Alan Kay Interview for nerdear.la 2020

From Viewpoints Intelligent Archive
Jump to: navigation, search
living organisms can survive in their own waste products
foreign
hello everybody uh
thank you for inviting me to talk
i thought
my bio
is very very brief
pretty much everything i did uh [Music]
came from my research community
and
nobody owes more to it than i do
i'm not going to talk much about the research community
xerox park and
arpa but occasionally i'll refer
to it and
the
talk today
really is in three short
parts because i would like to get
to the question and answer session
i think that would be the most interesting for
for all of us i'm going to try and
get through the talk uh quickly
enough to leave time for questions and answers
and uh
maybe some of the things i say will help
with the questions and answers so
three parts are
about pollution
the kinds of poor thinking that humans do that
gets us in trouble especially
software and other parts
of the world how to find better
than the ones most of us
pick and then we'll get into the question and
answer session
so the first uh
idea about pollution here is this very old slide
that goes all the way back to the 1950s
of a cartoon character named pogo
looking at all the trash and
he says we've met the enemy and he is us
and that's even more true today
because the amount of trash
we have is
much much larger in fact
we can see that
just one of the trash cycles
in the middle of the pacific is almost the size
of the united states
and this is
just the surface that we can see
so they estimate that perhaps 85 percent of
it is below the surface
and whenever i see pictures of this it kind of makes
me think about software
it's just a really
immense amount of it
bigger than most countries
and one of the reasons we have it it doesn't
seem like you have to do any real thinking
to use
trash make trash throw it away
put it somewhere else and
it just gathers up and gathers
and of course computing is that way
most of the software that's made in the world
is not done thinking about the future
in fact most of the costs of doing software
are because of what wasn't
done in the present to
make the 85 percent of the costs
of software in the future much much lower
instead everything is done right now in the present
it's done too quickly
and we wind up with this immense amount of
trash and of course the software doesn't go
in the middle of the pacific ocean
so it's perhaps it's a little bit
more like a slum that's
the size of the united states
perhaps we're living inside
of this
software trash
or slum that's still running
most of it hasn't been turned off
and there are many
go with this i'm just
going to pick one for this
talk so one of
the historical reasons for how we get
into these kinds of troubles is that in
our genes like many
animals many mammals
especially primates our
natural instincts are to tinker with things poke
see what happens if i do this what happens if
i do that and not worry about the consequences
it was hard to get hurt 200
000 years ago and
so trying this and that
's natural it's not something we want to get
rid of if you don't know what else to do it
is kind of fun to tinker
but take something like clay which
resembles a computer in that
you can make the clay
go anywhere you want you can push it around you can tinker
it forever but it's
really hard to tinker it into something great
you just don't get
by it tinkering
into something that's wonderful
like this statue of
uh voltaire
what you get is kind of a mess
to get voltaire you have to
learn a lot of things you have to think about a
develop skills
what you get when you tinker is
omething like this child making a little cup
and when you see something like
this with software
you can complain and say hey
that's not good enough and
these days what people say is well this is
a start when i complained about the world
wide web and especially the web browser
more than 25 years ago i said well
is just a start this is a start we'll we'll make it better
in fact they didn't
many of the things that were wrong then especially
with interaction and
end user creation of things
are as bad today or worse
so what you get when you start off badly
with a tinkering frame of mind is
something that looks like it was tinkered
into existence which is great when a child does it
but you really don't want adult professionals
to be putting a zillion of these out
into the world so if we come back
to tinkering historically we should ask well what came
after tinkering and the answer is
engineering which is
making things using principles
so by trying things out you find ways that
work you remember those ways that work
them up they may even be literally
a cookbook or they're like a cookbook
and then you have a set of things that
are likely to get you faster
where you're trying to go and to get you something that's
better now you still have to make the thing
so making is part
of both tinkering and engineering
but basically in
any real field
that involves making things we want to be able to move from tinkering
engineering so that the goal is engineering
and of course even bad engineering has
principles there are bad ways to cook
things using principles so we also have to ask
are the principles good enough
and after engineering
thousands of years after engineering
the kind of mathematics that we recognize today
was invented about 2500
years ago and it also involves
principles but the principles are about reasoning
and representation
and about 2 000 years later
what we call science today
was invented
one of the most important inventions of all time
because its principles
are about how we negotiate with
the bad stuff inside our heads our brains
ask the deal using
the kinds of belief structures and
memories and everything else we have in here with the phenomena
that we experience in the world
and so it winds up being a negotiation
and it is a deep skill
it's not something built into the human race
it was invented very recently and it's
one of the most powerful things that we've invented
it's also not understood very well
and people who are good these days
work
in a sweet spot
of all four of these things
and if you think of yourself as an engineer
these days you still know mathematics
you still know science you still know
tinkering you think about yourself as a scientist
you still know mathematics you still know engineering
so where you are on this scheme
depends partly on your personality
but basically the people who are really
good use all four of these things
and try to choose which ones are the most
reasonable at time to time so
get into this sweet spot is part of the
idea of being a modern
practitioner in a
stem field now
if we look at computing and ask where is it
it doesn't look so good it's
mostly tinkering still today a little bit of engineering
tiny about a bit of math and almost
no science and
most of the engineering is in the hardware
why is that well it's
get away with murder
with hardware software
people
can make things that are quite dangerous
but in fact because they can be debugged and fixed and patched
and so forth they get away with
much more than any piece of hardware
can so much of the engineering and computing is
in hardware and the vast
majority of software people
think it can be done without learning about engineering without
learning about mathematics and without learning
science and that is
kind of where this mess comes
from now when i started
off programming i started
off like most everybody else
which is not knowing anything
uh important about it i happen to start
a long time ago around 1961
in the air force
but i think my situation back then
was similar to what it is today and that somebody else was
choosing for example
what computers i worked on
they were choosing what operating system in
the case of this machine it didn't have an operating system
which in some ways made life simpler
they chose the programming language
for this machine it was assembly code
development system well it didn't have one
the design or lack of design was represented
in flowcharts back then
done by other people
the legacy system were huge
punch card systems
and the job for us
coders coders were
people who turned flowcharts into machine code
so we were essentially compilers of the higher
level language of flowcharting
and nobody cared
my opinions and i also didn't know
much so that was a situation of course
after after this i went on to
do slightly higher level programming on
supercomputers still mostly in machine code
still with
other people choosing the machines and the choices and so
forth so
now it had happened that i had gone to an engineering
high school and
i had gotten
a undergrad an undergraduate degree
mathematics and another one in molecular
biology so i knew
something about engineering i knew something about mathematics
i knew something about science but it
just wasn't in the computing part of things
which i used as a job to put
hrough college so i was a journeyman
programmer then in 1966
i accidentally wound up in grad school
my theory was spend a year
learning something about computing and avoid
getting a real job and avoid going to grad school
in math or biology and by
accident i round wound up in this
advanced research projects agency
research community that by 1966
had already invented quite a few things they didn't
invented the first parallel computers
real-time computers with displays
and pointing devices they'd invented
the first interactive graphics system
of a modern kind time sharing
uh artificial intelligence
there's already a first little personal computer
the tablet the mouse
hypertext and
they were starting to talk about uh making a
packet switching network which was called
the arpanet they hadn't done it yet
so that was the world that i found myself in
and it was full of people who were
completely unlike the people that i'd worked
with in the air force the people in the arpa community
they all had extensive
of them uh phd's
in engineering of some kind
like electrical engineering or mathematics
or science
and what they were trying to do
take what they knew about these developed fields
these difficult fields that have been developed
over many decades and hundreds of years
and to see what they could use
from this experience in this new
world of computing and
so pretty much everything i had learned
as a journeyman programmer was worthless
the way i went about doing it was worthless
so i wound up having to learn
again but this time
from this point of view of engineering math
and science that this community had
community not only made
its own software made its own programming languages
but it also made its own hardware whenever it needed it
so it's basically full spectrum
computing and in order to get
any kind of degree there you had to learn how to do this full
spectrum computing of hardware and software
and there are some
really famous people who will
be known to you today perhaps the patron saint
of programming don knuth
was there and he his
degree was in mathematics
and turned himself
into what i think we'd all think of
as a great programmer i thought he
was is still hanging
in there doing well and
one of the things that he tried to get people to understand
is this saying premature optimization
is the root of all evil
and it this was a
hard thing for him to say back then
because you could hardly get any computer back
in the 60s to do anything
without optimizing they were just really
slow like they were a million times slower
than your cell phone and
more than a million times smaller
but don pointed out that
most parts of most programs
doesn't matter how fast they run what you
really want to do is get the program right
identify the places that run too slowly and then find
a way of optimizing those
without disturbing the fact that you got the
running and
there are many other sages back then here's
bob barton um my
world's greatest uh computer designer
designer of the furrows b5000
he was a mathematician also and
the things he like to
drum into our heads is that good ideas don't
often scale and of course there
are many other principles from back then
i could make a a
three-day course out of just the principles
that these people used in the early
60s but for this talk i thought i'd just
stick with don's idea that
premature optimization is the root of all evil
and ask well what is it that we should be
thinking of then because we still want to
do something and if we can't optimize
what is it that we should do
well one thing we can do is try designing
that's a new thought to a lot of computer
people they just jump in and start
writing code and think about well we'll design
it later but it doesn't work so well on the other hand
you have to write something because it's
really hard to design a whole system before
trying things out it's too complicated
but you start designing
if we are not going to optimize prematurely
somehow have to separate meanings from optimizations
and this is something we can bring up in the question
and answer thing of what are the different ways of doing
this what does it mean
to separate a meaning from the optimization
for now let's just think of it as a meaning is the
write no matter how slow
that can be debugged
and
allows you to know that you
this meaning captured so like if it's
if your idea in one part of
the thing is you have the idea of sorting
then your little module layer
would be the simplest way to do a sort
it could take forever to sort but it guarantees to be
you something that
you can use for testing
