to-face communication you have
to have all of these just voice files it's being able to have essentially is
embittered now though lately people
giving this talk recognition many people
from the 60s who had the pure vision and
money together the hard thing about innovation is getting anybody to
recognize it when it's done because if
it doesn't for rational scheme to even
see
personal computer I did it's brown 1968
and they'd gotten a little smaller by then and this is a desktop special kinda
like steel it's a weight about 175
pounds it had to distinguish a character
it first was it was an incredibly good coffee warmer now these little black
packages that hold the silicon Macintosh or an IBM PC have about 40 or 50 of them
that constitute the computer back then
you could get maybe eight gates on a chip and so this machine had about 700
of them crammed into this space and had a fan that was like in seven the other
thing it was in Israel positively
supposed to the people a little bit
moist and fuzzy down inside very much
the way people react to APL today and I
was disgusted I designed the programming language just one of the first
object-oriented programming languages very powerful simulation graphics the
sorry to close out it had multiple
windowing lots of different ideas but I finally realized that the big flaw in
the design was McLean's statement I
don't know who discovered water but it wasn't a fish I was a graduate student
and the puberty right of being a graduate student is be able to learn a
stupid system that has 500 illogical commands in the show that you can do
something with it every graduate student goes through that rite of passage and by
the time you've gone through it you've become entirely incapable of designing a
decent user interface because you've
adopted the idea that well what a user should do is simply remember everything
that's efficient you can
of control keys and double bucking control keys for every different command
and that's all you actually need doesn't
work and the revelation came to me at
night when I first got its work with
logo in bridge school in Lexington mass and it was a there's a revelation to
realize that something that had formerly been done in what we used to call the Glass House which is the air condition
that you could actually extend a substantial part of what computing was
all the way down into the world of the child and that was when my head did a
huge I went of IBM used to think of IBM
is railroad laying its own track
deciding its own destinations making up the timetable if you wanted to go anywhere he had to go on IBM schedule do
whenever you want it I thought of angle Bart stuff as being and I thought and
when I did this previous machine I thought very much of it as being a vehicle when I saw a pampered stuff I
thought oh no this stuff is not a vehicle it's much more important than that medium this is like paper and
pencil it's not something we would hold from children until they're age 18
that's something we wanted children to start using from the day that they are born there's another system done around
1968 by some people at RAND Corporation
Rand had a bunch of economists that they
tried to build computing systems for the
Economist I said look we like everything that you're doing first but none of us
can type can't you do some we're used to writing on notepads can't you do
something about that so it's already in 1964 here's what they did
it's called graphical input language
this movie is from around 1968 flips
the system and a new way of interacting
with the computer there was much more natural much more like with people had already been used
to doing
[Music]
no keyboard at all slow arrow recognizes
he wants a box and makes one now it's recognizing his handwriting the printing
in the box is being used as commentary only in this case the box is slightly
too large a change in size
[Music]
[Music]
[Music]
[Music]
I noticed that this is the end here he
sees it he hesitates we can just then
make the change without any command modes whatsoever this system changed my
life when I saw that I felt I got a chance to use it for a half an hour
sinking my fingers right through the glass of the display and touching the
structures underneath you notice that the system is analogic no menus you
never are looking away from where you're working and every command resembles the
thing that you wanted to do you want a box you draw a box want a character you
draw a character you want to get rid of something you scrub it out it resembles
the thing that you're trying attention
go away from where you're working credible system still one of the most
intimate systems ever designed I used to show this movie to my group at Xerox
PARC every three months so we wouldn't forget what we were trying to do this again was
done on a monster system the size of this room I should mention that both
sketchpad and system were his room sized
computers but in fact each of those computers was slightly less 512k Mac so
what we're seeing here is not we're seeing stuff done by single people these
but what we're seeing stuff is stuff that could be implemented on the micro computers of today and simply hasn't
been because the level of thinking about this stuff most people who implement on
microcomputers have no sense of that this stuff was ever done maybe Newton
said I was able to see as far as I could see because I stood on the shoulder of
giants it's because of this refrain I'm
as smart as you are but a relevant
