Alan Kay Speaks at BAAI Conference

From Viewpoints Intelligent Archive
Revision as of 00:34, 20 November 2021 by Ohshima (talk | contribs)
Jump to: navigation, search
English (auto-generated) Click for settings
Alan Kay are so diet off it
hangs we see that Polly Donahoe Dodgers one night yes
Shanghai agenda and then
Hamlet leadership team
chumhum tempo
she paid ntlm continental
Alicia coach automatically
Cuccinelli entering Geeta Benton casita
yes I too met he entering
a trade hotel kuzia seongho Iranian
hopt Sutra new tell-all reunion
piano melody
Fortran to TIA Antigua Yamaha jatropha
Medici to come see our denim
poppy show you my kuchi Koo Yong goo hyung sutiya
- belches - session 1 yo
G Rho G 20 G to
go Don Watson teaming with a giant axe user instituto
de castilla Iranian
culture rotation Judy tranches entonces
si si yo chill intention in - Oct
7 it'll be a little silly Gigi
Hadid Atocha reunion show a
contingent ratio Quixote Emilia Mattapan
en todo el carro con tiempo de
Santa will yah yah I am fatigued velocity
tilde yokujin convention will
run deep here sweetie - casita will
run yeah - yeah Julian country period future
or the eponym she which she had
I get paid into your own countrymen gentle and Ishita
Allan K chose
L ve chose Cydonia fortune su
su tu tu - ooh because each are more sensual
Anita consolation Dante akaku
tsuyoku don don t hesitate written
he impose even curse official person
so and Humberto and de
de shiran simple yet important in
way emergency to eat oye bioship
a shoo-in cicada Louis Vuitton simplicity
Tony onomatopoeia ha ha
Sutra for incontinence in chickens oh ji-ho
tion-tion-tion solution Yin Zi
Shi Tian a fatality 22 cents a meal
Samir sample some pinto beans a hotel
or intentional food injure and clay to
vilify the chief salam 11g cannon
salsa we like in case ammonia built
in the orientation module 33 Shivani
- ha she will make you require shoes and
we like to see routine consul city like hoochie
ha ha ha
see here do you know
welcome to the BAA I a
conference and we
are going to have this you know almost an
hour you to talk about few things
you right away here is a
very enthusiastic about
particular you know you you have this particular
note of how the big hall
so I assume you
have a presentation first and now let's go the
presentation then we will come back with our dialogue
ok I'm gonna need to share my screen
sure here
networking yes perfect okay
okay so this
is just a few notes
to help lead
us into the discussion which is the main point of this
and these five topics and questions
were ones that were part
of the context I was given for this
little talk in discussion and in order
we all felt that this white
paper I wrote last year for a MacArthur
Foundation here in
the UK which
is all about large
and they were interested in how are
some really large challenges dealt with in the past and
there's something that can we learn for it can we help
educate the politician so I'm
presuming most people haven't read this
so I picked a few things from this
here on
the screen is one of the URLs
to get this but
possibly the UL URL has already been
sent around so
anyway this is a white
paper looking at some
of the big back to them
efforts on intense
critical challenges in the past
when I think about these things I
always try and think about what's the larger
culture that any problem
is embedded in because larger
surround often gets lost and
for the stuff I think about I always
start with humans and all
of the cultures on the planet so we
have several thousand cultures and
within these cultures and we are in
ours besides
all the things that we're doing we
have duties to society those are things we need to think
about when we're allocating our time we
all have duties to bringing up the next generation
of children whether we have children or not
here are world
views we have to deal with most of the
disagreements between people of goodwill
comes from them actually
having different world views and
we have sort of an think
have the same world view as we do but we don't so
you have to take that into consideration we
have to understand what schooling is and what it should be
there's this idea of richness
where it's
not about work it's not
about duties richness is
expanding our ability to have an
emotional connections to things and
then there's
this problem of how do we get
from day to day with our basic
needs so these
six seven things
that I treat as a context and especially
for here but also the way
I think about things I think of these is being embedded
in large issues
but I just picked
12 of them
so you can pick your favorite one the
problem is all twelve of them have
been let to get into disastrous
so we really have to work on all 12 of
them and that's one of the big problems is how
not to get distracted for
instance we have a health problem
which is very important people are not doing well
at it but it's also distracting from
other issues that
have to do with the planet like the planet itself is
starting to die in various ways and so another
issue that's hiding there that's
going to have a much bigger impact on human beings
then the the pandemic is
going to be the climate problem
and if we imagine a child
being born this year well
at the end of the century there'll be only be 80
they'll still be alive we have to wonder
how are they going to get there could they
get there with the planet in better shape than it is now
rather than in worse shape
so one of the
biggest problems here is imagination
using conventional thinking so
Einstein one of my favorite lines from Einstein
is we cannot solve our problems with the same levels of
use to create them so
however we got here whatever
we think of is normal which is how we got
here it's not going to work to get us past
this to fix
these things and so right away one
of our biggest issues is to deal with stuff that
is outside of what our society thinks
is normal and people
use the term crazy for stuff
that's outside of normal so that's what we have we
have to learn how to be crazy in really
good ways and
thing that comes along with this is
if you haven't learned how to think skillfully
and to realize where you are requires
strong thinking you are in big trouble and this
is where most of the human race is
o here's where most people think we are
some memory from childhood in the
past take this as the
planet it's no longer so green it's
really kind of like this but
in fact we have these glasses on so we can't see
even where things are now