things and also you can use for testing out the optimizations
and once you have the system running no matter how slowly
you can then carefully add in the optimizations
and you can add them in in a way that they
do not pollute the meanings in fact you
can use the meanings to test the optimizations
and by testing
the the optimizations you
want to constantly be able to turn the optimizations off
see whether this is that part of the system
still runs or not turn the optimizations on
see if you get the same results
you can have the meanings and the optimizations running
at the same time until you trust the optimizations
and so forth so there's a whole scheme of things you can do
here and in fact
these kind of things are done in regular engineering
disciplines what i call real engineering
so the designing and separating
from optimizations is called computer-aided design
in regular engineering
these lead to simulation
so you want to simulate
before you commit to doing
the much larger work of packaging
so you have to have a way of simulating what does that mean well we
can talk about that when you
add the optimizations in you better add them in through
the cad tool and then
let the simulator worry about how you
test the optimizations against the meanings
well you have real-time
requirements and your program
might be running too slowly so you just find
a super computer this is what the rpa community
did maybe these days
as you're purchasing super computer time online
from amazon or google
any organization that is targeting
things like this should be paying more
upfront in order to make their actual
program development time uh faster
with
and finally when you get everything going what
real engineering does is it starts thinking about fat
and fab sometimes
requires some additions into
the cat for instance when you're
making a bridge or a
table where you have two beams
connecting each other you might have to put a gusset
in there to help the connection
itself might not be strong enough
so the little things that you have to add in there but you want them to be
minor okay so
that is a typical thing that we find
over and over again in real engineering electrical
cad it's the way everything is done today
nothing electrical
especially no computer chip
is done without completely simulating every aspect
of it before it ever even goes to
masks and silicon
every form of mechanical cad
mechanical engineering
this and the
simulations are particularly critical
because in a jet
are many parts internally
in a jet engine that are actually
their melting point is
temperatures that are in parts of the jet engine
it's really interesting how you make jet engines
to allow these things that would melt
not to melt uh
biocad so here's something that's come along
after computing and
yep the biologists knew enough
to for bioengineering to make
real cad systems with real simulations to allow everything
to be done notice that
all of these things are done using computer
techniques computer displays computer power
to do all of this
so the other engineering fields are using computers to really
help them develop
nanocad
but when you look at
in if you go into most companies and look
to see how people are developing software
you don't find anything like this
what the
what is it whatever
it is it doesn't look like engineering and if you look closely
it's really a
set of facilities for kind of tinkering around
kind of organizing thing but it's mainly aimed at fab
it's much less
aimed at design
almost never almost never do you find
sim as a key component
it's mainly aimed as though
yeah we can do everything right
we're smart enough to be able to write the end
code and have all of the stuff and intertwine
the optimizations and so forth
so this is kind of a yikes
now
what if you don't have the cad
tools most people don't for software
partly because they don't exist but even
the great development systems many people don't develop
using them well what if you don't have those
and what if you don't have a super computer
and the lesson from the past
is you should still do these things
every single one of them you just
o put the optimizations
in more carefully and a little bit
earlier but you still must do the meanings
first so this is a discipline
fact the difference between
gineering and math
and science is the three math
engineering and science they're disciplines
disciplined when you're actually using them
so why don't most
computer people do this well let
me there are lots of reasons but
i'll just pick one is part of it is because our brain
really doesn't want to do all this work
that's why it's a discipline so
we're actually poor thinkers
and we have about 200
known ways that we think poorly
so i'm just going to pick one
way one little part of this
very complicated set
of bad machinery we have between our ears
and i'm going to use
as an analogy
our thoughts or our context as
like a flat surface
it's colored pink here
and our thoughts
will can move around
the surface like an ant crawling on a
table we can pick different directions
if we find an obstacle
we can figure it out how to go around it
so notice everything that we're doing
here is like what we call thinking
but every single thing is
pink and we don't know it's pink because we've never seen
anything but pink so this is a context
that we think of as reality rather
than something that we believe in or something that we've only
been brought up in
okay so
the other thing is uh
when we're seeking goals we tend to cling
on to our goals very very strongly
so if our goal is b
and we will often pick b
in there
so let's try and get to it
we'll start working on it but
in fact the problem space might not actually
flat it just seems flat to the ant
so we start moving towards b and
all of a sudden it starts getting really difficult and then we slide
into a ditch
why are we having so much trouble but we keep on fighting
towards b partly because
in school we've been told
we must solve this problem and
problems are usually given to us by other people
why do we cling so
strongly to these things well
let's go to africa here for a second and
the way
he's going to catch this baboon is he digs a hole in this
termite hill there's the young baboon there
watching it you
put some seeds
nice tasty seeds in the hole here
then he just backs off and
wondering what what's going on here
and that's curious remember baboons are primates
like we are
now looking around well the guy's over there
maybe i should take a look
now when he gets close
he can start smelling the seeds
reach us
in and he's got the seeds
but now we want to let go of the seeds
you can let go
anytime but he won't let go
so the hunter just comes over to him and
ties him up pulls
him out and he's got the baboon
so let me just tell you i won't tell
what what happens after you can find this
uh movie on youtube
and
the baboon does not get hurt
he hunter had a really interesting reason for wanting
to capture the baboon and you'll find that out
but for here uh what we need to understand
is the seeds
this
cognitive uh glitch
which we have baboons have
and many other animals have
it's called loss aversion once we have
something we don't want
to let it go it becomes much more valuable
more valuable to
this baboon than his life because the
hunter might have wanted to kill him and eat his brains
ame thing
with us that goal
becomes more valuable to us
sometimes in our life
and often uh
prevents us from actually solving
problems uh that need to be solved
well if you knew more like
actually a an explorer you know
well when things get tough maybe
we should re just explore around
maybe we don't have to go up over this
hill and then down the valley
and back up in the hill and stuff maybe there's a way
around and this seems like well
we're going we have to go away from the goal to do this
yeah you have to go away from the goal
going to take you a longer distance to get to the goal
but in fact it might take you less time
and less effort to get to the goal
in fact by looking off to the side boy there might
even be a super highway i've colored it blue because that's
my color for science science
is a super highway again you get go off out
of your way get onto this thing you can go three times as fast
and you get to the goal faster
but if you start using science
you might
decide to invent a plane just
fly over the darn thing
but once you start doing science
you've got this even bigger idea is wow there might
be a non-pink world
there might be a whole blue world up there and
only in this blue world is there a goal
c that's the one i really wanted
i didn't even know it
and by the way if you think
about this in terms of your education if you went
to a a really good university
then you probably
went in there with some goals like b a
really good university will change you so much
that it will show you things like see
and your life will never be the same again
so this idea of getting trapped
in a context
and the idea that there are other contexts and
it takes some real effort to get out of
seems real to these other contexts
that might be more powerful this
is the message from this section of the
talk
so how can we get
in the frame of mood to try and find
these bigger goals
these bigger visions
and again this is a huge area
the things uh i was
interested in still am
is the idea of personal computing
back in the 60s we used to say to people well person
is the first word in personal computing so you
better not try and do personal computing if you don't understand
uh persons and of course
people today do not understand
uh people at all and so the
most of the apps that are done are
have terrible user interfaces and
are actually not very good for people
so this is a first lesson but in
fact this is a little bit of a red
because a person all by themselves
is being punished
solitary confinement banishment
we actually exist in a
society of people and
so when we talk about personal computing we have
to talk about humans in a culture
we have to understand what humans
are
that we share with other humans the thing that makes us
human and that will give us great insights
into ourselves because it's
hard to look at ourselves when we're in the pink plane
and just see anything but pink
and in a society
well we've got duties like voting
we have duties to
the next generation whether or not we have children
every
culture has its own world view we need to understand
those and get more powerful ones
every culture
has some form of schooling we need to pay
attention to that there's this idea of richness
richness is
outside the purely
pragmatic we have large
respond to art and love
and friendship and
many other things that are not strictly
pragmatic we need to
care for those and then
we have the thing that most people are worried about way
too much which is livelihood how can i get
a job many people go into programming
because they think they can make money at
it and this is a very bad reason for going into programming
you really shouldn't if you're not going to go into
programming to try and improve the world you
should try and look elsewhere
and these things
here these seven
things are part
of especially in our day and age
are part of a much larger context of what
you might call grand deadly
important issues
i just have 12 here and you can think of more
these are the things that we
worry about also
and these things are global now
there's a large enough population
of the cultures
have enough power to affect the entire world
and are thus affecting all of these things for everyone
pandemic now
that is now worldwide and it gives
people perhaps a bit of a sense although
i think people still are not taking it seriously
and they certainly are not taking the climate seriously
enough because
this world we live on is dying and we
so a child born in 2020
is going to be 80
at the end of the century and you wonder what what is
going to be left for this child
but if you're taking the point of view
that i'm advocating here the other way to look
it is wow 80 years
things could be much better for everybody in the world
80 years from now if we just took all of
this seriously and learned how to think better
and the context is even bigger than the world
it's only been about
100 years or so since we realized that the universe
is vastly larger than our own galaxy
and it's only been a few hundred years
since we realized we lived in a galaxy
we are now aware of most
uh people on the planet in the form
thousands of different cultures all with different beliefs
now all in communication
through our technological structure this is a self-portrait
of the internet but it stands
for all of the technology that we've made
over the last few hundred years
and we ourselves
are a complex system
and our brains are
even more complex systems
so one of the contexts here is this idea
that what we need to be aware of is
ystems that we live in and the systems that we
and these systems are all interconnected
and intertwined so
most of them are invisible
so the four ideas here
first science
is not just about finding out about how
atoms work or how photons
work or how stars work
science the larger
idea of science is the set of heuristics
with to get around with
brains this is the way to
this is why everybody should learn science
systems
is one of the strongest ways we have
to think about complexity and almost
nobody learns systems it's not taught in classrooms
at least in the united states and
over here in the in the uk
and most computer people that i've talked to
