Alan Kay Interview by Dave Marvit (2013)

From Viewpoints Intelligent Archive
Revision as of 22:50, 6 December 2017 by Ohshima (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
are there important
problems for  technology to solve you
within the scope of technology the ambit  of technological solution and if
so what  well
I think the you
know if you take  the route of technology technique
it  really means things that humans
make so  it's a much
larger word than the way we  use it today and
the Latin root was ours  from which arts you
know the word the  arts
didn't mean the Fine Arts back
then  but they meant the art of building
this  and that and the art of cooking and also
the art of dancing and the art of music  and stuff but basically
the term the  arts have gotten kind of co-opted
by the  Fine Arts in our era
so if you take a  look at the the
synergies between human  beings and their creations
you know  they're
more I think would be very  difficult to define
a human being  without looking at either
physical or  ideational
artifacts
because we actually  create language
out of a propensity  towards
language as we're growing up and  that language has a lot
of commitments  and meanings
to what the meaning of the  culture is so
so one way to
you could  you could imagine a
degree in technology  that was different than a degree in  engineering
that just included  engineering is one
part of it  that
could actually deal much more with  the
whole system so you
know I argued  with the young kid there
the  eye-tracking kid because he had started
  off with this idea which i think is a
  corporate mantra that
you know the  future
is going to be humans
fitting - I  mean machines fitting to humans
rather  than humans fitting to machines and it I
  don't think that that split actually
can  exist there's no question
but that the  machines we use change us
well they  don't change us we
change us machines  are just doing things and
it's it's our
interpretations of the things that  change that
change jeans are just  catalyst
they're that's because you know
  we have these cause and effect things
  it's like you insulted
me and
you know  unless unless you actually poke
your  fingers in my eye all you're doing is  creating
vibrations in the air and every
other every part of the insult is  something that I have
to interpret  myself
and react to right
so if people  were actually dealing with
actually going on you couldn't be  insulted
unless you chose to be so
so  what I asked you was it
was only  response was reasonable and you broaden  the
definition of technology has  typically used to encompass
an enormous  scope you're just which is cool and
we can sort of put that into question  and say okay given the enormous scope  and
then we're given the smaller scope  but I
corporate hat a little bit here and  trying
to think of okay we have
this  muscle this techno developmental
muscle  at a large corporation and it
would be  good to put it to use to good
use if  that's even possible so
the frame us a framework in which I was  asking the question
  I think the well let
me be elliptical  again but if you go
back to this
  interesting phenomenon that
if you look  at
the daily use of
people with say  computers of whatever
whatever kind
it's  hard to find any actual new
any  technology in there that is newer
than  thirty years old as
far as the  foundational ideas
so the whole idea of
  graphical user interfaces is came
from  the 60s and the one we happen to use
  today was invented at Parc but it wasn't  invented
in isolation of previous ones  the mouth
both the mouse and the tablet  were invented in
1964 most people don't  realize that the
ARPANET started working  in 1969
and it just morphed into
the  Internet the ethernet was
invented at  Xerox PARC around 1973
object-oriented
programming yeah so if you if you go  down the line then
as I pointed out that  if you take a
look at what people are  doing on the
on the web what I see is  mainly better-looking
graphics than we  could do 50 years
ago but what
I don't  see is angle Bart's ideas
so Engelbart  routinely
shared screens
in real time  for the entire work
and so  you know it
was an in an integral part  of the system
it wasn't an app that you  use it was actually
and if you're gonna work with somebody  and often
you could see the other person  if they're
more than two people  collaborating their
initials would show  up on the cursor so you
they're pointing at and this is all in  the 60s and
there's nothing about it is  hidden so
what I see is something much  more like a pop
culture today where  people
are completely indifferent to
the  past and it's
not that everything in the  past was good but it's real
ose the stuff that was better than the  idea
like Engelbart was smarter than  most people today and
so they can have  all the pleasures
they want inventing  something and having
that egocentric  satisfaction
that I did this but  basically it's
just a form of pollution  because
talking about something that's just for  a single person's
pleasure we're talking  about something that which everything is  done
is Rama file to two billion
people  now and that you just can't allow
people  who you
know whether willfully or
unconsciously
are you know just  polluting
the system because they don't  know any better  yeah
on the other hand we've
got this paradox  that
a lot of the
solutions in the past  that were better than the solutions  today
are not the solutions that we need
  because the problems are different  well
no it's even for the same problems  so
if you talk to the angle bard Ian's  back
then they knew they were just doing  a version of
the thing it happened their  version is
was a better set of thoughts  than the web thoughts
which were really  crude and almost
nothing of vision in  them but
the angle bargains would say  well you know you know don't
build a  religion about our stuff because we
have  to do some more qualitative iterations
  because there's
so much user interface  burden in
doing this stuff so when we  were doing
the you
know the small talk  stuff with
the GUI and
children and  desktop publishing and all that stuff
  it's your ex part the
technologic  technological stuff was
relatively easy  partly because we had
some geniuses so  he had this guy Chuck back
you know  they're all in the wall over there so  Chuck
Thacker Butler Lampson Metcalf did
  the ethernet I don't know whether stark  Starkweather
should be in there did the  laser printer so these guys
are really  incredible
and we we
had people besides  me who
were very well-versed in
user  interfaces but in fact user
interface  design is this
because
it is a design and something  that
fits it's really hundreds of  experiments and
almost every experiment  fails
in one way or another and so the  people who do this
successfully are have
  tools that allow them to to try ten
  different kinds of things a day and of
  course because the users learn you have
  to get a new batch of users rather
often  to in order
to see how users react to  something like this the first time
out  that's true in the generative side and  also the
you do put out in the world change as a  result of what
game theoretic moving target  yeah and so
one of the ironies of the
park GUI which has been you know it's  two billion
