Alan Kay Talk at Virtual Heidelberg Laureate Forum 2020

From Viewpoints Intelligent Archive
Jump to: navigation, search
okay it's
my pleasure to introduce the next speaker of this
morning whatever depending on your
time zone it's allen k recipient
of the acm am touring award in 2003
for i quote pioneering many of
screen and go to full screen on slides
and in the upper right hand corner
is a place for your technicians to put
hey will
see what it is i hope
did the technicians get this
yes
they did and andreas
we're going to have to you're going to be in the way of the slide
so if the technicians will they can take me out
i'll be happy yes well
they yes okay
there we go
and okay
um hello everybody
uh i have a simpler bio than
the the one that they
gave you which
is just that no
his research community than i do
and that of course
was the arpa
information processing techniques office
uh in the 60s and
then xerox park which was an outgrowth of it
in the 70s
and so the simplest
uh piece of advice
i can give anybody unless
you're truly de novo
is just find a good research community
ours
benefited tremendously from
this slogan the goodness of the results correlates
most strongly with the goodness of the funders
we had i think the greatest
funders that computer science research has ever
had and so
much of what
we did and
that is idiosyncratic to
that particular research community and those
funders so i'm not going to
cover that today i've given longer talks trying to
explain this research community today i thought i'd
are maybe more generally applicable
and that would lead as quickly as
possible into our question and answer
the direction of this talk from
questions that uh
some of the junior researchers
emailed me over the last week so
you to them and some of them will
see some of their questions answered
the whole talk is really
about pondering i've divided into two parts
the first one
is just dealing with
the tyranny of
the present or the tyranny of normal
and this is an idea from arthur
kessler book
act of creation
and he says imagine our
mind as being
like a flat plane and our thought processes as being
like an ant and the
ant doesn't know it's pink
because it's the only uh color that
ever seen it can wander around it can do
all kinds of things it can find obstacles it can problem
solve uh
and every once in a while it will
have an outlaw idea
maybe in an unguarded moment
but the aunt has been to church
gone to school the aunt has
tried to get funding from a government funding agency
and so there's a big ker splat
that wipes out that little outlaw
idea get back into the pink there
but maybe the ant is taking
a shower just waking up and all of a sudden
this
outlaw idea gets much much bigger
it's a kerpal and
when one of those gets big enough
makes you see oh here's a blue
plane i didn't know that
my thinking could be more than two-dimensional
but here's a whole other world i didn't even know about
maybe i'll start exploring that and
course once you found
a an exception
to the world
that you thought was normal you realize oh there's
lots of them everywhere
there is there's a possibility escaping
and getting into a different set of thought processes
so the slogan i made up for this many years
ago is point of view is worth
80 iq points often
you are the context that you're
in uh that makes the difference
and certainly we
in our era are in
a context that was laid in the 17th century with
invention of science a lot of
what we do depended on what happened then and
years or so earlier like
leonardo was we would lack the context
to do anything with ideas like he had
another way of looking at it is from marshall
mcluhan i don't know who discovered water but it wasn't a fish
the ant doesn't know that it's
in a pink plain because it's only ever seen pink
and a question for you to ponder
while i go on with the talk if
the pink normal is sanity
because we equate normal with sanity
what then is a blue thought or a green thought
okay here's another way
to look at it uh from the
generally in school we're at
given a problem b we're supposed to
go towards it and be successful most
of the time but
in fact lots
of real problems also
exist in more dimensions
than school problems or most problems
eem to be so in the ant here doesn't
can feel gravity
can feel things are getting more difficult
but just feels the difficulty
ant has been taught well
then it will just keep on trying
it will hold on
to this particular goal in this particular way to get to
kay here's a
guy in africa he wants to
catch a baboon
so he pokes a hole in a termite
nest here
here's a young baboon
him do it wondering what's going on because he's curious
like all us primates are
and the hunter
puts some seeds in the hole
and just goes off to the side
the young baboon is thinking about this
curious
curious
better go take a look
i can smell
oh seeds
grab onto those seeds
whoops he's stuck could just let go
of the seeds but he won't let go the seeds
he's a primate just like us
he will just hold on to those seeds until
a guy comes up and catches him
now this is a cognitive
that is called a loss aversion
it's one of several hundreds of cognitive biases
that we have this is one we share with