outside of special
community most computer
people know very little about systems they think
programming is creating an algorithm
rather than designing
a system and being part of a system so this is
a very weak view then two ideas from einstein
we cannot
solve our problems with the same levels of thinking that we use to
create them this is true of the small in computing
the way people have been going about computing
are just going to make things worse the way we've gone
about health the way we've gone about climate
it's all going to make things worse if we
persist and einstein has this nice
definition of insanity which is
over and over and expecting
that's what we're doing and again you can
if you think small in computing that's what we're doing
computing is hardly different today
except in scale than
it was uh 60 years ago
it's really a shame
the scaling is just making things worse
because almost anything that was done 60 years ago
couldn't handle
scale very well we had to
invent entirely new methods to do things
just even like the ethernet or the internet
to make personal computing and the user interfaces
scale across billions of people
so scaling is another issue that we could be
talking about
so we can wind up here
talking
about the world's greatest hockey player
his name is wayne gretzky
a thousand more goals than anybody
history and he was just a little
guy wasn't big he wasn't tough
took a lot of shots on goals his his
his percentage of
getting goals for shots wasn't very good and some people
complained and he said well you miss 100
of the shots you don't take so
his idea here is you you have to show up and
shoot on goal you can't
worry you have to do all your planning ahead of time you
whether any particular shot is going to go in you have to keep
shooting and
he said well a good hockey player goes
to where the puck is a great one goes to where the
going to be and he didn't mean tracking
he meant getting to a place on
the ice where some teammate
could pass him the puck
where he would have a clean clear shot on the goal
so he would could understand
the patterns of uh all the players
on the ice he could see well if i get over there
somebody can get me a thing and then that will give me a shot
on goal there so this is a completely
different way because most sports especially
sports like like hockey or football
are basically tactical whereas
gretzky was the greatest
player in history because he was strategic
and
he furnishes a good analogy for
an example that i'll end with here from
so the first
idea is uh
you have to
have ways of coming up with ideas
and as i mentioned in the beginning my way was
by being embedded in this incredible community
that was full of rich visions and
interests in helping people so i didn't invent
the idea of personal computing but this
is what i worked on for my thesis
in 1968 and while i was working on it
i
ran into seymour papert who had
been he was a mathematician
like me and he had realized some really important
things about computers
and how children could
learn mathematics by taking advantage of what
could actually do and when i met
him in 1968 what
i saw was something that i understood
and understood beforehand except i didn't
understand it i was in the pink plane
and pappert was in the blue plane
he had seen something that was right there in front of
all of our eyes that combined
what he knew about children and what
he knew about computers and what he knew about mathematics
that provided a
glimpse into a complete revelation
about how
uh children's science and math education
changed and that just
completely blew my mind and so on the
plane flight back to utah i had a blue thought
not a pink thought the
oh this makes
computing really important makes it
more like reading and writing and if it's like reading and writing
children have to do it and pappert is showing us how
and that means children have to have their own
computer and it better be really portable because you don't
confine them inside they need to be able to use it outside
and arpa was working on uh
besides the arpanet working on a wireless version
so it would be connected wireless it would have a flat screen
display and so this cartoon
is what i drew on the plane back to utah
and i started
thinking about it and
community the thought process
went kind of like this well
you have this idea seems good
might be bad but one of the things you should
look around and see if there's are any exponentials that are going
to help in the future it's not possible now
so you can take the idea out 30 years
like to 1998 or 2000
and ask what does this
idea look like 30 some odd years from now
and the answer is oh yeah absolutely going to happen
moore's law will guarantee that we'll be able to
do this but we don't know how people will use it we
don't know what the user interface would be like so then the
critical part is bringing it back uh 10
to 15 years
bring it back to about 10 10 to 15
out in this case it was like 1985
1986 because
when you can get something within
10 to 15 years you can bridge the computing
gap by just paying money
you can build the
function of a computer 10 to 15 years
from now you can build it now
it's going to cost 10 20 times as much
it's going to be 10 or 20 times too large
but you can build it and you can build a bunch
of them so that's what we did we built
a little supercomputer for everybody at park in fact we
built about 2 000 of them and this
allowed us to have
a window to invent the software not
just the the operating
interface but a whole bunch
of the software that would be usable in 10
years or so from then
and this computer is
fast as what you could get from
time sharing interactive time sharing and allowed
us to do two things
there's a whole bunch of stuff that we didn't have to optimize
we could just program the meanings
of user interface ideas hundreds of them
and do a dozen experiments a day
is what we did to invent the
part gui that everybody uses today
the other thing we could do is by optimizing
uh
in the both of these in the way that i mentioned
uh that allowed us to do the
applications of the future uh
in 1973 1974.
this is microsoft word
as it existed in 1974 at xerox park
than 10 years before it appeared
uh commercially but it was quite
possible to do because this machine was actually more powerful
than a mac or an ibm pc
of 1984 85 or so
right so
the simplest way to think about making progress
here is always find ways of computing
in the future and even
if you don't have a super computer there are ways you
in the future because you can make a future
architecture just using
software alone that is going to get you much further along
okay
made it to the question and answer and with that
i'll turn it back to my friends
for the next phase of this
talk thank you very
much
is
in
can you both hear me okay yes
yes good so again
being with us alan and
so as i was saying i am not worthy to take this conversation
so i will just let you talk with
with hernan okay i'm not
worthy either but anyway i'll do it
no no
no thank you thank you very much alan it's it was
amazing the talk i i always enjoyed your talks
uh it's a new way your worth
ways of thinking and new ideas and
and that's great not too many people do that so that's
amazing and i like to start talking about science
because that's something that i know you like and
you talk about it also and uh you know you
always encourage scientific thinking in
in people and i remember uh an example
that you gave in the squeakers dvd like 20 years ago
to quincy jones where you
you showed you know you told us that we science
broaden our vision and in that
way we can you know see beforehand
going to happen a long time from
from now like for example aids and so on
um but in your
that we don't understand science very well
my question is why is that why it's so difficult for
understand science and to think scientifically and
how can we change that well i think
if we look historically
and if we think about science as
what we think of as modern science
we it started really uh
400 or maybe 450
years ago now the greeks had
scientists but they didn't have
science because science is not just
individual scientists but it's also a community
that helps debug fondly held
notions that
people as human beings have
right so zionists are humans as well and
so um they like their own theories
and
one of the one of the hallmarks of the existence
of science is debugging
of ideas and even
ideas that seem to be backed up by experimental
evidence the idea is it's not just
about what one person
thinks is going on or tries to demonstrate
is going on if you go back historically
we can trace back humanity at least several
hundred thousand years we know that
the uh the female line
of humanity goes back 200 000
years and so anybody
the question well how how could it possibly
have taken us 200 000 years to
invent science what was so unobvious
about it and part of the answer is
uh our social
uh cultures
relied on
storytelling so regular
language convolved
with storytelling and using stories as a way
of remembering things and explaining things
and a story is like math
it can be consistent
but it doesn't have to have anything at all to do with the real world
and the other part
of it is if you look at optical illusions
the thing that anybody
who experiences an optical illusion has to
realize is oh i'm not seeing what's out
i
just think i'm seeing what's out there but actually what i'm seeing is
something manifested inside my own head
and that's why i'm seeing some sort of parallel
you know the simplest one is just measuring
holding up your thumbs yeah and i can see
in the in the video that the closer
one is about half the size
but in fact what i'm experiencing is the closer one
is about 80 percent of the size
because our brain knows they're the same size
and so i'm seeing something about a third of
a second late that
is a manifestation of a combination of my
beliefs and understanding with some input
from the outside world putting together into
a kind of a story which is presented
back to me as reality and of course
the though right now we have a perfect
example for the world to see in the
american president that we have who basically
like all humans but
in a very public way is projecting his
beliefs out onto the world
and taking that as his reality
and that's basically what humanity is all about
you see it in computing all the time
also if you just look at the comment section in
any slash dot or
this stuff you see people constantly
projecting their own beliefs
and desires as reality
so it
also happens that this is the 400th anniversary
of one of the major
starting places for science
so in 1620
a guy in england here by the name of francis
bacon wrote a book
uh called the new organization of knowledge
novum organum scientia
science actually meant
knowledge and
the gathering of knowledge it doesn't didn't mean back then what
we mean it today and in this book
he points out that humans are terrible thinkers
in a many many ways
he picks four his four favorite ones
are were terrible because of our genetics
in other words our brains weren't made to think they were
made to do something else
our cultures
weren't made to think they were made to survive
our languages were made for
stories not for representing
ideas very accurately or being able to use
to help think and then our academics
are teaching
uh facilities uh
will often teach ideas that have long
been debunked
we are doomed basically
yeah close close to it so
so what uh what bacon called for is what
he said was we need a new science
and what he said is what what science is what
this new science is a new way to get knowledge
is basically to come up with all the heuristics
and methods we can come up with
to get around with what's wrong with our brains
so this is the big idea about science
and it's not taught in any school in the
united states that i know of no we are not even
close to that i mean yeah big idea about science
is not about uh
no life on mars and
certainly not what you find in a science museum the science
museums in the us and the uk have
hardly any science in them at all they're full of technology
which isn't the same thing at all
but bacon's idea was much larger
so just to make everybody feel better or maybe worse
the
this 400th anniversary of some of the most
important ideas of the last 400
years has not been mentioned once
in any of the british papers
so the the people in the uk are as innocent
of scientific knowledge and history
everybody else in the world and they have no idea
that science actually for real got started
what happened to us in our field
in this software development with you know the the
people that created the development is has
even a bigger problem which is
because uh
when you start off with software
you're starting off with something that probably isn't
going to kill anybody so it's not like
building a big bridge or building an airplane
or something and so you don't really
have to know very much about engineering
write a program you certainly don't have
anything about science to write a program
and you don't have to really know anything about math to write a program
yeah so
uh because simple programs will still do
something it's a little bit
more like this game that was around for a while called
guitar hero yeah where you
could pretend to be a guitar player
and things with things would happen and
a lot of the attraction of programming uh
when the 80s started up and the attraction of
the 8-bit micro computer was
just to touch it