important thing is we don't start from these ideas to go back and be smart as
Ivan is and that's hard to do we need to
start from ask ourselves what what has
changed in the last 20 years know the
1868 was a little one inch square flat
screen display developed at the University of Illinois when we saw that
we realized that the destiny and we're able to calculate out when the destiny
was going to happen Gordon more Intel for a 1965 that
predicted where silicon was going to go for the next 30 years he's only been off by about 4% over
those 20 so anybody who wants to know what's
going to happen in the next thing you simply go to that 19 paper and read it
you will find out why do you have a
flying a plane designed by an engineer
don't ever design in a fly by an artist
the big difference between the two you may enjoy sitting in the ground but you
may have some worries about how it might actually get up again Gordon Moore Dec
12 con was going to go by seeing this display we said okay sometime in the
1980s there are going to be millions of these what would it be like if everybody
had one could anybody and the lights
there's no hardly anybody could use these machines and so we said well the
real problem is and this thing as a medium pencil for even for things like
portability means you can close to 0.5
herniation per block is not portability
anything that forces you to decide whether you should take it along is not
portable we like the this metaphor when thinking
about children we said what would it be like for than the one pencil and all the
children had the time share it what would it be like for the classroom in
the school where all the pencils were and the kids could only go there for two hours a week in fact thought so we
decided since we couldn't understand how to design for adults very well so what
we would do is think about this lap
Dynabook we called it in the context of children and say that anything that
calls itself a personal computer has to be accessible to children or the whole
point is so this machine was built at Park in the
this is the 4 workstations first
bitmapped about is actually is a
Macintosh money to create the 1980s 10
years ahead of time we said this is what we want to do and
we eventually built 2,000 people all
over Xerox have these the ethernet was invented the same year and so was the
page a second laser this was concept no
are you at all evolutionary things almost never work
out this is done okay this is what it's
going to be like in the 1980s let's build it now this is as small as we could build those in your lap that was
as small as we could build it but it had all of the functional care that's great
had only bits because liquid crystal
display of the 19 a nobody
direct think so silly as to only because
we're simulating a display of the future why people do it with honor off dots
today I have no idea I think they're just slavish ly editor and the design
here was saying we don't want to have to undergoing to do so the first thing we
wanted to get rid of is any responsibility for what's what the fonts
are we didn't want to put the funds and wrong because the message of a bitmap is
you can have a kind of or any kind of display that you want let's make it
programmable the first font editor a page of text on that display we're
working with the idea they said hey you know hey have our composition our first
drafts of our time
because it misses the point you know this looks too looks too set in stone to
chiseled and bronzed so one of the kids put in his own handwriting so I'd rather
have my handwriting front first - fine
now I don't mean it yet and of course
you can not only do crisp computer graphics but you can also do sketching
sketching is tremendously important because up till the bitmap display and
these techniques design on computer and
I daresay very few people design on your computers the reason is is that because
of even when you have computer graphics it's usually too crisp and solidifies
you into a point of view before you want it done by making messes and then trying
to pull out order from the mess so we're
very interested and then of course we
knew about windows but one of the things you can do with a bitmap display that
occurred to me in the shower one day was that you could act as though the windows
were overlapping and that would be terrific magnify the effective area of
the screen and also give you something that was like the metaphor of dynamic
paper on a desk so that idea was thought up in these stuff so this this slide is
twelve years old this is the kind of interface of people now Mac that's an
apple that the letters IBM for
now important thing to think about is
this is something I heard in the talk
this morning that access isn't enough one of the 80% so it's going to be done
is going to be end-user I think not just
that end-users were going to make decisions but that end-users were going
to the 80% of programming those being
done I fact I think the figure two
percent one hundred percent in a few years maybe two percent will be special
authors but in fact the important thing about change is you can't anticipate and
therefore you build in rigid rules into your system it somehow has to be
sculpted we designed language calls
completely object-oriented and it has
okay all that word visitor otta morning
Jules it has inheritance it has share
ability and in fact the early version simply with children could actually use
it as an authoring language so not only were they accessing tools what their app