and the problem is we have to imagine into the future
what's likely to happen on the current
set of things it could be the future of the pandemic I'm
picking the climate as a big thing
so we have to be able to deal with all three
of these views at the same time
and the
problem is that science is our imagination
amplifier but most people aren't scientists
especially most politicians
most of the boating public in the United States
are not scientists or in the UK and
they do not
like thinking outside their
on the other hand what
has happened over the years is
that every time
there has been a threat that is
recognized by the general public including
the politicians there's usually a big response
to it war is the favorite one
you because
should be epidemics too but
most people are not taking co-head
seriously at this point
few places have but
most most places haven't especially
my country two countries
US and the UK really haven't taken
it seriously so if we look
in the past here too
all of them have had
some sort of threat involved
with them I suck the Empire State Building there because it's
one of my favorite engineering projects ever
the the whole building from
the raising
of the site the demolishing of the previous
site to occupancy was less than a year
and it was done by less than
3000 people it's one of the great building
in designing and planning jobs in history
and the reason it was done that way was
because the Depression Great
Depression in the US had
hit the very year this got funded
so part of the motivation here is we have to
get this thing up in a hurry and another part of the motivation
show everybody what building a tall building is
really like because another one like this is
not going to be built for decades so
the other entries
here which are all described in this white paper are
from world
war ii and the Cold War
so atomic energy Bletchley
Park the UK and
us joint project
on radar which was the prime
technology for World War two and
then a lot of funding continued
from the in the Cold War
and part
of that funded ARPA 1957
and ARPA funded a an
agency to do
research research that was not top-down
looking into information
systems and Xerox PARC
was one of these groups just funded
by Xerox about ten years later so
the process
of this right hand panel
in the Cold War came out of what
the group doing radar learned about
how to do science and engineering together on
a large scale to
combine research into the unknown with
doing practical engineering so
most of the successes in the realm of computers
research in the u.s. many
of them came out of the
process and it was taught to grad students generation
after generation
so a key thing about this kind of unrestricted
spending without
a lot of top-down supervision
that a key thing that is missed is
the amount
of wealth that came out of doing this spending
was spent to deal with a war but
in fact the most of the
industries in the u.s.
happened after World War two happened
because of World War two most of the technologies
that are prominent today came about
because of World War two and so
an incredible amount of wealth that has been generated
far far more than what the outlay there
was no need for Wars to
do this at all but
that's just the way we are and
to give you a sense here Park
came out of the ARPA community ARPA
research started in
1962 so
they're 12 to 14 years to get
the payoff and the larger
community started before here's a
larger look at the
DoD ARPA merging into Park
the two groups key
from Park came out of the Arco community the
internet was a joint project
and most of the technologies
that we did at Parc came out of
prior research at ARPA including some of our PhD
if we take an even closer look at park we can see most
echnologies that we're still using today
that I'm using to give this talk and
so we call
it eight and a half inventions the half is the Internet
all of this was done by about two dozen researchers
just 25 researchers
for the for these inventions
really inexpensive
about 15 million dollars a year in today's dollars
sheep returned
so far has been over 40 trillion dollars
this recreate created industries rather
than increments
o this prior art and all of these things but each
of these created industries of their own and
valuations in the stock market
of American companies came about because these
inventions so what's interesting is crazy
funding of research that
nobody believes in and believe it believe me nobody
did when we were at Clark they thought we were crazy
here are funding
people who are up to the task and you're willing to take
thirty percent of the results you're
going to get an enormous return
just monetarily but
also you're going to get an even bigger return for
society and that's what we're talking about today so
people say well Xerox
make any money on this but in fact they made billions
on the laser printer and paid for park many
many times over 50 or 60 times over
okay so here's the key point and
that is if you're interested in doing
oodness of the results you're going to get correlates
most strongly with the goodness of the funders and this
is not in terms of money most
funding today in the United States is
not good funding
in the terms I'm using good here
it's mainly funding for
goals picked ahead of time
stun top-down it has
making money as an end result and so forth so
what I mean by goodness of the funders here is
the funders actually understand this
other process which we're talking about today of
how do you get stuff that hasn't been seen before
and here's one of the great
ones of all time the guy
who set up the ARPA eye PTO funding
in 1962 and he
was a visionary
and he would not talk about goals you did not
have goals visions and
a vision is something that gives you a sense
of what the thing what the desire is but
nothing specific there's nothing you can work on directly
we have to pick the problems from the
vision so here is his vision that set off all of this
stuff computers are destined to become interactive
intellectual amplifiers for everyone
university universally networked
worldwide that's all he would ever say
yeah when they ask well how
going to do this he said well I don't know but I'm just going to give
money to people that think they can help
and will take a percentage of the results
and given what we're trying to fund if 30
or 40% of the results are successful
we will change the entire world now that's exactly what
happened so what so these
visions they're kind of like little magnetic