primates this is also how they catch monkeys
in burma and
i won't tell you the rest of the story you can find this
video on youtube uh but i will tell you
hunter does not kill or hurt
the baboon the reason he caught the baboon was for a completely
different purpose which you'll find interesting
okay but what if we didn't have laws
aversion but one thing explorers
know they have a heuristic that says boy if things are really getting
tough we should at least explore around
the hill and the and the gully
there aren't everywhere
and so we might have to take a longer route around
but actually we'll get to be faster
we'll just have to travel a longer distance
and in fact
away from this goal we might even find a super
highway one of the super highways we use
today is called science
science get off to
can go much much faster
and in this case you travel longer but you get
to be much much quicker
but
you could also use
science to boost engineering and invent an
airplane and just fly over the whole thing
that takes a little time but now we've
got something that's generally useful
and even better once we started
thinking in this way we might find
one of these blue planes
and on the blue plane we might find
a goal c that is much much better than b
was and in fact if you think
about it this is
a key
to even thinking about your own schooling if you
went to college trying to
get to b and college showed you so
much that you came out with a c that you
never knew existed before the college has done its
job otherwise it's only
helped you on your own goals it hasn't helped you expand
now here's a a barn that burned
down one that burned down in the 17th
century inspired a japanese
poet to say oh the barn burned down but now
the moon was more rewarding to him
than his barn
that gives us a way to think about
how do we think about ideas do we think of
them like matter
they are separate things they collide with each
other like words or categories that have hard boundaries
or do we think of them like radiation
or light or processes or relationships
and these superpose
so it can shine all of them on the wall at the
same time and we can
there we can see interesting combinations
other they might not
be compatible directly but by
not fighting each other we can get some ideas
about what we can do with them
and guess what the the moon might actually
have been hiding all the time behind our categories
but in fact in
the radiation idea of ideas we can find
the moon and of course in
california we like t-shirts
o this is a good t-shirt for researchers and
other curious people
and again these categories
have this sense of reality
they're so well defined
they're so uh set up they've
been used so often that they actually can get
in the way about thinking about things
well here's uh kessler's book
act of creation written in the mid 60s
it has many wonderful ideas not
just this two planes intersecting
he has a whole theory of creativity
and humor and science
and many other interesting things he was
a very famous writer who later in his life turned
his hand to the behavioral sciences
and a more recent book
which has many things
to do with the way human brains actually work
by kahneman who won the nobel
economics also
with his partner tversky
uh coined the term cognitive bias
and i uh
if you look at the wikipedia article on cognitive biases
and the cognitive bias codex
find all sorts of interesting things including
the fact that you'll be able to think of more cognitive
biases than are listed there we have a
of them now if you think about it if we know what our
cognitive biases are as human beings
we should be able to create heuristics
to deal with them
think of them as dangers out in the
world we have to make up heuristics to deal with
these things okay
the second part of this
which is really part of the first part
is that generally
speaking and certainly
in my experience almost all the good stuff
that happened in my research career
of more than 50 years now
came from finding problems the problems
that were around the problems
nsf likes computer people to solve
and so forth are ones that
can be explained the solutions can be
explained so they're really more like engineering
but in this golden age
of 60s and 70s when a lot of new
stuff got done a lot of it got done because
arpa and park were willing to fund researchers
to poke their noses around
with their own conceptions of the problems
most worthwhile looking at
so you can think of it as widening context
closing your eyes
the perfume that's around
find the perfume that smells good to you
so
this context idea is interesting
we still have to explain
people the person is the first word in personal computing
why do we have to do that because the
user interfaces they do are so terrible
so they're much more interested in the computing
part than the person part but if you think about
it this is not a great term because
um humans
don't exist by themselves
out of the woods or in a cave
we most of us view that as punishment
and so what we're what we're actually
embedded in
and we can't be human without is from the time
of birth to be embedded in a
human culture and that culture
is where a lot of the
learning and attitudes and worldviews and other kinds of things