just to feel part of this thing that was happening
the problem is none of the content happened and this is very
common in pop cultures
so pop culture will develop its own music
which is usually much much simpler than
uh develop music it will develop
its own notions of knowledge
and science and you can see it uh
[Music] all over the world
through uh social media
right so this is a universal
publishing system as
viewed by uh
by the majority of people who
haven't had the good fortune to undergo
education in basically the 20th or the
21st century so we have this enormous
culture that
is not very far
really from the middle ages
and we can see that by looking at
most of the world and most of the world's leaders
have reacted to uh the covid
think of it yeah that's right please because
anybody who's actually understood
an eighth grade biology course
these days those
should know exactly what's happened there's nothing that tricky about
it real question is
how infectious is it really but
as far as what a contagious
deadly disease without a cure can
do it should be something that every adult
on the planet
at least in the first and second worlds
should be able to respond to well
now new zealand did a great job yes yeah
they had the right kind of leader the leader was
able to get business people and the politicians
both parties together and get everybody to agree
to this how she did that i'm not
sure but he did
do it and it results in there so if
you prorate new zealand's 25 deaths
with a population of 4 million
you can see that almost every other country
in the world has been needlessly killing off
tens of thousands to hundreds of thousands
of people uh unnecessarily
just because hardly anybody is educated
enough in something that you learn in seventh
or eighth grade biology so because business
life basically
well i don't think they think of it that way you don't
no they just don't see it i
think if you held a gun to the head of a businessman
that said would
you rather live or would you rather
uh stay in business i think they
choose life most of the time most
of the time yeah most of the time there's a famous
there was a famous joke which i
is told better in the u.s okay
which i won't won't tell here but it's quite funny about
that so no i think the big problem
the
lack of imagination
so if you think about what that our brain imagine readily
well we can readily imagine gods
we've never seen them but these are things
that are part of our subconscious
and so and we have experienced them in
dreams so we can imagine things like that
more readily than we can anything that
that is related to science or to think scientifically
yeah if it's if it's really small if it's really
fast if it's really new
and this is where bacon comes in because what
is okay or or
sentence here so we can go okay
next question is yeah yeah so
you know the term artificial intelligence is
used widely and very loosely
these days
but if you think about it what is it what is artificial
intelligence well it's artifice
it means making something you have
so
somehow you're trying to make some sort of a
process that is going
to act intelligent and
impressed with any such process
unless it's more intelligent than a human
and so if we look for something on
the planet that is an artificial
process that's more intelligent than any human
there's only one and that's
science itself
science is a better scientist than any scientist
and science
itself is that process
that makes a group of people much more intelligent
than human genetics human culture
human language and human academics and
so uh
so this should be a simple idea
but partly be
because of the way education goes
uh science has been
relegated to just another belief system
yeah and it is a belief
system but it's a belief system that's backed
up with a lot more than most of the other belief systems
and that can reflect on itself and it
can change based on most mistakes that it makes
and those things that's basically what other belief
systems don't do like religions not really
i think argentina is a catholic
country and one of
the most famous philosophers in history was saint thomas
aquinas and reflecting on
christianity from the standpoint of
greek philosophy was exactly what he did do
so you don't want to you don't want to use that
i think the thing that's
interesting about science is
the particular methods it uses to
try and deal with the noisiness of our human brains it's
basically an error correct detecting and correcting
system if you learn how to do it
and it's basically a skill so
it's not something you know and some people
it's like in music some people have a little more talent than others
but in develop music talent
won't do it you say you have to
practice and develop skills and
the same thing is true for thinking and thinking
is uh in the large
is primarily the province of what's what science
is about so in computing
uh you can hardly see this recently
recently meaning the last 30 or 40
years so the in
the 60s and 70s i started in
like 1961 or 62
the people who are doing what
call computer science today
and what we call software engineering today
started off as real scientists
and started off uh
as real engineers yeah we weren't having undergraduate
degrees that's right there was no
no programming career or
computer science career that's right and so the people who went
into it were interested in it
and just like i did it from another perspective
and that allows them to to see
we as developers don't see
it was a special thing because don't forget there
still ibm yeah back there
and there was i i learned to program in the military
in the air force and the programming there
was just as uninspired
as it is today yeah
what was special uh was that the the
research community that i just luckily
stumbled into in the mid 60s
had started thinking about
what the computer actually meant
and for them it was the next 500 year invention
after the printing press
and many of the things that were going to be important
were how it could go qualitatively beyond
the civilization building inventions
of the past several thousand years like writing
mathematics science
press and so forth and that
o this idea and ibm
did not have this idea the air force did not have
this idea but these people did
and because they were in a good place and because the cold
war was going on there was extra money in
the department of defense uh
one of the people who had this idea got funded
in a big way and he spread
the result is most of the technologies
that we use today which are not invented by ibm
not invented in the air force so
a fortunate thing the processes
that led to those inventions
hardly exist today yeah
see those kind of research
and in current industry at all yeah
the opposite is all the opposite just short term
i want to do this very fast and
yeah the talks i've given
about this in the past the line i put
up on the screen is the goodness of the results
correlates most strongly with the goodness
of the funders so
the rarest thing is a good funder because
if you look at the bell curve at the top of the bell
curve in every generation you're going to get uh
you know super clever people
to draw and so the thing
different than was not us
we just happened to be the lucky people
there the difference was in in the funding
and the funders found us
and uh encouraged us to follow our instincts
and so we've got what we've got that's what we don't
have today yeah that's right so
let me go back just one more question about
science and then we can go to to software and and that
stuff so let's let me let me play a
devil's advocate here okay i
just for a little bit i'm not sure if i can do it right but
anyway uh you know science is great you
it as the most one of the most important achievement of humankind
and uh but science
is what brought us here i mean not science
we humans using science as a
tool is what brought us to the situation where we
waste everywhere with the climate
you know the burning woods and
those kind of things because science
in science in in some way allow us to
think better how to build stuff
but in the other way
not ready to use that powerful
tool so
scientific in thinking can help with
e consequences of things but the actual historical
fact is that
uh business
technology and the industrial revolution
led us to the problem we are what
science did was to make engineering
immense
and the industrial revolution that was starting
yeah was to
add on to this immensely powerful ways
it transformed engineering from something
that was essentially cookbooking
making cookbooks of things that worked
doing things by uh by principles
being able to actually derive uh
physical uh results
and estimations and new
kinds of materials and so forth so science
is a culprit in the sense that it opened
up the possibilities
and the power ratio
or what people can do
but basically when i look at
history what i see is
today what i see
is people uh
just trying to get ahead
in various ways and for instance in
america i don't think most heads of businesses
are really all that aware that they're part of a
country particularly
the multinational ones and certainly
uh if we go back 200
000 years ago we have to look
at something that we can find in all mammals
that
that we have also which is trade-offs
between two powerful forces
that are very different one is cooperation and
one is competition
the
more powerful one unfortunately is competition
in the end
it takes a very very special person
not to try and save themselves
even though we know cooperation
better than competition it is and not
only that uh you
know the reason uh there are social species
of which we are one of them is because cooperation
uh even if you're in a wolf pack
allows the pack to
and you're in
a baboon troop it allows the baboon so
uh cooperation
evolved as
uh all the other traits evolved
underneath that though
is competition and so
you know sociologically
people feel deprived if they are put into
solitary confinement or banished
but as soon as they get back into society
they start competing
and it radiates
out from them to their
you know there's a saying in some of the arab
countries that uh me against my
i against our family
our family against our neighbors
o these
uh tribal uh identity
things radiate outwards
and there are many science fiction stories written about
how to unite the human race will just uh
attack the world with aliens
and all of a sudden then the humanity can see itself
as a bigger tribe that's right and
get together and fight the aliens because yeah
so this this is the this is uh thinking of
completely immature beings who have
no real sense of of what's going on
the big deal in business
is at least in america
is most of the ceos i
know are not
really aware that the reason
they're making money at all is because
they exist within a cooperative structure
that was set up 250
years ago the wealth
that they are tapping into is there
because of the cooperative structure and then they're competing
underneath it so again just just to finish
this off to go back 200 000 years ago
you ask well
if we go back before there's hardly any
culture there's probably never
a time you know because uh primates have cultures
oh so before that even before
go back okay when we have
almost no culture
there's almost no language yeah
but we ask well what uh how did things
uh managed to survive back
then and the answer is genetically we
have drives to reproduce
that are very strong yeah we have
uh drive
to find food and water
so hunting and gathering
is predates any kind
of cooperation if you're
by yourself you're still going to be looking for food and water
yeah and if you think about hunting gathering
it doesn't scale well yeah
right hunting and gathering one of the scalings
of it is to steal things
one of the scalings of it is to
uh strip a
cooperative area dry yeah
so there are many many variations
of this and so one of the
complaints that you could have about school
is that schools do not
teach at least in the united states again
schools do not teach the
children anything important about their own species
these are kind of taboo subjects
what are human beings actually
and we just we aren't very
pretty as a species when you start looking at it from that standpoint
but the good part
of looking at it that way is we get to see
how a
civilization can be
kinder can be more cooperative can
be smarter all of these things
and that is something that uh you need to learn early
because certainly mo i would say most american adults have
idea about this at all from their standpoint
his battle for survival
yeah competition is is being taught in
high school and schools on sports
and all those things instead of instead of cooperation yeah
yeah and rhetoric and governments is we're
not competitive enough what
how could what does that mean on a finite planet
hat's crazy
i don't mean as a metaphor i mean this it is literally
insane yeah well it's what is happening to
yeah and of course the you know scientists
were aware of the climate problems
uh starting in the early 60s
yeah yeah they knew
about it and and the people the business people
tried to not to the idea to spread
out yeah yeah it's not if the business
people really understood it they
uh could actually see
one of the one of the biggest problems with lack of history
is besides
trying to avoid blunders that
done early
we reinvent the flat tire all the time
as you said reinvent the flat tire all the time
but the other thing is we miss opportunities
so for example whenever we've had