to make things if you think about this
in relationship media we don't think
people are literate if all they can do is read so that means all I can do is
consume material done by others we say people should be able to write at least
a two page memo page essay is a
rhetorical tool you can change somebody's mind with it you can clarify
something the equivalent of that is to ask what can you do in a couple of pages
with your computer system
here's answer to that
so here's a system that's quite
recognizable to us today it's like Mac
so it's an object-oriented graphics system it has a menu of commands down at
the bottom back window in the middle here's the color palette off the screen
this size move it the only thing interesting is it
this is not bought in the store this movie 1975 12 years old application the
application was designed and implemented
by a twelve-year-old girl after just three months programming and small talk
the size of the code to do the entire application was 50 lines this is a one
pager
these professors are horrified because they'd be signing this for their
sophomores and juniors one of the things
let's think about and one of the ways of
in your new architecture is not just whether you can integrate things and not
just whether you can have modules but if you don't have a size reduction in the
size of your code of at least a factor of 10 and a reduction in the complexity
of your code by a factor of a high your run run here's the system done by a
15 year old who was interested in ham radio feely sophisticated he has a
little pop-up menu he uses there for often you open and closed dot he has a
menu of components down there it's also
done after about programming and small talk it's an 80 line piece of code about
a page and re rotated my programs done
because before at a PhD has been awarded
and it's good so I figure we were making
some progress
final question we should ask is well how
much of a medium is this it is really
all of human existence so I got from the
woman who is my accountant
about a year and a half ago this is her
daughter 22 months old two months shy of
her second birthday she world without it
was not intosh 'as both of her parents
work at home her parents when I found
out the little girl was interested in computers like to which she rejected she
wanted what her parents had because that was part of her environment and all I
don't believe in user surveys I did one
with this young user and I discovered she vastly preferred PCs with installed
hard disks
treat it surprises can use the mouse
without looking at it but in fact everybody can and I've never seen
anybody this young do this but it didn't amaze me too much that she was able to
pick out visible macpaint here and
change parameters and so forth so I saw
this I was intrigued but I wasn't overwhelmed until I saw what she to get
a new sheet of hits to close a box in
the windows saves her all uses the
pulldown to get a new one and she's off and running again and we discovered that
after this we took many many hours of video tape of her using a variety of
applications and we discovered she was 70% literate in the Macintosh generic
commands so she could even start up in an application like PageMaker
good here make some marks print it save
it get another one and so forth so she could handle she was basically 70%
literate in what we call the Macintosh user interface language at age 22 months
and in fact we discovered that the mouse which is the size of a brick for her
Kazak to use at that age than a pencil
she could do things with it she couldn't do because it was had enough mass to be
stable for so there's one answer to the question the access so you get with a
well-designed user interface all the way down
and that's incredibly important because it's only which is gonna take place we
learn things from birth remember technology all that stuff you
were born right we don't think of
watches as being technology some of us think of television like television but
many of us in the room think of television as part of the environment and our reactions to it have to do with
whether it's a round I burned it as a
child or whether it was something that was thrust upon us later on which
technology opposes the technology so is
maybe just quick so I think I'm out of my time usually in this talk I go
through what the the psychological reasons what we have goals that make
user interface designs like this work the enough perhaps too much in the
beginning of the thing and so let me finish off with another user interface
joke and this one is a Winston Churchill joke and like all Winston Churchill
jokes it has to do he's at a party after a certain number of whiskey's and like I
should tell you to Churchill jokes because the both of them apply the first
one is he's had many whiskey's it's a very late at night very early in the
morning and they he's sort of peering
out Lee around the woman and the Duchess comes over to him and says mr. Churchill
you are disgustingly drunk and he looks at her and says Anne you madam are
incredibly ugly but in the morning I will be
indisputably sober so this is one way of
one way of dealing with your evaluating your information systems and see what
they look like in the morning the second Churchill joke is ready the hostess of
the party comes over to him and says mr. Churchill I'm so distraught I just saw
over here steal one of my silver salt shakers what am I to do and Churchill