come from if we just pick
a few things that come out
of this way we can say well person is
personal computing but in fact in
a culture we have a duty to our society
we have a duty to the next generation
it's not just about us
we have to understand the world views
we have to deal with
the schooling of the next
generations and the adults that we have
today there's this idea of richness
man does not live
by bread alone so this
notion of richness this is a tough one
used to be more embedded in schools than it is now
it's one of the most important things because
it deals with things that are outside
of simple pragmatic
problems and then there's the
idea of livelihood
earning a job but it's sort of the least important
of the six that's around and these
seven things together are too
small a number but it just gives you an idea if
you just pick seven things
for a context for doing research and computing
here's seven i picked
and these are themselves embedded
in much larger issues
here are 12 big issues there
were issues 60 years ago when i started
their bigger issues today
and
these issues are actually global
so they're not tribal
they're not local cultural
there affect everybody and
the global environment is dying
in part because
many of the people who are the most active
over the last 150
years we're not concerned with
the larger environment
so if you think about a child born
year who is going to be 80
century we wonder well are they even
going to get there is there any possibility
80 years from now
for that world to be better than the one that we have
now or is it going to continue this
piral so why do i put
here because this is where
my community
in the 60s and 70s took a lot
of the romance for
the problems that were chosen they were chosen to deal with
these large issues not
just helping engineers
and we can widen this out one more
level to thinking about humanity
in general which is recently discovered
a planet many people don't really get that
a hundred years
ago the
in communication with each other and our technological infrastructure
this is a self-portrait of the internet
uh is enormous and also
large
our bodies are very
complex systems and our brains are even
more complex so if you look at these we've got
his idea
of i hope what what that caption
says there it's blocked on my screen
it says uh the systems
we uh live in and the systems
we are so we have this context
with which to get these new
and they are connected
so it's not just just that each one of them is a non-linear
complex system the whole thing is very
complicated for most people still these
are invisible they're not taught
now we can get to this
much larger view of science which by the way has its 400th
anniversary this year from francis
bacon he defined
science that we need is that we
need to have methods to get around with what's wrong with our
brains
and we will learn
thereby this is in a book called
the novum organum scientia you
can look it up
so science is about 400 years old
ating from this bacon idea
systems are less than 100 years
old and not
even taught in most k-12
curricula
couple of quotes from einstein
we cannot solve our problems with the same levels
of thinking that we use to create them
it says don't stay the big plane has
more or less ruined the planet you've got to find
a blue plane or a green plane that
has stronger thinking methods including
ways of getting the adult population
of the world to understand what's going on
and
insanity is doing the same things over and over and expecting
results that's what's been going on
london watching
europe and the united states go
through successive waves of
a pandemic which as a former biologist
can tell you has nothing interesting or
new about it from the time
they started identifying the crucial parameters
of it a country like new zealand
went with what it
actually is and prevented almost
all deaths they've only had 23
or 24 deaths whereas
the rest of the world somehow just cannot
get themselves to deal with
what's actually going on but instead they want to deal
they hope is going on and what
they hope will happen so this is a tragedy
now in the 60s the arpa community
was devoted to these
many of these ideas the climate
idea by the way goes first big warning
on the climate was 1963
by nsf so that was 57
years ago uh engelbart
uh
right here uh
was a big thinker about this and
known today it's unfortunately just known for
the mouse which he said hey that's just that's
just a button on the car radio we invented
a whole car so that's been a
real mess and what we're doing right now is absolutely
what they were doing when you see something
like what we're doing now they were actually collaborating
in groups
and i can't explain this but
you can see this demo
idea here though is that
the demo was the smallest part
the demo 168 was the smallest part
1962 engelbart
wrote a huge program
right before he was funded by arpa
and a lot of what he did can sort of
up what the whole community was about
here's the way he thought about
what you needed to do he said look
humans definitely use
tools but the problem
with the tool the problem with a hammer is it doesn't teach you
much except you can hammer things
you'll eventually use a nuclear weapon as a hammer