fields hidden behind the mountains
researchers can feel these magnetic
fields and if you're over on this side you
can feel
north appeal the magnet going
you're over on this side you can feel the magnet going everybody
starts moving towards the magnet and you get
a lot of different ways of approaching what this vision
is and that is what happened
so whenever you get into trouble what
you need to do is to realize that most problem
statements come out of the current context
but most of the solutions you need come out
of a context that it has to be invented so
you need to retreat from goals and problems to visions
visions of how things should be and those
will help think wider so how did this
stuff work back then well this is the Pentagon
Department of Defense and there's a US Congress
every country has something like this
in 1957 the
Russians put up a satellite and scared
people and that
fear was enough to set up the Advanced Research
Projects Agency in 1957 and
these were the days of the Cold
War and so instead of putting bureaucrats
in there all the ARPA directors were scientists
mostly physicists and
in 1962 they funded the
information processing techniques part of ARPA
also stocked with scientists and a new guy
every two years the ideas if you stay in too
long it becomes your job no
your job is to be a scientist and what you're doing is doing service
so you do two years of service
helping the country and then you go back otherwise
you start buying a house and get a house Morgan's you can
start worrying about your job it's losing your job now
of course Congress still
wanted to be a watchdog they
wanted to be the top down so they'd all ask well
how is this relevant tell me why this work that you're doing is
relevant to the Department of Defense and unlike
today these guys would
that's not the right question to ask because
they didn't care about getting fired no that's
not the right question to ask the right question is how
is this going to help the country or this technology
or our society or our culture in general and by the
way these are direct quotes and
Bob Taylor who is an onlooker
there said well they'd stand up to the Congress
and in a polite civilized way attack
their nearsightedness their myopia because
they were scientific Statesman
so Bob talked
to Charlie Hertzfeld here Charlie
said well what do you want Bob and Bob said well
this network is what I want to do this is like 1966-67
Hertzfeld didn't ask
him what it was he said okay
Bob said well I was a 15-minute
conversation with my boss and
then Charlie said well
need to get this off the ground and Bob said oh about a
million dollars or so that would be like six or seven million
dollars today just to get it started and
Hertz fellows response
was okay and as
Bob said there was no R per order written there's
no bureaucracy nothing
the paperwork work was done about
a year later but they're able to start the
same day that Bob had this meeting and
that network that he wanted is
in case you don't get the joke here the
joke is this is how this thing that we're using that's
all over the planet was actually funded without
a proposal
just two people who trusted each other and it
was part of this vision of lick lighters
and what's key to this story besides what happened with
ARPA was that Bob then went on to set up computing
research at Xerox PARC and
so he took this way of doing things
into this Industrial Research Lab
and hired a bunch
of us to do the work so here's a key idea being
responsible doesn't mean you should try to control
this is the biggest mistake that almost everybody
in charge in bureaucracy makes
because they are held accountable by their boss and they're
buying a house and they're afraid of getting fired but
basically you
can't do it because
in the case of unusual research
if you're a funder you don't know enough
to come up with good problems believe me you're
in the wrong place and the wrong the
people who know enough to come up with good problems are out
research without messing
around with money they're not doing politics so
you just cannot do top-down on this
stuff and get anything
and it's the great people who come up with the
great goals what you do is support them so
being responsible means you try to manage
the support of these people who are thinking
about what should be done
realizing that perhaps 60%
of their efforts are not going to succeed
and the other thing that ARPA
chose to do was to
have no peer review
and the reason is that even great people
often poor judges
of other people's work just from rivalries
or just there in the different context
it doesn't seem relevant so peer review just
doesn't work at these top levels it
only works for mediocre stuff even
if you get peers and
in the u.s. like at NSF and it's after you never
here and there who do it you
never get a peer
so key idea top-down control kills
unusual research lots of money
gets spent you'll set up research labs all over
the place people will get hired people do things in the
US at least it is essentially
never worked
no labs had had a sign
all over going
down the stairs and walls the
sign would say either do something very
beautiful or very useful that
was the sign at Bell Labs either
very beautiful we don't care
if it's useful or very
very useful
but take it out there and go
to one extreme or the other so what you need here
is you need unusual thunders you
need unusual research managers and you
need unusual researchers
so back when I was operating
in the 60s and 70s it was ARPA
office of naval research Xerox unusual
research managers like Licklider Sutherland and Taylor
now the unusual researchers
interestingly enough are easier to find but
they won't do anything without
these first two slots being taken
care of so if you want a championship
football team you get a Beckham
you just don't go out and get somebody randomly
you get the best football player
in the world and hire him for your team and if
you want to do a Xerox PARC you get a butler
this is Butler Lampson
stratospherically Abel
and if you
look at the bell curve here
what you're looking for
people that are for your research labs are one
in a hundred thousand one in a million one in ten
million ten one in ten million while China has
140 of them pick
something to be unusual in China has
140 in this almost 5 Sigma
case and if you want to do this stuff
you have to find them and nurture them
and some
like this some of the researchers will be like
top scientists some of them be