a big calamity
in the last 100 or more years
usually
things have not been prepared well enough for it
these glamorous are usually wars not always
but for example the result
of the wars
the aftermath of the wars is
usually prosperity
and the reason is is that it's in war when
there's the extra investing kind of investment
that conservative people don't like to
make or anything
yeah when you're really
in trouble you might give a smart
person you don't understand some money anyway
just right and so the kind
of funding research funding that's done during a war
is unfortunately almost the only
kind of research funding that is uh
that has been done it's when people this is why the
institute of health in the united states
gets more than three times the funding than all of the
national science foundation national
science foundation funds all the other sciences
but the national institute of health why because
people are afraid of dying so they're much
politically they they can
imagine their own death they're afraid of it and there
is a goal that unites them to do that funding
yeah in that case the real
problem with all of the the problems in
recreating arpa and xerox park
uh in recent times
have been that the the mostly billionaires that
have tried it uh they
want to uh direct the research
yeah and i've told any number of them
i said well you shouldn't be doing
this because if you look at why arpa and xerox
park succeeded
the people who directed the research were the people
who were going to do the research the people who chose the problems
were these uh top
researchers that's right and with all due respect
sir you've spent the last 20 years
becoming a billionaire
researcher and so the chances that you can
pick a good research problem are essentially zero
you just want to be a wannabe
you want to be part of this and you're
not satisfied with your billion you want to do this too
but you're incompetent to do it so in
government it's slightly different uh
where the people who are responsible
think they need to be in control because you
know they are responsible yeah that's right but the
made arpa and park different was that the people
who are responsible knew they couldn't be in control
yeah yeah big lighter and bob taylor they
allow you to do whatever you want
for trying to get a culture
going trying to find talent
responsible so taylor
never suggested a single research
project uh at park and
lick lighter stayed with his vision
and people asked well how are you going to do the vision
know but i'm going to fund people
know yeah and they'll take
30 or 40 this is you know
this is more like playing baseball yeah it's hard
to get and it is interesting because bob
taylor was a psychologist wasn't he
and lee glider that's right so they they
knew how people how to get people together
stuff i think there is some of that
both of them happen to be experimental psychologists
okay we're not clinical
yeah yeah and they're
different personalities lick was
just a very nice guy
and okay you're special though he was
one of the inventors of cognitive psychology
so he's special but his personality was
also special taylor
was a uh
much more aggressive
okay he adored
lick glider okay and
uh when he when he
was part of the arpa thing he put a lot of effort
exactly why what
lick lighter did was working so well
and when he became head of head of the
computing research at park he put those principles
he was not acting by instinct
he knew how to do it you were basically yeah
he said uh this is what lick lighter
did by instinct
uh we can we can do this by method
we'll just do do
it this way and park
was a concentration of both
talent and method yeah and it was
amazing wasn't it well
i mean everything that you've done there
yeah well i think the most amazing thing
was uh the bulk of the the work
that's known today was done by you know
25 or 30 people yeah yeah everyone
every part of it and so the concentration
of of abilities there was
large and the concentration of method
the taylor's application
of what he thought was a good way
to do this thing was more powerful than the
than the more ad hoc arpa
management way that he'd learn from
so yeah so
so i always i when i look to
when people ask me questions i said well you know
don't look at us look at look
at the you know the four funders at arpa were
lick lighter ivan sutherland
larry roberts and then taylor
came and did park and so if
you if you want to thank somebody thank them
and you know
uh when we've gotten metals in
the past uh i mean my
line is that well um you know metal
40 years after the fact is okay but
the if you if you want a real
reward think of the reward for being
this work yeah back
then to actually do that's the big deal
and then as much of a reward is the
fact that the funders were giving us gold medals
and a lot more gold than was in the
gold medals they were giving it us
before the fact knowing that most of it was going to turn
into lead so
when you get funding like that and you get to do the
work uh you don't need anything uh
that's it that's it yeah
place in the you know the right environment
you create it was great and we
knew it was great back then uh
and uh our
appreciation for it went up by about a factor of thousands
after it ended
yeah yeah yeah right because i was in
it i had been out
of the regular culture
because i was in it in grad school and then uh for 10
years at xerox park and so i've
been completely isolated from the outside computing
culture for 15 or 16
years and i was quite shocked
when you get out of there yeah
about everything i
can imagine yeah yeah i can imagine so let's
go to that subject now to software
if you don't mind because i know i know
that you can talk for an hour by yourself
i i need to do something you know
you should interrupt me uh by the way
much okay
okay yeah but i'm not sure when to interact
when to know but anyway so let's
let's talk about software and basically software design
uh you know for a long time software design has been
thought as drawing uh
on paper you know and to the design
was to draw basically what you
thought the organization of the system should be
while you were running the system in your head
so you were imagining the system and then
making some draws and you know that was designed
time that was the classic idea as you mentioned
flowcharts in the air force uh you
know that was maybe thought as design but
uh you know we know that that doesn't work
at least what you know the history
tell us that it doesn't work so what
is for you software design because you talk about it in your talk
you talk about cad and those kind of things so what
activities should i
should be done in software design
i think the first thing is
uh i'm sorry i'm sorry to interrupt
you to finish the question because you also
talk about meaning creation of meaning
and how do you relate that to design i think
that the key point is there so yeah
i think the first thing the simple
thing is that
you know the thing i had to get
over in a hurry when i went from
being a journeyman programmer
in the early 60s to accidentally
research community
was that the
the average program
in the early 60s was relatively
easy to do
the computers were small and
so the main problem in programming back
then was more than anything else
get completely
mired in optimization
because you had to optimize you didn't have enough memory you didn't
enough uh cycles
uh so that the tendency
was to convol as people do today
yeah yeah we're still doing that having
that old school in in our heads yeah
so the uh however arpa
was uh
basically the people who
were our mentors first
generation people who were funded by lick lighter
we're basically all systems people
they have many of them
sage air defense
system which was
24 installations of
two computers the size of a
softball a football field
wow there was one floor
of a four-story building the basement of the building
was the power supply for
these computers so we could think of two amazing
and then the third
floor was operations and the top floor
had 150 graphics terminals
that were run by these enormous
vacuum tube computers and
the government built 24 of these block
houses wow all connected
into the radar systems of the of the country
and so that's what these
people were doing in the 50s and
so the monumental scale
of
people who came out of world war ii
and i should mention here for
people who are interested in history if you want to understand
where the mental framework
came from for the
this uh research work in the 60s and 70s it
came out of the mostly out of the radar work
which was jointly between
uk and the uk yeah i'm
primarily at mit but started
off in in britain and there's a great story there's some great
books read about it and
a talk that you had with a
on youtube that where you talk
about that i've got a couple of talks about that i've written
some papers yeah about
it also another monumental
effort was the manhattan
project
so most people aren't who don't the manhattan project they spent
you think about what the us was doing in
world war ii the manhattan project itself
cost more than one percent
of all of the defense funding
war ii so
it involved about 800
000 people it involved
making new cities yeah
because they went wherever they could find cooling water and they
built new cities they brought in school teachers they bought
in doctors they built entire cities they built
plants of acres and acres they didn't know
the best way to refine
get refined uh material
for the for the bombs there are four
known ways and general grove said well let's let's
just go all out on all four of them
and he was the he was the head of this project
and so his history is worthwhile
reading to read about uh
you know something at scale that scale
yeah yeah and the same thing is and i've used
this many times in talks is my background partly
was in uh molecular
biology and you have my favorite book up there
yeah yeah behind you yeah
that's my favorite edition of the book the third is oh
cool the red one the right one not
one that's i think that was the sweet spot
and uh they haven't read it yet
little bit big one of the best
reads ever yeah it's about a thousand pages long
one week one week of reading
you're in uh
so if you come out of that background computers are
tiny tiny tiny little things
and systems
are even the
largest human-built systems are small compared
to what biology pulls
off and of course a lot of the stuff in biology
uh can't be applied to computing
just because of the nature
of the way the materials themselves work and
what things are like at sub microscopic
scales but a lot of it does
and another book that was very influential
was uh christopher alexander's
book which is his phd
thesis called notes on a synthesis of form
yes yeah he repudiated this book
because he got hippified uh
after he went to berkeley went into a different
subsequently wasn't that bad
but this phd thesis book was
really interesting and his
his main example
done after this fascinating
discourse of how you think about complexity
and design and the conflicts
and finding them and
modulizing things and it's just great it's
great today it's just a great book i have it here
somewhere yeah but the end
the the end example of it is
taking the uh
new uh a new
uh village in india from
scratch so it's going to have a
population of a few thousand people and it has
you know a thousand or two constraints
of every different
kind in the thing and how are you going to design this
and this is great because this is not a typical
if you're looking for computing stuff
uh people are start taught programming
examples that are nothing like
or should be nothing like anything they're going to do
for real yeah yeah yeah that's right
much better off dealing with us
a live running system
like say small talk yeah or even
javascript although the it's so ugly in there
but it's live so
system that you can look at
you're much better off learning how to program in
the context of the system because
then you start having to sift into
your mind yeah it's
because you have to to realize
that if you change something you can break it while running
so it you know you can do you can learn a
little bit of driving a car driving around
a field or a parking lot yeah with no cars
on it but really driving is learning how to drive on a
street where you have stop signs and you have that's right
there's all these things you have to worry about and it's very confusing
and the process of learning how to drive
is learning how to create things
in your mind for dealing with all
of the heuristics that have to be done in real time but
so design is again one of these thinking skills
and part of it part
skill is how to hold
in your mind simultaneously
uh things
that conflict or seem to
rather than trying to resolve them too early
and i forget whether i did it in this talk or not but
about you know our ideas made
or like categories where they can't interpenetrate
or
they
hello
i think i lost i think
we lost yeah oh we lost you
the last yeah 30 seconds chris
from my point yeah so the question is are ideas made out
of matter yeah
so one idea can't is basically
antagonistic to another or rd is made out of radiation
so you can shine lots lots of colored
lights at a wall and they superpose
so you can see all of them there they don't interfere
with each other you see interesting combinations
and that's one of the states you have to get into
when you design or when you think about anything
when you're dealing with complexity