when you grow up so hammer learning about
hammering in a context-free way doesn't really
work and
what happens with this is that
why ends up building
rather poor heuristics
in our minds so the first thing engelbart
said look we have to have education and training
the more powerful the agencies
that humans get to control
uh the more different we have to make
reactions to them we can do that through
education and if we do that then we can use
powerful methods that have been invented over
the last 400 years and we can invent more
and we can come up with stronger and stronger ways
to represent our ideas especially now
the computer so to him this
quintuple here
was one augmentation unit the
human merged with these four
things and then the idea is
humans do that really count are
done in groups so what we want to do
is make groups of these augmented
human beings where the group itself
is augmented in a similar way so
that is a really big idea uh
doug was in despair
the last 20 years of his life
on saying doug don't worry just stay alive
they'll eventually get it and
he died and the world hasn't got it
so there isn't a system like this
except some a few specialized
ones for specialized scientists and engineers
but for the general public there's nothing like this
well let's
take a look at the world's greatest hockey player
why was he so great
hings they complained about was uh
his percentage of
getting goals was low but he got
han a thousand goals more than anybody else in history
and he said well you miss 100 of the shots
you don't take so you got to shoot on goal
worry about whether it's going to go in or not just have to keep on
doing it and then the big idea is
where the puck is going to be
don't follow it
go to where somebody can pass you the puck that gives
you and you're you've lined up with a clean shot on the goal
so we can look
at that as a strategy and here's a
little piece of history oversimplified
but in 1968
i was in grad school i was working on this desktop
computer called the flex machine this is
a self-portrait of it on its own
screen from back then
and around
october or so september october i met
seymour pappert who was
doing computing with children
and paper and i were both mathematicians
but he had he understood something about children
that tapped into some
very aware of but i never occurred
to me that it would really work with children
i never made the connection at all i was
till on the pink plane and when i saw what paper
was doing i realized he'd come up with one of the great
ideas of the 60s maybe of the
20th century and that changed
my idea about what i was doing completely
to the point where on the plane
utah i drew this
cartoon a blue plain cartoon
of what children actually needed
and it's not so much the tablet
idea that's what people home in on
but it was the idea of taking
bart was trying to do with
human adults and
understanding what augmenting children
actually means which means
to change the quality of what's going on
their ears even as they learn these
tools so that was huge so
that was one of those kerpaus
and here's the thing you can do when you have an
idea you think is good
look around into the future to
see if there's a vehicle that's favorable
like we had moore's law
a prediction that went
out towards the end of the century
odd years so take the idea
out 30 years and look at
it out there and say
30 years from now would it be ridiculous if we
didn't have this and my answer was well yeah it would be ridiculous
moore's law says we can have it
so we should start designing it now because the design
the hard problem well the way you do that is you
this idea back to 10 or 15
years out and ask is there
something i can do here and the answer is yes
and then here's the key thing
10 to 15 years out is just
dollars away from where we are now
meaning you can
by spending a lot of money now to make
a supercomputer you can
make the computing power that's going to exist 10
or 15 years in the future as a commodity
so you spend that money that's what we did at xerox
park thanks to butler lamson
and chuck thacker especially chuck who
made this super computer
and now you've
actually invent software
rather than just piggybacking on
the software of your day you can
do many experiments and if you have a super computer
wow
you don't even have to optimize a lot of code you can do
experiments to understand
new ways of doing user interface like the one we have today
if you do optimize you can do the
kinds of applications that will exist 10 and 15
years into the future such as microsoft
word which was originally done in 1974 at
parcc and so for computing people
the simplest heuristic i can think of
is if you're trying to do
future you have to
understand
that this
as nice as it is looks sort of similar to
something that we were thinking about a long time ago
it's nice that this is this is the not
not even the present this is the past
so when i go into a research lab in a company
or i go in to graduate school i see students
to do a thesis in the future
today rather than
using machines that cost probably eighty
or ninety thousand dollars
like i see futility
it's almost impossible
progress when you're mired
in the entire infrastructure of the day