like top engineers but some of them are going to be like this and what
I'm showing this this is an ad that I suggested
Apple make back 40 years ago as
to what personal computing should be about and
what I meant when I tried to get them to do was
to do this because
the real process here is not to have a domain
of right brain and left brain the
real process here is to have the
ability to fluidly
merge all the different thinking kinds
of thinking you have an aesthetic thinking you have
into the thinking you're doing
so here's another key point
ARPA spent a lot of money
net generation of researchers
consider that one
of the main research results I had to fight Congress
on it but they
said hey this doing
this vision is going to take 10 15
maybe 20 years so
we have to we we have to grow the next generation
of researchers and we can grow them in this
in fact almost all the computer people at
Xerox PARC came from this ARPA process
we were all young I was the oldest person there I was
old everybody else was
younger butler is only 27 that's what
Bob Taylor wanted he had seen all of us do our PhDs
okay last idea
here is this idea of mad money
mad money is money that
you have in your pocket
that you spend on a win
can spend it on something it might not even wind up working and you don't
money that is not aimed
at any purpose so
the number one thing you have to ask is
in any organization or country
is how much mad money is there
if you have a country that is SuperDuper organized
they will tell you we
have no mad money
we plan for every cent we can tell
where we spend every cent on this and we'll show you in the master
playing with okay so that's not very creative
and most companies
want to show the stock holders
that they've been spending this money wisely and
here's the pre turn on this investment they hate to show
line items where there's no return on investment
for five years okay some add
money so here's the way to compute
it for a company Ardi
and a company is usually five to fifteen percent of revenues
most of it goes towards product
mad money is
1% to 5% of that
so the range here
is about
it was just tiny fractions
of our total revenues
in a company and if Xerox PARC
was about 50 million dollars a year today
that means that
taking the stingiest approach
to this we're only gonna take 5%
of revenues for R&D and 1% of that form
and money still the first 14
fortune 418 companies
are going to afford to do a Xerox PARC not
a single one of them half
no matter how much money they have
and if you go to the higher end fifteen
percent of revenues like technology companies do
and you're going to spend five percent of that on the
art on the mad money
like Xerox did that
means every single one of the
fortune 500
fortune 1000 companies
so let's look at it four countries
if we start with say Japan and Germany or so large
countries on up spend a hundred billion
to five hundred billion for
country R&D I think
China's is about two hundred and forty
or two hundred and fifty billion dollars a year the
US has maybe closer to five hundred or something
but it's about two percent whatever it is it's around two
percent of the GNP
and so
we take the same mad money figure 1
to 5% bad
money for the lower-end is 1 billion a year to 5
billion a year of stuff you should spend without
worrying about what happens like Japan
spending a billion dollars a year without worrying
about what happens
and so a billion dollars a year is
66 Xerox parks
eems like should do
it right and
for a large company like China
or the u.s. it's over 300 -
over 1,600 Xerox parks for each country
what this means is
the world collectively
and especially the large
countries getting their act together
actually have plenty
of funding have plenty of
potential talent to
actually deal in an unconventional way with these
enormous problems we have the problem is is that
most of the people in the world cannot see that we have these
problems or cannot see them as threats
or fears or whatever it takes and
so I'll leave you with the final quote by Einstein here
which is insanity is doing
the same thing over and over and expecting different results
that's what's been
going on for the last 40 or 50 years in computing
for sure and
I daresay in most
things and most especially for
these 12 critical issues
that we have to deal with right now so
thank you very much and I
guess that was about a half an hour rather than
I talk thank
you well thank you very much Alan
for the great talk and very very unlikely
the what
what do you just talk
about a large part of it that you actually summarized
very well your note the
big hall and you asked me list of
you know 19 roots of
Rd management to solve immense war
challenges and but you also mentioned
you also I to summarize the very well
but you didn't mention your talk is that
there are tremendous barriers and for
people to follow through your advice and that
90 rules including you know give
you the money you give all along and
the mad money and give
the meta money and the less of great people to the great thing
don't ask them how much they have professor
made and but
people just couldn't do it so you measured
about this barriers and you have this
concept of a six core buyers in the so-called
oestrogen paradox
could you talk about that a little bit well
so a good a good place good
reference for all this
is the Wikipedia article
that is called cognitive biases
I should have put that up in a
slide cognitive biases and at
lists I guess probably closer
to 200 things that are wrong with
human brains so this is the
result of anthropology and psychology
over the last few hundred years one
of the things we've been interested it is
why are we such poor thinkers why
do we require so much training
to do good thinking
why was science only invented a few hundred years ago instead
of a hundred thousand years ago and so forth and
so there are many so
something that every scientist knows about
is called confirmation bias so
when you have a theory
confirmation bias and scientists do it to is
any evidence that will strengthen
the thing you already believe you
give it much more weight than evidence against
the thing you already believe it really costs people
something to unbeliev something and of
science you're not supposed to believe things but we
do anyway so the ostrich syndrome
we've seen it in
responses to the pandemic
well it's not there
it's not important well even if it
is there we can't do anything about it
so all of these things are
there are expansions of one
of the things that is deeply built into all mammals