thought for a minute and stuck his cigar in his mouth and went over to the earl and look on the way picked up another
salt shaker and put it in his pocket when he got over to the Earl he took it
out and we've been noticed sweet back
and the message is if you want more to
the message is if you want people to go
along with you then you have to involve them in the same conspiracy and modern
day computer design I submit is almost
entirely what the users need to do and
who the users are it's a new conspiracy
this that I'd like to see it first
where has been last in any design process thank you very much
[Applause]
any questions that is a dream of Apple
that we have and let me substantiate the
dream a little bit by pointing out that
these ideas that we thought up at Xerox PARC are now 15 years old and and
because iBM is now interested in them we know that they're obsolete it is that
pervasive networking which i think is really the major agenda that mean we
could say why isn't the telephone company been thinking about this much longer phrase which I saw in a slide
this morning was thought up by John and written in paper 8 and the notion to
deal with an information utility was thought up a long time ago it was
realized that once you have pervasive networking the kind of user interface
works with the children everything else doesn't work nearly as well anymore
take a Macintosh style user interface
hard disk work to it and the browsing
that helped you so much that we sell we
now have 8% of the business market about 90% of the machines we sell into
the Fortune 1000 company we sell because they can pay for the machines out of the
training costs that they save they save
about two weeks of training for every Macintosh that they said that they buy
and so all of those things that are on
there in fact start breaking down a little bit when you have thousands of
things that can consider and remember when I was a child I was visiting
somebody in Washington DC and I had a question I wanted to answer and I was in
library to that question I heard they
had a library see so I got my mother's
alright discovered that the card catalog of the Library of Congress was much
larger than my hometown library
discovery when I was there is that no person going into the Library of
Congress actually I'm still there the
people are people who are experts in the real strategies for using person I got
one of those Peabody's like that people
attached to me for the hour or so it took to my question long time because
all the things we think of as being modern user interface techniques start
falling down terribly because the user interface that we designed at park we're
reactive not proactive to volume level
of resources and need to learn new things and the imponderables to a or a
million different things then you need a proactive user interface that kind of
user interface was thought up actually again in the early 60s and given a name
called agents an agent is a semi
intelligent computer product take on so
structure you may work 95 and without
you ever seeing it it's one of the death season let me show you this tape this is
why
[Music]
[Music]
[Music]
you need to take Kathy to the airport by 2:00
[Music]
[Music]
[Music]
[Music]
[Music]
research
to recheck these figures of deforestation you predicted
[Music] I have a simulation that shows the
spread of the Sahara
[Music]
[Music]
[Music]
[Music]
[Music]
[Music]
would you leave a message this is your mother I know that you're
there [Music]
that's why we say the the IBM attempts
to do user interface are sort of like putting béarnaise on a hot dog that
these ideas to do a user interface doesn't live here as it takes to refine
it many of the people that have been working on this concept the agents for
more than ten years now exist ten years
ago people who have come to Apple in the number of years and by the time 1995
rolls around there will be enough networking and so forth so that all of
the user interface techniques that we
now think are a vanguard commend me with something that is much
more proactive more anthropomorphic and
probably squadrons of the agents rather than just a single one that's what we're
working on very hard to do because the
again we're doing a project a very large your children again working with agents
because the one of the hardest things about agents is that they have to be able to learn and there's nothing in
artificial intelligence right now to
bridging the gap systems and what even cats and dogs are able to do and so
we're doing what you might call basic research very basic research in trying
to produce artificial mentalities that can actually be resourceful and learn
for example US Apple may have some
environments
mm-hmm well they do want to talk to us now and what I'm saying is is that there
there's a tension between Apple believes and Apple is not like Steve Jobs was
where the company just go it alone we
would do in IBM in five years and so forth on the other hand the prevailing
wisdom and even people on the Apple board six years ago really thought Apple
should do a PC crumb that was but that
was the standard and if you want it well
you have to work within the existing Santa you see the problem is is that there's a creative tension the standard
rock you in to obsolete ideas
domain that actually release you
difference between defining and data structure which is the the worst thing
you can ever do because then every
procedure in the system thinks it knows something about how to use that data
structure make any change you have to find all the things that have