the last thing you should try to do is solve a problem
the number one thing you have to do when you're dealing
with complexity is to try and find out what's going on
yeah try to find a way of looking at what's going
yeah the problem the problem finding instead of
the problem solving grinding is and that was the big
one of the big things i got in arpa yeah
they realized the problem finding is that and so
when you get to software
and think about what's the software possible the
pro software process is
trying to deal with
e kinds of things that alexander was dealing with
and the first thing you have to have is some
understanding of what the meaning is
before you start trying to to optimize so
alex so if you look at alexander the i should
have brought brought the book
i have it in my room i don't have it here but yeah
i have it here
oh my
god
this is london so i don't have my real
library here but
i have about a thousand
whoops i have about a thousand books here
just just a few and i
need some yeah so this book
yeah you can't
uh praise it too highly
eah
i read it it's really interesting it opens your
mind yeah so here's his final
design of the village
uh but of the most interesting
things is how careful he is
yeah so appendix
one is the worked example where he
looks at hundreds of constraints
and tries to put them down very carefully
if you look at the requirements and of course
the
real question is are you actually saying anything when
requirement down in a lot in a natural language
so
the belief we had when we started thinking
about this stuff and you bet again
the arpa community
invented computer-aided design
and general motors
was doing also doing a project but
i think was most strongly invented
in sketchpad by ivan sutherland
and then especially by his brother bert
and others at lincoln lab and
mit had a huge numerically
tool project
that was also about computer-aided design so
the idea is you and a lot of the computer graphics that was
then was to try and see whether you could define
shapes using
graphics that a program
could look at the shape and see what a
five-axis milling machine
would have to do to make that shape out of a
piece of metal and so there's a lot of
that and of course the
shape that you want like when you make a flange
yes uh
that flange might not be strong enough
in its yeah
in its simple flangeness
yeah you might have to put a fillet
you know which is extra metal yeah to make it stronger
on the bend part so you can think of that
as something that would be revealed
in simulation so
sketchpad when you designed something in sketchpad
you didn't just you weren't just doing a drawing
you would get something that because you had all the
constraints and the simulation to see what yeah
so and and going back sorry to interrupt
you because we're running out of time and i have a bunch of
questions and i want i want to go to one at least one
questions from the people i don't want to be the only one asking
but um you know you mentioned small talk and
you know that amazing uh system
dan ingalls and adele goldberg
other people and you know working is
with small talk as you said it's like working in a living
system where you are changing it while the
system runs but also one of the main
features of small talk is the immediate feedback and
you you know brett victor he he gave a great talk about
immediate feedback and how important that is to design
stuff to work with staff so how do you see
immediate feedback in in software development how important
you think i think you know
put into my talk
session i just did another talk
okay okay no but you didn't talk about immediate feedback
in your talk yeah but i might have thrown in
uh four stages
yes yes uh the tinkering
engineering followers that's right and i was going
you forgot art why you didn't
put art in there also oh it all
is oh okay okay because what art
art the the greek
word uh art comes from the
latin ours and
the the greek root which goes back even earlier
is techni okay so technology
literally means anything that
humans make anything
okay so what what
happened is in the 19th century uh
painting and sculpture took the term
art oh it used to be called the fine
arts okay okay let's see i
see yeah that's like uh it's like
uh a.i used to mean something different than stuff today
that's right yeah yeah object oriented
used to mean something new today
so people people do this but in fact
in many of the presentations i do i show
these four guys sitting
the larger context around them
is art because it's everything everything
that's that's why artificial
means something special
yeah doesn't mean
something bad perfectly means
something that's made
okay is
intelligence that is made that is made yeah
so yeah so
uh so if you look at those four things
uh and the point i make is
in modern times you want to be
in the sweet spot at the center
ight yeah that combines diagrams
yeah so you want to choose when am i going to tinker
scientist tinker
engineers tinker mathematicians tinker so you have to
tinker that's right yeah you have to be you
have to make things principled it's not
just an engineering but uh like i have also
have a degree in pure math so if you're
going to make a mathematical proof it's at
least as engineered as a bridges
it has to have that integrity
connectiveness in order to be
considered a good proof yeah so
yeah and then science has this extra
important way of being a
between what's inside of our
heads and the kinds of phenomena that we
deal with out there so it's it's a meta
thing that's much more complex than the rest of these
ideas and much more powerful
so so if you try and apply that
map software into it it's usually
way off uh yeah yeah
bit and wearing
uh hardly any math no real
science so if you try
and map that into some modern system
that you might try to make
one of the number one things i think anybody would do
and certainly we did
what we could do back in the 70s
was to say yeah we have to do cad and sin
you have to be able to do computer-aided design
which means we have to be able to continue it's
graphics that we have to continually develop
the thing we have
in mind so we can at least fasten on the meaning
of it so the compute
the cad part is meaning it's semantics
and the
simulation part is because
beyond a few sentences
you shouldn't trust 20
or 30 or 100 requirements yeah
that's right without debugging that's right
that's that's something we do actually in software development
with testing you know there is a
test-driven development i don't know if you heard about it uh
yeah sure but that the problem with that
is that you know
if it were good let's let's
you're doing it well yeah and
in theory you should be able to run
uh you should be able to on a supercomputer you should be able
bring uh the thing to life right if
is covering everything uh the truth
at's right some parts of
it yes so the biggest problem with
want to have tests but the problem
is it loses the larger integrity
of what we need in designing a system
that's right yeah so i think
it's it's not that you don't want to have tests but i think
trying to start with tests is
and again
sometimes i throw in a lump of clay yeah
yeah you did can you i did okay so
it's pretty hard to debug a lump of clay into
a really nice piece of sculpture
yeah yeah and but you got to test
everything continuously right yeah you
have to have a vision of what it is that
you do yes but starting starting with the test
it's like thinking about the meanings first instead
of thinking about the no no it's it's thinking about
criteria the question is is
you know the way to think about the meanings
is to think about the simplest thing that is
like the thing that you want
that gets you no so starting off uh
uh asking
about bathroom facilities
is not going to lead you to a village design
no of course so what you need to think about
is what is the simplest thing that's like a village okay
okay let's see that gives you the system
okay that gives you many of the main things that the system
has to have and then that's not going to be
uh nearly
detailed enough or scaled
enough but it has its it's
a vision of the whole yeah
i think we don't have too much time
i'm sorry to interrupt you again i think we need like six hours
three ten days to talk about everything
uys are organizing you you
know too much alan you know too much it's
difficult it's difficult too
but let me at least ask you one question
from the audience um we have one from maximo
prieto he he admires you because
and he his question
is what can we as simple programmers
not the owners of the business do to
provoke a change in the way we are forced to program
today i think we deserve to program in a better
way yeah so well i think that's the big problem when i
was a professional programmer
you know it was somebody else's machine somebody
else's software somebody else's
problem so i was
you know at the bottom of this
of this machinery on on the other hand
what we did have
lot of this was in machine code
but the nice thing is uh the
machine code systems back then
many of them had really good macro systems
and so anybody
who wanted to survive back then and particularly
with regard to this question
the first thing we would do
a bunch of us who are involved in these projects
is to spend a fair amount of our time
each week often each day
working with each other on macros
that would give us higher level
blocks so basically programming language
dsl basically making
because otherwise you're at
the you're in the wrong place
for uh changing your mind so
about one of the ways of thinking about this thing is
uh and certainly was the way
people approached it and the arpa community really approached
it was let's
admit that human beings are terrible thinkers
that means we are terrible thinkers
what can we do about that well it means that
most of our ideas are going to be mediocre down to bad
let's just put that right
hat's right let's
as a you know that's the yeah let's
face it we still want to make progress so what
can we do well what we need to do is to be able to
fix things so we can change our
mind and what are those things we can do well
starting to happen back when i was a
uh when i was a programmer like one
of the machines i programmed on back then did
not have any index registers
really basic yeah
but you don't need an index register you
can go in and modify addresses that's right
what index registers are is late
binding something
yeah give you both safety and
a way of changing your mind more easily
every time you use a an indirect pointer
you are allowing
allowing
the possibility of sticking something in between
that's right to change that that the
other way would be static but in that way it's dynamic
it later on yeah idea
is virtually nothing that's important
should you know the actual address of
and one other thing and i've
this you know long ago a big
although it wasn't nearly as big as
it as it should have been became a really big
revelation later on
but the the file system
uh on the borough's beach
uh rose 220
in the air force so there were no operating systems
then and they wanted to exchange
tapes with files on them
and the way they hit on doing this
and i don't know who did it but
the idea was that the and these
piles were long because i won't go into the
weirdness of the tape drives back then but
so the front part of this was
bunch of a vec you know
an array yeah of pointers
further on down in the file
into the second part of the file which is a bunch of borrows
220 programs
machine codes yeah those are machine code programs
yeah and then the last part which could be
the size of the rest of the tape were all
the data records and the way
you way you uh read a tape is the tape
always got read into the same place in memory
and the only thing that was standardized was
what the index positions in this
first array actually meant
like so it was basically a
class with messages pointing to
methods basically that yeah yeah and i
real you know and it was great
we we called that data driven programming back then
yeah okay because what what
want to do is write your own code
to understand what you needed to do
was just understand you know if you want to find record
150 and of course there are different
sizes but it didn't matter because the code
there knew what size these guys were it knew
whether where the data is right
so nothing nothing was fixed but it was
dynamically decided by the
code yeah i got there in 1961 and
that was that was already in place
yeah wow yeah so that
was how you did things on the on the
borough's 220 for air
training command records
and but you know if you look at the principle
behind it yes it's like the principle of oop
that's right basically your abstracting and sketchpad
which was done about the same time had
the same idea sketchpad object
was
mostly a bunch of indirect points of the
thing called display and sketchpad
was just a pointer at a
specific position in this data
record for the thing and you just
jumped in direct through it and it would take you to the procedure
that was most well suited for displaying
that particular kind of thing there so
even though in the sketchpad those things
were more like prototypes than that classes is that
i don't know if you look if you read the thesis
yeah everybody should because no he
had uh even something like
inheritance oh really well
no no it was
sketch paper they were not prototypes like
that you cloned because i thought that they were like
cloned to have another one
not like no no okay okay
okay so but anyway once