that probably
most animals that can move is
when you're in trouble basically
you should either find
a place to hide from it
or you should prepare to fight so
it's called fight or flight and this stands
in my way people deal with things verbally
so it's it's a part of human
nature and people are willing to get
past their human nature when they're really frightened like in
a war yeah and
you mrs. battle people you know when
people really got a frettin they are see another
story you mentioned about
you know Dhamma today what we call it
a modern time it started was because
code walk because the is a threat people
filled from silly union that time and also the
fact that so it is launched and sfotch so
uh you know when when people says
it's danger then they may start
to act and acting in the right way you
know so uh fear my question
is it's been damaged you know the last four
months or five months now it has
coughs the were so much trouble you know economic
wise and people like wise
thing you know is that big enough threat that
you think you know can waking people up and
start doing things in the right way
doesn't seem to be
has depended on
how the government
governments structured
like New Zealand did the right thing and did it quickly
the u.s. didn't do the
right thing and is doing the wrong thing
and it going in another wave of
it so but I think overall
the expecting
know mainstream people to
do the right thing here they just can't imagine
it until for
instance maybe a relative gets it
so when a relative gets the
disease and baby dies from it or has a terrible
time with it or they do then
they have this local way of assessing things
but for war
is prevalent in human
history and so
the threat of war has
been enough in the past for instance it was enough
radar started in the
UK seven
or eight years ahead of the Battle of Britain
yeah when the rest of the country
did not believe if there was going to be a war with Germany
and the politicians were trying to avoid it but the people
who did the radar did it anyway and Britain
came out on top when
the Nazis did fly over so that
was a reaction of what you might
call eccentrics who had some
funding who had the ears of some people but
a certain radar was not a national priority in
the UK nor was Bletchley Park
until there was an actual shooting
war but both of them had been set up ahead
of simile in the United States there was a billionaire
yeah who got radar started
in the US because the government wouldn't fund
it and he didn't care he happened to also be
a physicist and he also funded the start
of the joint project with with MIT before
the US was officially in the war so
that made an enormous difference but you
can't give the US government any points
con you can't give Congress any
points for reacting to what was actually
going on u.s. in those days was completely
isolationist okay
we have talked about the
kind of a history of modern
computer science development teeth from
the tapa days and that's
did I just say one thing got on the deep
but when the D the D was put on to
ARPA yeah it wasn't ARPA anymore
too because Parker has ARPA
in it McCain's nothin
because what that when the D got put
on the processes
that had been going on before it got changed
so uh wait
a minute is that yes today it
was that D and the processes changed
that got Taylor to try and get money from
a company that's how Xerox PARC happened okay
that's important Xerox
PARC happened because of DARPA couldn't
do it anymore okay oh yeah
but you know in
inth a say they accident remove the default a
while for a few years and then they put
back so and not less let's
to you you also mentioned in passing
very quickly you say the last week two years of
computer science the research you know there are there are awful
things that if they use old tricks and 20
so uh yeah that's
uh that's not long little
discouraging but uh you know but
I I still dare to ask e how
do you feel about a to loss to
sixty years progress in AI
Oh aye that's easy
because again if you're a biologist
I used to be and
you know something about
the human way the human brain works
an enormous ly large part
of every mammalian brain that most
brains are like
machine learning
they're basically quick
and so
a big machine learning system is kind of like a super
pigeon and you
get a lot of good things out of a super pigeon and in order
to do something right now that
we would call real intelligence
the name AI got stolen now
now it's called general AI
this is one way of telling that things have
gotten very very wrong so I
even lost four years well
I'm with you when the people are who are doing
it co-opted the
name with a subset of the meaning to claim success
Wisie called AI well people
wanted to have success and so they decide
work on the hard problems which of course are the cognitive
parts and so the
cognitive stuff has been very poorly funded
over the last 40 years really
poorly I'm speaking from the standpoint of the
US okay because I'm mad weird
and you know some
people have worked at it one
way or another the the
people at Vulcan Labs we're looking at cognitive
AI Doug Linnet
of the site project looking at cognitive AI but and some
people have started looking at now with this drive
towards explanation
and drive
towards having a user interface
that humans can deal with that
user interface if it's going to work in human
terms it has to have a context similar to
common sense world that human beings have hardly
any work has been done on this percentage-wise
in the last 40 years and so
I think that the
what ml
has done is
impressive and fun and it's
needed but I wouldn't confuse it with real AI
not even a little bit
whoa at
least you agree you know machine learning as Pascal
you know through the tremendous progress in the last ten
years and the deep learning in particular
yeah now we finally have some networks
that can work can solve problems
you know now you have machines that can recognize faces better
than most of humans and
solve a lot of real-world
problems but do you have a pass
am I like some song you know and then
let me come back for the Cosima of the means theme of the conference
here is the next decade of AI so
what's your prediction Oh excitation
oh your region you meant about
vision or Cisco I think the first thing is
to take the term intelligence seriously
really seriously just
like one of the problems with
the thing called computer science is
that term science was taken seriously
in the 60s and