assumptions about it versus doing something in which the object is able to
protect itself and complain and even adapt so yes I agree I agree with some
of those ideas but I also believe that the that in one sense given how by the
is all right I think we're at a place where companies can't afford not to be
current with things that were done 15 years ago figures and so I think the
companies at least have to say what we're aware of this stuff basically
should something like well the Objective C people say mumble mumble rejecters see
people didn't invent programming
and asking people who invented object-oriented programming what it is and what it might do so I think going
through second secondary areas and hearsay and stuff doesn't help help very
much either and the important thing to understand is that no user interface counts until
you've tried it out for many many years
variety of users I mean Scully had to say to the
educators at edge accom that please don't expect this by Christmas this is
not a product that Apple is doing it's it's a way of thing flaws in this we're
gonna redo the movie because it was it was done rather hastily and the people
who did the movie missed the important idea that the thing is portable and so
number one you've never print printing destroys almost everything that's interesting about information and number
two he would have it with him so at the end of the movie instead of abandoning
his mother he as his mother said I know you're there he would simply close the book
so getting these visions set up is a very important thing getting them set up
and so that they're at such a high level of aesthetic romanticism because the
romanticism is the thing that gets as you implement a set of design is that
after you've been implementing for five years is to have something of the original spark left in the final result
this thing in Hollywood called war mode so about four to six pictures done in
war mode and that is when they get to the point where the deadlines are just so crushing they just stay up night and
day people write software this way the
problem with the war mode Zion is done during it and so you wind up with this
incredible pile of baling wire and stuff whose only structural integrity is the
structural integrity of the programming system that you're that you're working in and since they have almost none
generally then you're indeed it's just
that simple yeah I think talking but see the talking
shouldn't mean the least common denominator it's what I hate about television television communicate
communication on television means that we're not going to assume that anybody has really learned the English language
human eggs and pratfalls it's not whit
whit requires you to understand language and something that you can allude to and
I don't think that's good enough I think the a better way and that's I
have a huge objections to systems like UNIX whose notion of inner processor
communication is let's send a string for one process to another how can UNIX be the great white hope
when it wasn't actually a good design when it was done in 1971 and just
because it is better than even worse designs you say I was saying at Xerox
PARC that relative judgments have no place in art doesn't matter whether Bach
was better than Beethoven it's not not an operable question the real question is what's the difference
between great and not great UNIX is not great but doesn't matter how much better
it is than anything poorer than it I think we have to go above and a my
feeling about what protocols are about and what networks are about is that people are thinking very much along the
same kind of limited lines at Unix you don't want to send strings because
strings have to be parsed at the other end there's no context that goes along with strings you want to send something
like objects that means that the whole set well about what it means object
3,000 miles and have it be viable and able to function able to explain itself
able to integrate itself all of those things at the other end there's no way
that people are going to do I don't
believe any configuration system is going to be able to do that and the reason I say that is that some some of
the best systems that were ever done we're done by the another group at Xerox PARC
system called cedar that in fact had very similar architecture to the one
that you're talking about here and it was it was better than anything that
gone before but the configurations manager was a nightmare to deal with
because it was never quite sure when you hooked in the component whether you were
hooking in something that was really going to do what you wanted it to do or whether you're simply hooking in
something in which the interfaces were compatible right one thing that have the
interfaces be compatible and another thing for understanding that the actual
functions that's going to execute for you are what you want that is the
nightmare of doing components and inheritance any other questions yes
the statements that power supplies for power source was the biggest impediment
to something like the Dynamo getting that time frame the last six seven eight
months shouldn't the variety of things
the if you look at the laboratories of the Japanese companies then you will see
the displays the first displays that are good enough to be on a Dynabook and so I
I say that those are here and the power consumption our nose is not so bad the
power consumption on the processor is powerful enough to do a lot of these