you know i saw a sketch pad
a few years later in 66 in my first
in grad school along with simula and
it was having seen
this this idea like five or six times
a row because the borough's b5000
had a variant of it also and
in some ways even more sophisticated
and i was just dumb you know it just took me
four five six times to see the thing then finally
error is holy this is really an enormous
good idea it is actually
cosmically important so
let's make it scale now that's what you
did basically yeah so once
course once i saw it was
out to
yes
you know and again again it's
it's pragmatics that hurts
the idea like it's the simplest idea
that there is that if you have a computer
you can do any computation
you can represent any data structure
yeah so it should be a complete
instant
deduction that oh if i want to make
a computing system i should just make virtual computers
because that will allow me
to define anything else and i have
a universal way of doing it and people still don't use
what they call oop today to do that yeah
they just can't see it
here's something about a computer even today when they're
a lot smaller that with
where the pragmatic reality of the
computer you know contrasted
virtually or what it is semantically
yeah yeah so what's once you can
move from pregnant so one of the ways of looking
at is one of the biggest problems in programming in every
era and i don't even know what
decade this is for me i've encountered them
but it's going to be 60 60 years pretty soon
six years next year
six decades wow but in every year
it's the same thing with the same problem
get caught up with the pragmatics
yeah with the immediate with the immediate
instead of the yeah and the number one
optimization and pragmatics
hat on
and we can all do it once you put that
on it is almost impossible to design after
that's right completely incompatible
idea it's very difficult
to change it it's very difficult to say
to understand that you did something wrong it's very difficult to
see that you see that you did something wrong
so yeah so for me uh
you know there are a lot of different ways of using a dynamic system
and it depends on the personality for me
i like to do
when i was i
like to do this idea what's the smallest
village and i
just do that from scratch okay even
in small talk in small talk i wouldn't go in and define a lot
classes or any of that stuff i'd use the
the thing and called workspace which
had grass space yeah in it
and i would use a rebel with
i say like a rebel but with asteroids
because you know in a rebel you're like working
in one line but in a workspace you have the whole thing yeah yeah
you can have many many things in there and
it has its own uh address space
so what i would do is just do
a complete system the smallest thing
that's a village and everything i mean i would do the
loop i do the user interface i do every little thing the
could do trying to get
into my head what the
uh primary focus of the design
because the important thing about what
what oop is at least the oops stuff that we did
is that it's a module
system for us encapsulation
percent yeah yeah
yeah it's not there uh you know
occasionally you might have to fall back and define something like a data
structure but and when you
do that you're actually recording danger
basically it's a module system and the problem with
system is it doesn't tell you what modules
hould come up with right the module
system just gives you a way of protecting one set of things from
another
the whole front part of it is how to think
in terms of modules and what
to do in his his thesis program which
he wrote in fortran was a thing that attempted
to find you the
best set of modules out of a complex
set of constraints still a good idea yeah yeah
but basically and to me that's getting
the meaning of the thing down so to
me meetings dominate and then you have
this interesting thing in
most programming languages today
is there's nothing that because of this
fact that you can use in directions
that you can have a set of meanings
and they might include some tests
but if you think of the uh
yeah suppose it's just something like sorting yeah
and so
the you know the the meaning
for sorting is that the
output is a permutation of the input yeah
that obeys some some relationship
and prolog will even uh
sort for you just on that basis
does it really does it the hard way but
it it gives you the and of course
permutation is not easy thing to describe
it's a trickier kind of
thing but uh but there are 50
known ways to sort oh sort yeah
and they all have different uh
pragmatic ranges
and they have different conditions and so you can imagine
if you're going to do a sorting thing you have something that's
going to protect the meaning
so you can grind let it grind if you want
yeah yeah or it could be a really simple sorting
program that you're sure really does sort
yeah maybe or maybe we combined with tests
because both of these things but you keep that over on
the side and then on the on the right hand side of the page
you write down your 50 sorting routines
and you head them with the conditions
under
which they should be invoked so
they look this is just what we did when we were writing macros
force that's right basically
inputs oh this is an enormous
array so uh
i better not do a bubble sort here
or one of the side conditions
here is uh
this uh this
system has to be uh
easily updatable
so you might want to use a b tree
right you know you don't want to use a hard
array you might be better off going to a b tree
which gives you incremental updating
on the thing and still gives you pretty fast sorting
yeah yeah of course small talk has any number of
these yes different uh kinds of things
and then your module
is which is called sort
is the thing that
done the cad sim part
fab part
is all the all these
routines and the rules should be you should be able
to turn off any and all of the optimizations in any part
of the system and just have the system slow down
that's right but still working if and
that isn't true then you've done
a bad design period
that's simple yeah you're just playing at being an
engineer yeah in computing or you're just
flying at being a programmer even
yeah but going to to the last question
so what what would you tell to a programmer
today to you know
are working what can we
do you know a tiny people can we
do something or not i mean
because as you've mentioned many times complexity
currently in in our you know development software right now
is so complex uh you know working
doing web application is so much
difficult today that doing applications like
20 years ago yeah but so okay it's in
some cases you're just not
going to beat the system yeah
that's that's some cases just that because
the the what's imposed
on you is just too much however
like suppose you are doing web stuff yeah
uh so the
the first thing to notice is that
underneath javascript
is a dynamic
language subset of javascript
can be used as a real
target from uh
other language development yeah yeah
because uh you know there's a
good garbage collector yeah
yeah and quite a bit of it
lacks a couple of reflection
but i i advise people
in javascript to do a thing that
ago which was to
make a preprocessor for javascript
yeah there are
many there are many things like that today yeah it just parses
so you feed javascript to it
and it writes
things in a way that uh what you do
is reflective because it adds
that stuff in there that's right okay okay
so the basic idea is that for almost any
programming language uh
the chances that it's going to really fit what you need is
low and the whole point we do
should be open and with all the source
code available like small talk that helps
that helps a lot yeah i'm sorry sorry sorry
to interrupt you yeah but in
but you know small talk i don't advocate small talk today
because it was the world's greatest thing
really in the 70s but
uh you know what's
mall talk today is how well it compares
with more modern things
but that's because the more modern things are not very good
small talk is not
what i would use if i was going to do
a major project today i would
make i would do what we did you know the other thing about
that community back then is uh it didn't bother
us to do major tools yeah that's right sometimes
our part of our skill set so yes something that
today we're still difficult it's like
we build tools for all the other that's right
for other people but not for us uh
it's that but it's also just people not learning their
field yeah most people
are are happy to get paid for doing x
yeah and uh that's not very professional
so if you're professional you learn
field and our our field
is primarily
there because software
was invented so we didn't
have to put up with a fixed set of
facilities from hardware yeah let's think
about what that means it means that if you have software
you should never have to put up a fixed
set of anything well if you're
just treating uh
you're treating really badly design software
most of the time as though it's some sort of machine
that you can't do anything with
so this indirection principle yeah
so i i suggest just being subversive
you basically
you start working on
a much better way
of doing little parts of things
and at some point you can
uh see if you can attract more
people to doing this you gotta
you know you you do it over lunch with
beer you gradually
but but before you do that
it helps you know the first ones you do
you're not going to want to use as tools
yeah but you have to get started
and if you are going to try and
get around the problem that we have today
then you're going to have to be a lot more skilled
than most programmers are
yeah so i
like i said i was uh
kind of a standard program because i was used
programming to work my way through college so i didn't think of it
as a central thing at all
i was i had a got a math degree in a biology
and i programmed uh
to pay my tuition yeah but when i when
i got into grad school i was in this research
community that was really serious
about big things
and so but they also gave
the grad students a lot of leeway
so i had the time
and the freedom and the
resources there to just
learn a whole bunch of things that if i'd
been more of a professional earlier on i would
idn't know how to do a language i didn't know how to do an operating
system i didn't know how to do a
system even okay doing
programs and uh you know and
but a couple of years later
uh i was more
skilled on this and i understood
process of
because that graduate school
arpa
did not require
people to do individual
theses although usually
ou did but usually
they allow you they gave you the freedom to
do whatever you thought it was it was in a
big context and so the uh
you could work on big problems and
you know write up the first two
a big problem if that was good
they'd give you a phd if you worked on smaller problems
they wanted you to finish something in two years
but they didn't care what
they wanted was two years of world-class
work to get to give out a phd and
you didn't have to write little papers
with your professor's name on them or any of that stuff
okay they didn't want you to write papers
well completely different to what
oday what they want you to do is to write a thesis
right that's the whole reason
grad school it's not to write papers that's right
they want you to do
uh world-class stuff for two
row and write it up and out you go
so i was in grad school for two and a half
years and uh
and i wasn't i was not
nearly the fastest fastest one out
during my year uh in that community
was john warnock who
is famous for doing adobe oh
okay but john was a uh
was a staff program he had a math
degree had a master's in math and he
working had a wife and a kid and he was working as a
staff programmer for the university
and one of the kids
on the grads uh grad school
students went to him and asked them a question and
john saw a big answer to it that happened
to be the first really good way to do continuous
tone 3d graphics
he was the inventor of at that time
of that and so his
he was a grad student less than six months
and his thesis
was actually 16 pages of text
and nine pages of pictures
bingo out he went
and falcon is a multi-millionaire because of adobe
that
uh so that was a
good way of doing things back then because it it
emphasized not what you're doing
school school was basically
something that was a support system for
doing the kinds of things you were going to do after school
that was the way they thought of it and um
so the so the main thing about it is
uh there was a lot of scrambling but there was
not a lot of competition yeah that's right because
these are grants for were for entire
departments and so the professors weren't competing
the professors didn't have to worry about tenure
graduate students weren't competing
was a completely different environment
today but i like i like what you said that uh
should be subversive to
that's what that's what software
is that's right that's right that's science
the big problem with it is bad software is subversive
in really terrible ways
like the like the pollution
that i did and yeah yeah so
if you think about it uh you know in medicine
you can you're allowed to put on a band-aid
without a doctor's uh
degree but not allowed
to do a hard operation without
being certified and nothing like that exists in software
and so forth yeah yeah yeah and and if when you get
out of of college in in medicine you have to
stay like four years in
stuff you have to
learn yeah yeah and so