the 70s but now it means engineering in
United States at least if you talk to any
computer science graduate student and ask them what their
field is they'll give you an engineering explanation
not a science why so that you
term that got co-opted is object-oriented
meant something
in the 60s and when it became popular
because it was very successful Xerox PARC
everybody wanted it and so an enormous number
of languages started appearing that were called object-oriented
that weren't at all see this happens
over again because people want to be successful they
want to feel part of the winning team
you can buy jeans you
know jeans that we wear that have
the Harvard logo on the back do
you want to feel successful I'm
conniving cotwell this is part of the brain
it's called the magic of contagion
is these guys are successful I
want to do what I'm gonna claim I'm doing it I'll do a subset
so yeah so take it seriously
draw a high threshold for what it
means and don't worry about you know quit writing
papers on stuff that
isn't interesting what about one
of the reasons that ml got going yeah
is because of kohonen
showing that certain
kinds of matrix math was
equivalent to certain kinds of perceptrons all
of a sudden academics could write papers with math in
them if you look at what that math
does it's not that interesting or impressive it's
basically a big correlator whereas
judea pearl who I think is
a guest of your conference I hope he's
a great play yes he calls
ml curve fitting
yeah if you call it what it is you
might want to work on something more if you're
working in the field of AI if you're calling the little thing
you're doing AI even though it's using
thousands of machine cycles and special chips and
everything but if you already working on a subset you can't claim
anything interesting about the large field so
you know main advice to give
to any young researchers hey get real
don't worry about yeah
there's always no question about
having the right papers to get your PhD and
having a deal look the people who
make the breakthroughs here are people who just don't care
about that yeah I don't want
I don't mind the funders funding
everybody else but if the funders omit funding
the people who can do the breakthroughs because of
the orthodoxy of this stuff that
isn't it then it's a disaster and
I think of a lot of what's been going on in computing the
last 20 to 25 years in general as a complete
disaster a great
point you know I I'm sure many audience
I have heard you uh you know many
in the audience has heard your demise and
I want you you know your
memory data but you know you you you you misses some query good point
you're not on paying too much attention on
paper publications and papers and
today maybe citations all those things so
one question I have for you is when
you were at a park right
when you were working on this the graphical user interface
you know and there's
not you know you people will argue they're not much but
much massing it you know that this is not
use differentiation a differential equation
and and how
do you convince your boss now this
work there's going to be a picture and in this change
the entire interactive mode of computing
and that will lead today you
know internet whatever in mobile devices
you know read the paper look at my talk I
never had to convince my boss of one
not one thing
because the bar and
the boss who's Taylor he
just hired me and said follow your instincts that's
the only direction that's the
and Taylor never told any researcher at Parc
any project that he was that he
us to do never great great
high teen don't make me talk or
neither did you think about my in so far
take a good with other and whatnot what am I know
I was paraphrasing what
are you your sister said in Chinese yeah but then
let me follow up on this same
topic you know how do you then generally
carry your idea be accepted in the community
you know and that that's more important right ya
basta you don't have to do whatever you want do better then how
do people come yes they your stuff it's
great okay now most of my ideas haven't
been accepted in the community that's okay but
is there a few right that your what you
got rented programming and there's a graphic interface those
not the
the main things that are called object-oriented
languages aren't anything like how
I described object-oriented no I lost
on that one because I thought
people thought they knew how to program and
the graphical
user interface at Parc and that's an interesting
different talk because we can run
all that software on today's
machines I give I give talks
using the exact software
from Xerox PARC in fact that Steve Jobs saw
so people can see what he saw and
if I were to do it you would be shocked at
some of the things I could do that I can't
do in a PowerPoint or
or keynote
for instance you could the whole thing was
live it wasn't a static presentation the
whole every part of it was programmable
and every part of it
was safely examinable so none of that
stuff is around today
so no what happened was
if you look at the these technologies
coming out of park ones where
there was no competitor
even really opinion wise
those made
it like an example of that well for instance there was nothing
like the laser printer right nothing
so company
immediately doing that there's nothing like the Ethernet
right everybody had tried to
do a local area network that is the only one that ever worked
and so that was
easy people thought they knew had a program
so the new programming techniques and
languages and stuff that we did a part were
more or less rejected
and people adapted the user-interface
to a subset of it partly
because the Mac was so weak
compared to the what we what we had in
park that it could only do a part of the problem so what
you're getting out as a subset the
thing to take
a look at is how bad the web and the web browser is
compared to the software of
the 80s forty years
till think so well
the web web and the web browser go back to like nineteen ninety-three
so a twenty seven twenty seven
years and in most cases you still
can't do WYSIWYG editing in the web browser
that's pathetic and web and
the web was developed on machines
that had WYSIWYG editing on them
so what you're seeing here is a bathtub setting
and failure of people to
do the work okay since
we are on this topic the graphical user interface
right and there you also mentioned object-oriented
programming idea yeah people did
you know people do their own a different way wrong