we were really lucky at xerox park
because our funders and at arpa our funders
were willing to fund
the from scratch development of both hardware and software
that is a risky business
so you have to get some chops to
do it but if you can do it and your funders will fund it
that is a way of escaping from the pink
plane okay
i'll end with dunning krueger i think everybody
knows what that means it's a
these are two psychologists who studied
uh people who were
too stupid to
know how stupid they were
uh here's a nice cartoon
this is about elitism
the passages said these smug pilots have lost
regular passengers like us who thinks i
should fly the plane and of course we have a president
he united states who thinks
the plane and the passengers
of them we'll find out
in a month or two just how this plays
out but here's the big
idea the big idea is for all levels
of ability human beings
tend to overestimate uh
uh how good they are
we all you know it's actually a
heuristic it helps to be overly
optimistic can't just
cower in a cave but in fact it's
deadly and it's something that has
to be understood and dealt with so
occasionally there are
people who undervalue themselves occasionally people are more
right on it but generally speaking
kahneman has it right
that
gener generally speaking most people are completely
unaware of how poorly they think
including people who are professional thinkers like
most of the time i'm not aware of how poorly i think
but the big difference is i know as
idea that most of my thinking must
be poor must be poorer than i
think it is most of the time
my ideas are mediocre down to bad even
one of these kerpaus so just
that as a heuristic really helps
and the other problem with being really
smart and i've known some
work with some people who are a lot smarter than i am
some of them were so smart they
had the extra 80 iq points they were
so smart that they
just
tried to bowl every problem through just from sheer
intellectual ability and lots of times they were successful
and
some of the biggest inventions though came from people who are not quite
that smart they realized
they just didn't have the intellectual capacity to deal with this
this level of complexity and so they'd
invent a new kind of programming language
that would relieve a lot of the intellectual
burden
so the point of view equals 80 iq
it could mean a minus 80 iq points
not plus it's
whatever context you choose you can choose a really terrible
context okay
let's do the q and a and
um i more or less used up my
think yeah but we started
thank you thank you alan so
we will see how many questions from the
queue we can accommodate
um i start my video thank
you so the first question is which cognitive biases
most actively working to counter
in your own life decisions
well i
the one that has always dogged
me is uh
not feeling
of my results is
o i've always been down on uh
my efforts and a heuristic
that i came up with uh that
works some of the time is whenever
feel like
i'm not doing well
enough i asked myself uh
well uh what's what's the quality of your effort
and
if the quality of my effort
if i couldn't be putting more effort in
to the thing i start feeling better
because
quality of effort is something you can control by willpower
but the complexity of the universe
means that you can't control your ideas
to be as good as
okay that really leads to
question what approaches do you recommend
prove that an abstract idea actually works
i guess that to prove something to the research community
we end up categorizing ideas like matter
rather than light well
so my research community uh
was
first it was made up of former
scientists uh in
the mathematical physical scientists sciences
and from [Music]
you know established deep engineering
so there were no undergraduate
degrees in computing
back then and this helped a lot because
all of us had been through the rigors
of dealing with highly developed fields
uh with complex
theories and complex ways of trying to
understand the theories and vet the theories and
and of course in engineering uh
one of the ways you vet things is by building models
and then you build actual bridges and
the hippocratic oath in engineering
is the the plane must not crash the bridge
nut must not fall
so the arpa and park research communities
basically
didn't spend any time arguing with anybody that
was what was nice with the funding that we had
and our view was there
that was hugely
interesting about what we were trying to do
that you could prove mathematically
although we used math in a variety of
ways and that the only way you could
vet your ideas was to actually build the artifacts
so that little
personal supercomputer the alto that we built we actually
built 2 000 of them and in today's
money they would each one of them would have been about
120 000
uh xerox uh paid for those
uh ivan sutherland did his thesis
of inventing computer graphics and
much of personal computing
in the early 60s on a
sage air defense computer that costs
uh probably 20 or 30 million dollars
and arpa gave him time on that
computer so
so in the world i came from
uh you basically built anything that
uh you uh thought was interesting
and particularly