things it's not so bad it's about a wad or so and the the thing that is
difficult it's at a rapid speed as
possible to build a Dynabook this very instant if you had a system they switch
in and out you know there because the system would have like ten or twenty
megabytes of RAM some of the RAM would
be used the way we use files today although there would be no files in it
but some of it would be stuff that you weren't currently working on and a
couple of megabytes would be stuff that you were currently working on and if the
system were able to idle the stuff that you weren't working on then the power consumption would be reason and the
power consumption that that we thought was a minimal is something like 10 to 15
hours of being able to do anything that you want it and right around the corner
but I think to me the dining book was never a piece of four it was a
head-mounted display with a thing you could slip in your pocket - to me the dining book was always a service concept
and there I think you could say that simply translating the Macintosh unit
user interface into a 2-pound system
with twenty megabytes of RAM is not it Macintosh is is
is the first person criticized and
that's that is my opinion of it you know it is a Apple the heck of a risk to do
it I'm glad they did but in fact no matter how you slice it the Macintosh is
an embodiment of things that we thought were important in the early 70s it
doesn't reflect things that have been discovered since then and the kinds of
integration it has although the best in the industry are not even close to what
you really need just to give you an example the one of the big principles in
user interface design is what we called results mode interaction the idea is
there you should always be working in the context is going to be think about
the Mac it does that a little bit spreadsheets and other kinds of things
that run on it but in fact if you think about making up a document what do you
do you shouldn't get your and you go
after this application maybe get your text and you go after this one you get your pictures you cut and paste them
bringing them back in why not just work in the context of your document and you
have a super document in which the paragraphs are actually windows into
your favorite top ten applications right that's better integration Macintosh
doesn't have it nobody else does either and that's just because the Apple people
decided to use the cut and paste metaphor that was one of the things we tried out at Parc a long time ago it's
not the best way to do it so I know the
thing to be a dining book for me I'd like to see it reflect what we
understand what we had Apple feel very
critical about in our own user interface I think he could do it Dynabook without
going into the agent idea I think sort
of a Mac you know a Macintosh double star type user interface that was also
programmable the other thing about the Mac user interface and other interfaces
is that they're not programmable in any way and the major idea behind the whole
thing was that you're supposed to be able to finish this delete three files by simply picking them up with your hand
and dumping them in the trashcan but you shouldn't have to do for leaving fifty three files and then
somehow we had solutions for this apart somehow the user interface would teach
you what the symbolic equivalents were of your concrete action wanted it to do
something more and powerful and you have
a reason you shouldn't have to read an ms-dos X to delete three file but for
reading fifty three files you shouldn't have to do it by hand and so the big the
big slogan that I made up for that interface is that doing with images make
symbols and that the concrete actions of
your hands and your eyes should lead you to symbolic renderings of those things
that are more powerful through abstraction that doesn't exist that will
in the next few years and that will be a heck of a good user interface before we
start bringing the agents onto the world the agents are gonna take a long time to
do because the agents raise expectations that can't be fulfilled another way of
thinking of using interface is the most important thing you can ever do in a user interface is to subtly broadcast
what it can't do the Mac works partly
because almost every guest a user make is fulfilled people an enormous number
of percentage of people never read any of the manuals on the Macintosh and I
pointed out to Apple that since that's true we should never put anything important in a manual so first as most
Mac users don't know that you can extend a selection by the thing called
shift-click of holding down the shift key and clicking that will extend an
existing selection because there's nothing in the user interface that tells you you can do that it's in a manual and
if you haven't read the manual you never learn that and that is a very bad thing to do the agents are exactly the
opposite to use shouldn't that can't be
fulfilled their answer for morphic so we immediately expect them to understand us
like people we meet expect them to do things like people no way only we do
this film we're going to redo it as we did it at Atari some years ago we did a
bunch of agents where we use cartoon characters
you can also broadcast through there exactly they are now if we were to do an
agent an agent today we want something like Porky Pig or Elmer Fudd where you
just have very low expectations the
agent ever brings anything back that's interesting to use they all boy that great there may be something like Pluto