way and so you know just just a
little bit gaza yeah so what are you preciate lasting
jobs actually stoning your idea and let it angel
man look I I told both
Steve and Bill Gates
take the whole I remember we were
today the public domain Xerox
let us write papers wasn't
a secret visit I wrote a scientific American article that
was published two years before Steve
up that showed the whole user interface and everything
everything he saw in 79 was out
in the world two years earlier so there
was nothing secret about when I told this guy both
of these guys is look take the whole idea
don't take a subset of it and mess
it up take say stealing
I don't yeah no we're not none
of us had Park tried to get
rich that's okay by
the way to your point about you know don't
you care about X Y & Z no I
myself by the quality of the effort I put into
anything that's the only thing I control control
other people's opinion right boy great
plan the second thing is that the
people at Park and in
the ARPA community in general had an aesthetic Comeau connection
we really
really loved Lick Leiter's vision
I mean loved it like you
love a lover we
really wanted it wasn't it was an emotional
need that we had that we were called to it like
people are called to religion or to being a doctor
or to art and
so what so what we and the
the hiring the
selection process Taylor was looking for people who are essentially
artists who happened to have deep
scientific backgrounds
and like I said
there are there are people like that I was one of them many
of the people at Park were like that and boy
an artist
might be a little disappointed if somebody doesn't like
painting but not much
like because that
isn't why you're not painting it for to
sell the thing yeah artist
art is about I I
have these feelings that I would like to
manifest so that
other people get that's what you're doing
and if if you feel like you've done the highest quality
effort you can possibly do on this work
what what more can you do you can't
let some okay hi
III need to pull your bag of it I better you know so
you you have you know in park on your
peel in your career you have event little awkward ideas yeah
but there was a story you actually went to visit
y combinator in the
in 2015 you did a graffiti on
their slogan yes they say they
will invent the product people
want and you you cross
out want and put a leak so
what is what was your thinking behind it and
yeah well we want sugar
we want that we
want salt
we want companionship these
are biological need biological wants
there dries we want stories
way to make money is to
make a technological amplifier for any human want
because a technological amplifier
in the Industrial Revolution creates
essentially a legal drug
1/2 can you have these things
we have to have these things so let me just explain
the last thing and they don't quit although
what a need is is something like education
everybody needs deep
education but many
children don't want it marketing
people hate the idea of needs marketing
people want to sell to people's wants
educators are trying
need and they're trying to figure out how to help them
gain this what they need but it's
we it generally requires a lot of work to
deal with a need and much less work to deal with a lot
say so our society
is messed up in the over balance
between the fact that you could make easy money supplying
nee supplying wants but
it's gotten out of the habit of doing work necessary
to take on a new need like
learning lots of things you didn't know before
that are completely new ok great
you know the interest
of time you know I I need to stop here
I want to stop with one last question you you only
have you know you only need to answer you one sentence you know
if you if you want to name one great
product now you have a seen in the
last 20 years what that
great product
the most the most fun thing that
can be found online right
now it's a company
it's probably going to be a product
but you can use it right now for free
so the most
fun thing I think over the last 20 years in the
computer realm is is
called croquet
croquet is
version is the fifth of five deep
research efforts over the last 20
years to see if
the thesis of Dave
Reed at MIT and the 70s would actually work so
it's a mass of something
that is as large as the Internet and
massive deployment of pseudo time
computations is this is up at
the level of importance with tcp/ip
ok also I
have to cut off here and again
I was thank you very much for your enlightening talk and
the interactions we have and I know
we can chat back space for another hour or
two you know and there are also many questions on the audience
but I have to cut it here was thank you so
much and hopefully next year in the be a conference
will inviting you coming again and we
have another fireside chat thank you Thanks
with a bid and
salud even a jab
you know a machine apart you know show me doc hi
just wait a few minutes we want to yeah
okay yeah we should take
a photo yeah so this is a virtue a
way of taking photos kind
of sweet oh yeah oh yeah yeah oh
yeah oh yeah
first I need a wing one here
anyway I want to see how
one question from our director
of the BAA I okay
Alice thank you very much for your presentation
in fact I believe we represent where is higher
many people in China but I
have a question question is that why I
don't know Biagio that
have okay but only why and
Park in the world No ten
knots many market
than what so what what the what is the reason but
I I want to further extend your
idea on the Med money you know even we spend
for example in China a firm present but
it meant money but this do not park in
China what the reason for that that way
I think so I'm the last thing I think
the thing is to look to see what the researchers
think their goals actually are I
the reasons that let's say the
other ninety-five
presents money is no more money
compared with your mad money but
the clever people is limited
the no more money they cannot spend
their time on free it's flow its
flow region so that's why no parking no
real you know me occasionally
so princess in
biology and exam there are several examples
of things in biology that
are like Xerox PARC one
of them is called Jane Elliott labs in
outside of Washington
DC it was set up by the Nobel Prize winner Sydney
Brenner and
but generally so
one of the interesting things that has happened is
certain people like Paula Helen who
is one of the Microsoft guys he
decided to and he has billions
so he decided that he