uh so to go to a serious
subject it's if you know if you know about
the boeing 737 max
yeah yeah yeah so my one of
is a pilot and he used to to pilot
that one so yeah so there you have
an artificial intelligence that knows nothing
doing and
uh it was allowed to be made
right we know there are people on board
it doesn't know anything about flying
because if you're at an altitude of a couple hundred feet
you don't correct for a stall by diving
the plane at the ground that's right every pilot
so so uh
boeing allowed people to
make a technological device
that uh should be
called in artificial intelligence
because it does things that humans do
and uh but it has no
certification the people would
have no certification and so
none of none of that is being
seriously in medicine the number
the first thing in the hippocratic oath
for doctors says above
all do no harm that's right that's right yeah
ave that those kind of things no such thing
no no and the equivalent in engineering
is the bridge must not fall
the plane must not crash
and engineering is now starting to violate that
yeah they let that plane
there's no way you should ever let a plane
right and so
what's happening is this pop culture
amateur and
i should i i like the term amateur because
lover okay
but basically people who are not skilled enough
for the responsibilities they have
are and may not
realize it may not have the faintest
idea they may think that's right safety bug one
bubble sort program they know how to program
and they're being hired right
and writing gazillions
of terrible code much of which is going
to be is almost impossible
to deal with you know so i i say
well this is like a big fire
so in a big fire you have to decide what parts you're
going to just learn let burn out
and other parts you have to isolate by driving
okay through it and so forth and
think about what software is it's exactly the same
idea right you have to
uh the
much of the stuff look
just to retreat back to something that should never have
happen is the way the web and the web
was done i was i was going to ask
question because there is one number one
the number one telling thing to me is
virtually no student i've met in
undergraduate or graduate school at ucla
can tell me what's wrong with the web
in the web browser
that means uh the education
is completely failed yeah because
they only teach you what we use right now and
know what used to be or the principles or the problems
full circle to where we started which is
people are projecting their beliefs on the world
like one of the most pernicious beliefs in computing
is uh simple
darwinism yeah we must
have the best stuff in the
and i used to give lots of talks until i gave up
showing stuff from the past that's infinitely better than the
stuff today no because they think darwinism
just fits
yeah that's right if you have a stupid environment
you're going to wind up with with a
stupid result of evolution and that's what
we've got remember the
really important thing is that
people were probably as clever from
the iq standpoint 100 000 years ago
as they are today yeah right
but uh the
so uh ignorance resembles
you can be clever as hell but if you don't know anything
you might not be clever enough to
around that and liam my grandfather
my grandfather used to say that it is worse an ignorant
person that a dumb person because
is ignorant doesn't know it
and that makes more more harm than
somebody that yeah yeah if you have a smart
hey know it and they don't yeah so
and that's basically i think computing attracts
that kind of personality there
used to be and also it attracts a personality
that tends not to like humans very much
they're more common
well i think it's because uh
you know many people in computing uh
have asperger's i have a bit myself
and it's comforting to deal with machinery
yeah yeah because you don't have to negotiate
with it yeah that's right that's right and
it doesn't have all of these you don't have to be polite
yeah yeah uh you don't have to try and
figure out what the the other person
is thinking about you don't have to do all of these things yeah
yeah and so the thing that saved me
was doing theater whereas
theater is the perfect thing for
a person who is uncomfortable around
people because you have
to because it's theater is basically the
an anthropological way of getting at the anthropology
of humanity
interesting yeah because
understand why theater can work
if you think about what it is and think about
why people can wind up crying
in a completely artificial situation yeah
okay and how to arrange things
happens you understand a lot
if you're still uncomfortable with them
but you understand more about what's going on
and you also have the understanding
needed to design a decent user interface
if you don't have that
uh the user interfaces you'll make are hopeless
yeah and you'll retreat into
world of simple machinery
simple yeah
so it's very possible to for people to
create these bubbles that they
live in that really have nothing to do certainly
if you look at uh the software i've looked
at it's too much software to make a blanket statement
world now but
the software that i've looked at in
mainly in very large companies and in
the government uh indicates
that hardly anybody
who is good at design was ever in the
process because the
objective is to get things done quickly
on time and in budget not
worry about the future and that's where we get all the trash
in the pacific ocean from that's right it's all about
getting the bottle of water to you quickly
cheaply and who cares where the bottle
afterwards that's right that's right that's basically
why i use that analogy to start off
my talk that it's it works very well
uh unfortunately software is much more invisible
than the trash in the pacific ocean
because you can't take pictures
of
you know gazillions of terabytes
of crap
you can get an idea by looking at
a company that should be doing a lot better
uh like google
and ask the question
how is it possible that when you retrieve something
oogle that it's not showing
summaries of those web
just picking random
stuff out of the web pages but it has to index the
web pages to be able to do the retrieval at all
so why aren't they indexing the retr uh the web
pages are about
and putting that in there is meta information
why after all of these year now of course i've asked
my friends at google many times
you guys are doing all of this stuff you
got ai chips to do this and
but you can't do the most elementary thing
in your primary product
to make it any better than it was uh 25
years ago right so
what's the
problem
can you hear me
hello
hello can't hear you
see i was complaining about
the web and it took
revenge
hello hernan
yes can you hear me now yes
yes no i don't know what happened something was wrong
with my well i was i was complaining about the web
and it took revenge yeah
so what i i was asking you what did i say
when you told them that when you asked them that well
you know if i i haven't tried it on
recently but i used to every couple of months i'd send
norvig an email he's a good guy
charge of all this stuff and well
he went to the basically the
the one of the ways of
looking at it and you could ask them why
the things that really did happen
and i think it has to do with
how school once you start valuing
a's in school
pro process
that makes the problems progressively easier
right because people complain
if they don't get an a if they're valuable that's right
the only goal for them is that yeah
then you have to go to easier and easier problems rather than
uh and so i think what
happened is certainly with uh
ai which is actually a hard problem
what they did was to substitute uh
various forms of machine learning
and uh you know perceptron
type stuff which is which
you know one part of our
brain for a fair amount of our brain does a lot
of that but so do so do pigeons
yeah they decided to uh
and the nice for them the nice thing is
cohonen years ago showed that
certain forms of matrix algebra
were isomorphic to simple perceptrons
and all of a sudden that meant so
that wasn't always known okay
was a guy tubo
i don't know yeah well he was the guy who did that
and uh it put a mathematical
basis on it
and it suddenly allowed academics to write
papers with math in them
and since they were supposed to write papers
uh all of a sudden
academic ai shifted
abruptly yeah from just
just machine learning and giving up what
is now called general ai yeah yeah
this is like i have to call object-oriented
programming dynamic object going to program
now
because the term i
away from me and the same thing
as the term that mccarthy made up
which was you know in which machine learning
was a tiny tiny part that's right that's right
take it away from him and
change and change it change the meaning of the world yeah
and so uh a friend of a friend of mine
actually wrote a paper somewhere about this
thing and he called it colonization
when something is
successful everybody wants in on it
and so the they get it by colonizing
the term
in american
uh k-12
uh mathematics yeah
uh they tried to reform it
yeah they got so much pushback that
solved the problem by renaming arithmetic mathematics
that was the solution mathematics in school
have
to realize this is the world that we live in the pop culture that
in yeah that's right it's all about labels
it's about designer jeans
buying a t-shirt with a label
got one right on on you right now there you go
that gives
you the illusion that you understand what i mean
yeah of course gives
other people the illusion
thank you thank you for you know make
me feel like that think
about it do you know what i really meant well i
think so but i'm not sure you know
i'm on email so
i i i know the history i i
'm not sure if i understand exactly
what you meant i've gotten very few emails
about this because people love these
slogans yeah this is why people love
organized religion because
the slogan sounds good yeah
yeah but you have to realize the slogan is this the
can tell that's right that's right
have the context so it's the meaning
yeah and
so well
no the meaning can be what they want it to be
oh yeah that's that's better yeah that's the way there is
a university building in padelbourne germany
yeah that has your phrase in there yeah
building and when i went they asked me to come
celebrate it and i asked the audience whether they
knew what i meant
so i showed them some of the variations
okay okay and it's like like point of
view is worth 80 iq points which is another favorite
one it's on t-shirts and yeah but
i i didn't say whether the sign was plus or minus
good point
and having a weak point
view uh makes you
appear less smart than you actually are
and the same thing with the best way to predict the future is to
invent it doesn't say anything about uh
whether it's a good future
see people inventing the future take a look at washington
and wall street
yeah
so yeah this is what i'm saying these
slogans i started making up see i think in terms
tell by the fact that i talk
and talk and the reason is that i'm basically
a book guy i don't think
in terms of sentences and i don't think in terms of slogans
but i started making up these slogans when
at xerox when when
i realized the executives couldn't deal with threesome with this
slogan and what they needed was
something that sounded good to them
a good way to
make them think what you wanted it's like
when i give a talk uh
the talk is just a commercial
for doing a lot of work
and most people uh try to use the talk as
ource of knowledge and it can't be
look at the world that we live in this world was not made
oral discourse this world was made by
thousands and thousands of these things
yeah these books that's right not bad from what
we're doing right now so if if this
session you know had
help us to understand
a little bit more of what is going on and i think the number
one thing to take from any talk of mine is wow whatever
thinking might not be right that's right
yeah
that's right you have to doubt about
o get you know
so i think make
the mistake of thinking that i know the answers yeah
don't but i am a professional thinker
so what i do know how to do is to
my own thinking yeah so most people
don't do that alan i you know
we have to uh you know the time
is up uh i
you know i would love to be like talking
for more like i don't know 20 hours 40
hours all the time you could give us but uh
this time we have to to end and
i think the last thing the last thing that you said is
how we can you know wrap up uh
thing um i really appreciate
your talk and it was amazing for me a pleasure
to be able to talk to you personally i i wanted
to go to london this year as i told you in the email
to the conference of the history programming languages sadly
it wasn't done because of
the covet but anyway um i think everybody
enjoyed your talk and everything
us today and you told us today
and i hope to see you someday and
for everybody to keep enjoying
what you think london is a great place to live
oh well i haven't i've been out
of my flat here since the last week
february whoa
because yeah yeah
they have one they have one of the worst death rates
uh in the world here in the uk so
and i'm a former biologist so
better be careful we want you to leave more
i am my wife and i
are casual we're really here okay
okay alan thank you very much i i
don't know if you want to finish saying something a small
than enough okay
it's been it's been a tremendous pleasure thank
you thank you very much bye bye
okay thank you alan
[Music]