you know Pluto ever winds up with the bright news paper back you say oh good
dog I'm not kidding because that's
that's one of the reasons why age agents have been thought about for a long time
at the big TV stations that get you very
very angry when they're not worked out and so a lot of the design is how do you
defuse user expectations but still make them want to use the agents for what
they're good for I've been working on it I've been working for almost a decade
yeah and it's a very interesting process of doing the design besides all the AI
and other junk that goes on underneath
yes there's this concept of agents that
you seem to be going towards is that in a really obviate point of trying to
design the intuition in the interface well I think because it appears that you
know it's icon or itself driven the user has to make a choice and it appears in
this example using agents he's just saying I have a request now you get it
for me well I think I think to the point I think they're there to I think a lot
of retrieval I mean a good example where standard icon driven things break down
is because this is an agent that we built at MIT a few years ago called News
peak and what it is is an agent that
stays up all night logging into information sources like the New York
Times and the Association or a newspaper
a personal newspaper for you to read a breakfast okay so it goes into Nexus and
the stuff and reads the articles in the fine Rondon vo disk that has 45,000
pictures of famous people Z for morgue
Iran Paris Francis mentioned it finds a
map on a map disk uses an expert system to layout the thing what you get is
something that looks like your own personal version of USA Today right it
might say in Afghanistan or might just as easily say your 3 o'clock meeting was
canceled today because one of the new sources that goes to is your own electronic mail but you've gotten
anything interested the night before and it tries to screen that and see whether
it's important a sidebar might say your children slept well last night I think
redefines news Telegraph move literary
remember that the Federalist Papers which were the justifications for why
the Constitution should be ratified were published in the New York newspapers
nowadays we think of them as something being so literary that most children
shouldn't even read them actually the
kind of thing of newspapers in the late 1700s the telegraph changed that by
making currency more important than depth Terry look since 1840 there so and
now we're saying that what we're getting is something that's basically a slice of
interest ok so there's an example of an
agent now think about the process for trying to get the agent to know what
things that should go after think about trying to set up any kind of information
system you've ever used trying to set that up forget it and the agent we did
at MIT does three-quarters of its workplace watching how you read the
articles and tries to get some what it's trying to do is come up with a sort of
what you might call a meta collection of notions for describing the kinds of
things that you're interested in that have no particular keywords or any other standard information retrieval
paraphernalia done over a period
gradually learns and the agent visits the agent at MIT periodically comes and
asks you about sir you have to see just
to check out and see whether you are so
I think that the having an obvious user
interface that has some metaphors to the world of objects and so forth it's
always going to be there but I think there's going to be enormous amount of things in which the the kind of
information you want has no obvious keys into it and the agent is going to be
gradually formulating over the life justice is going to be learning how to
understand you're speaking better and better one thing has been completely missed by people who do speak correct
each recognition so almost everybody who's ever done it try to do this is the
old time-sharing way of doing it we have to have a facility on our central computer that can I understand multiple
speakers and stuff like that in a Mac -
that's your Mac - Mac - squeezed down to the thing that's in your pocket can be a
speech recognizer the recognizer that lives you with you for 40 or 50 years
learning as it goes it doesn't have to understand anybody else's speech except
yours and that it'll be easier video
synthetic speech from a wide variety of sources the technology has almost been
ignored in the last 10 or 15 I stated
country in which changed changed
and things that were inconceivable and they're still thought of inconceivable receivable
we're often inconceivable twenty years ago because nothing was and nothing was
right for doing it and just as in sketchpad we have not done a sketchpad
today because the all of the assumptions that said sketchpad was not a good idea to try and do in the 60s are no longer
true but nobody remembers them so it's simply something that you know will be
done graduates or even an undergraduate
project and we'll have some more in the world these gaps are really just one my
favorite saying of Marvin Minsky's they are I understand something unless you
understand it more than one way I think
that's the best way no matter how much you think you understand it if you don't have two ways of understanding it or
more than two ways of understanding you don't understand it you're trapped in the context thank you very much
[Applause]
[Music]