was going to set up what he called Xerox
PARC done right and he
set it up in Palo Alto and he hired a lot of Xerox falcata
interval it's called interval yes and
well after
about six months or a year or so they invited me to
visit and
I was there for about a minute and I said well this isn't like
Xerox PARC at all yes
just then looking around
they had
some of these people who are so productive at
Xerox PARC er they were not being productive at all it
was easy to trace it back up to Paul Allen
who was not
content with just funding it he wanted to be one
of the guys and I
said to him I said Paul you you're a billionaire
you've done this that and the other thing but really
don't you admit that you've never done any real
computer science research in your entire life
know about it and especially what do you
know about it now why are you
trying to meddle with what's
going on you have to you're
not good if you're not going to actually try and use
process that Xerox PARC don't call interval
Xerox PARC done right you're
not going to get it and so the
thing I would do with with your young scientists is
for instance one of the problems
in graduate school today in the US
is the fact that the housing market
in the US has expanded
an additional factor of 10
in cost
over everything else by
inflation what this means is no
graduate student
at Stanford can afford to buy a house in Palo
Alto when they get out and get a job they
can't do it they have to go to Google
and even there it's
hard to buy a house and pal so the this
some almost nobody wants
to stay in and do research and if you
in and do research who is going to fund you well
not NSF because NSF
funding in computing is almost entirely like
engineering proposal left in your proposal
you have to explain to them how you're going to solve the problem this
is the proposal whereas
at ARPA you would scribble something
on the back of an envelope and say here's
the issue I want to work on and if they
thought you were a good guy they would or a good
gal they would say yeah that's that
was was in a response to the question
there now nobody ever asked me to justify a
single thing yet when
I was in grad school or at Xerox PARC they
just I walked in there they gave me the equivalent
of a two million dollar a year budget and I was
one year out of my PhD right
so one year but
they thought I had
potential and know
if I stub my toe on the thing while I was
just one of a bunch there and
the theory there was
was it's like going
out on a in a farm
what you're interested there is yield
not every single
thing being right you you're
growing everything as well as you can but
not everything is going to grow and no farmer
cries about that it's kind of the overhead
and the same thing is not
succeeding in your research when
you're doing this kind of research it's not failure its
overhead it's the overhead
it's like in football
the percentage of
goals made by a football player is really
low compared to the shots on goal
right missing
a shot on goal when you're playing football that's
not an error that's overhead it's
a difficult game and you're willing to pay
people immense amounts of money to
every once in a while score goal
in a game that's the way it works the
sports world understands is completely
because what you what
you're trying to do is something difficult a big
problem with bureaucracies especially large
ones is they are hiring so
they have to dip into the middle of the bell curve so
you're getting people that have to be managed
and all of a sudden you have started having layers
good point and you have
you have accountability ya
know you're much better off so in America there's the American
MacArthur Foundation
probably heard of that they have this thing called genius
grants so every year
they give five years of support to
35 artists in some field period
are no goals behind its whatever the artist does it's
just to get artists artists can't not do their art
right that's the definition of an
artist so if you get somebody who takes direction
in this area you
don't want them yeah nothing you don't
want them because they're not following their art what
you want is to get the best artists you
can get in there and take 30% of the result and be
in lighting you know extensions
hander you know there
is one thing I hear
is you know I look at through your 19 rules of
you know managing
the great lab
operated research or math managing man money
I think you imagine a number of them meter no control
don't contract control bill Domingo don't don't matter
are wrong and also funding people
not project which is which is exactly
what Pai is doing so
we're not funding project we have found in people and
we're also funding young people and
we don't want people apply we say you're
because you're gonna hear some money so uh
so high I remembered and
next time when I've got a challenge how
say you know dr. Allen case said
that okay so thank you very much Alan
and I hope well
I will hope to be alive the
next time this comes up oh that would be very still
that's thank you thank you
ha see ya see
ya Thank you Thank You Alan thank
you John thank you ah one
minute Randolph with John
good a su tienda utilize mm a deterioration
memory each Cena
your son daughter Judy
yeah yeah yo Sandow
to land a job in Korea each year by the hundreds in
Africa MobileIron Romeo should teach for GCC
that you're an Indian today sadhana yogic
on yadda yadda cadenza
you what yeah Lana and
you Felicia your country you know your G
also too Linda Janelle oh thank you and truly know
or not hi oh you got control of anonymity
and AI that you go to 90 okay
she didn't bother to my food are you good honey yeah
viola woman Ellie the cultural
nah homie kun kun Han holiday
honey you home ji Yi Li AO at the child home
at the funny head Tommy Turner
- Turner family I wore my
height and Halle lonely at home holla maybe
sing don't sing when she
pairen-chan gorilla City Arizona whoa
michio kaku la sandwiches jr. Shinjuku
now awamiya Chihuahua well
action I equip our
team Li Xiu Qi Gong ha
NK y ou you go feed her to me the mania
Archer the best way to predict the future is
to invent it Joshua
yahooeys huellas a function should
you travel I wanna set
or the session have a compartment
or chicken I am wandering a Google the
entity sudenly shyest
reaction concoction chronology order
at which another country also you
owe me Australian identity you don't wanna
tada super dad yet
sonic another child at longer that I see newly
particle will a maybe Holloway light
well actually