uh for the personal computing aspect which really was
arpa was about and networking
you had to build personal computing
means user interface no matter
what else people think it is in
order and user interface is something where we do
not know enough about human beings
to be able to design abstractly and
have uh the interface
work and i can tell you that i
think it is justifiably the case
that several of us knew quite a bit more about
human beings back then
just because we had studied
them we knew
more i think than most people who do user interfaces today
but in fact we had to do hundreds of experiments
okay
thank you uh maybe we can get one more question
the answer a bit short
in order not to over stretch schedule
too much the question is is the
innovation culture quote unquote something that
engineered or is it something that
from the right incentive structure within a community
i'm particularly interested in transdisciplinary
collaboration yeah so
the first thing is uh you know we're
using these hard category words innovation
is a terrible word for what we did
at arpa and park
and this was pointed out in the 80s by regis
mckenna who was a silicon valley
figure he said apple is innovation
xerox park is invention
innovation
is complex it's difficult
it's expensive it has
things but it is uh taking an
idea and doing the packaging
it's basically incremental
to a sea of ideas that already exists
otherwise you put it trying to load too much
into innovation which is what has happened today
you're much better off saying
yeah there's this idea of invention
where even inventions come from previous
things but they're much more startling
than most of the things we call innovations
today and yes and you
can do an innovation culture
that is precisely what arpa did
and uh there's a great book
called the dream machine by mitchell raldrip which is
the history of both arpa and xerox
park it's the only really good one i know of
i've written you can find some papers
i've written about why i think the culture
worked i just did one
for the ellen macarthur
foundation over here last year which
looked at these large efforts that
required a lot of invention and had
to be done relatively rapidly by top
people with unlimited funding and i picked five
or six of them you know the radar effort
the atomic bomb effort the code breaking
effort the
sage air defense system and then i
use these as a background to look at
the principles that came out of the
park thing and you know
community and invented so many things
that it's impossible
to say well this is just a fluke
right there's actually principle
in there particularly at park where taylor
who had been an arpa funder put
into work at park what he thought
made arpa successful and he would tell you about
it if you asked him so yeah
can do it the problem is funders today
confuse being responsible
with
the feeling that they should be controlling
things and in
edge of the art research that's a disaster because
funders have not been doing edge of the art research all
their life they've been accumulating money in one
way or another they're the worst people to try and do problem
finding so every time that somebody
has tried to do a park-like place
since then some silicon valley billionaires have tried
and some seattle billionaires have tried
it's always been a disaster because
want to play and they forget
well hey you never did any of this stuff
why do you you know you don't want to
a wannabe you made billions of dollars but you just
don't have the skills to do this far
out stuff you don't know how to think about it just
put the money in and go away
and the people who can do
found and they will think of new things
that nobody else ever thought of and maybe
30 or 40 percent of them will be successful
and if they're working on a big enough problem like
we were that 30 or 40 percent that
successful will change the entire world
and in the case of arpa and park it brought
close to 50 trillion dollars of new wealth
so it's actually a good investment it's the largest return
on investment on research and history
but the funders today will
it because they can't deal
with spending five or ten years of
learning curve just to get a factor of
more money they only want millions
and billions they they can't stand the idea of trillions
so if you look at the look at the comp
anies on the uh in
the stock market that have trillion dollar evaluations
today every single one of them you
t of the technologies that were invented
by the stuff that no funder will fund today
thank you alan i think this was a very
important and very deep insight
o summarize your talk thank you very much
for the inspirational talk uh we enjoyed it very much
sorry that we can't handle more questions
i would like to invite you to join some of the other activities
the week of the heidelberg laureate forum
look into the vr space and meet some people
whatever but for the time
much for joining us and uh hope
to see you in person sometime in the
future bye-bye
uh we will now switch to the next
which is a completely recorded uh event
uh it will be given by david silver
but i shouldn't get ahead of myself because i will
introduction in this pre-recording anyway
no pause here we just continue
[Music]
stay
[Music]