people and probably a
  million applications now have been
done  successfully in it it's
actually a  bicycle with training wheels
nobody knows the training wheels are on  it and
the reason is is that I designed  that user interface for children
and
one  of the things that was not a training  wheel
on it was never accepted by
Apple  and because it wasn't accepted by Apple
  Microsoft never appropriated it and
  because of that gnome didn't you
look at the way these things trickle  through and
so that particular idea
  which was
recognition of some of the  overlap
between children and adults  never is
not on any system today except
  except small talk which it always was on
  so what is that idea well the idea was  just that
it's not really a desktop that
  was Apple's misinterpretation
  even after we explained to them
not a  desktop it's or
if you want to think of  it as a desktop there are
an unlimited  number of them and they really didn't  get
that in that those a couple of demos
  we showed them and so if you think if  have
you know what could I do with an  unlimited number of desktops
answer is I can organize  but I
don't want to be application  centric because an application
is a  stovepipe what I really want is  something
where I can get every object  of every
kind that I need for this  project out
where I can do things with  them and I
want that desktop to remember
  their state over time because
most  people including children have
three or  four projects they're working on each  day
and instead what Mac
Windows did is to give you a situation  where you're kind of
and tearing things down and staying  inside of
applications and not being  able to integrate you
know in other  words what Apple and and
Microsoft did  was
was something that essentially
prevented what today would be called a  mashup
but you actually want it so if  you think in
object-oriented terms you  don't have to have applications
didn't have them at Parc you don't have  files
what you have is an unlimited  number of areas
which today you could  think of as web pages
and but you can do  all developmental
things on and that's  what I was giving my talk in
first so I  can
run so the important thing is it's  dynamic I get something
that's 10 times  as powerful as PowerPoint for
free and  there's a sorter so I can
show which  ones are those things that I want and  cool
and so forth so that was a one
of  the strongest ideas that we evolved
at  Parc probably around
76 or 77
  onwards never made it still
would be  useful so that so when I look at the
  interfaces today what I see because
I  know what it was and what
it could be I  see first a
bicycle with training wheels  on
but people can't see the training  wheels because they don't know
it's  supposed to be a bike and then the  second
thing I see after 20 30 years of  it is a
bicycle with training wheels on  it completely
encrusted with jewels and  rhinestones
because it's been decorated
  in a thousand different ways features
have been put on it but it's still got  the fucking training wheels on
it and so  this is actually the way things
go  because if people are not willing to  actually
like Apple did not want to hear
what the theory of that user interface  was even
after I was there for 12 years  because
they had expropriated it it was  now their thing
wasn't the
you know the  history of it was completely irrelevant  so
that was when I realized oh this is a  pop culture this
is the way it doesn't  matter how how good any
musician was of  the past or anything else the point of a
  pop culture is getting identity through  your own
actions a feeling of  participation it's not
the same as  where's Park was anything but a pop
culture Park was all about banding  together in
a pretty anonymous group of  experts
to make something grand
so the  part of the implication of this is
that  the the intellectual
ownership which  often brings with it resistance to
  change is is
it fundamental to corporate  models and if if
not then how do you  shake it well I
mean everybody has an  ego so
the real question any especially  any research
manager  ass is not does
this person have an ego  or not that's it's like
that guy saying  we're gonna have the
Machine fit to the  person
so the real real question your  ass you
always ask is what is the
  general expression of ego that this  person
exhibits and that the way they
  express is what allows them
to be
team  members and stars at the same
because you want to have stars dark  sparkles
full of stars but the key to  the
key to park and the arpa community  it came from
was how it cooperative was  how it was
able to cooperate was not was
  competitive up to
a point that was just  good enough to be
fun was mildly  competitive
but the truth was is
that  the 15 or 16 are per projects
cooperated  they swapped graduate students
they're  all working in on the same
vision but  they had different goals and
this gave a  lot of opportunity for
argument and this  was a was a community
that had learned  how to argue that's what I learned
when  I went to graduate school was for the  first time
in my life how to actually  argue because
when I was a kid I
  mistakenly thought that the
purpose of  an argument was to win it
  and but they you
know but argument in
the ARPA community was not that's sort  of what you do with debate
you know it's  sophistry learn
how to win a debate and  they
were not interested in that at all  and of course there were a couple of  exceptions
but I'm talking about  hundreds of people here
and the purpose  of these arguments was
to illuminate to  get out
different perspectives on  something and
park was a an argument if  you talk
to anybody who was there we  argued
incessantly with each other and  it frightened
Xerox actually because  they misinterpreted
it yeah taya well
I'm as strife  whereas
Taylor who was kind of the  genius
who had been one of the
funders and then was the guy who set up  Xerox
PARC he was a psychologist like
  the guy who initiated ARPA funding  Licklider
and Taylor never made
a  technical decision he trusted his  scientists
and his job as he saw
it was  to protect us from Xerox and
to set up a  climate
where these lone-wolf he he got
  most of the people at park a lot of
people were what you would call lone  wolves
he liked that because he didn't  you
know he wanted people who were  basically not amenable to being
managed  we needed management he
have a management structure and he  didn't
but then there's the question of  can
he set up something so that these  lone wolves will cooperate
when that's a  good idea and to
give you an example of  that there's no reason to cooperate on
a
programming language  pretty much everybody
at park could  invent and build
a programming language  or an operating system
so the idea is  anything goes there
but if you're gonna  do an alto
where and
the the mantra  there was that
everything that we do had  to be engineered for 100 users
so we  built a fake pdp-10 that
had to run a  hundred users as a time sharing system
and when we did the Alto we knew we were  gonna have to build
a hundred of it and  building a hundred machines
that in  today's terms would cost about eighty  thousand
bucks apiece  you need to have cooperation
and he got  it and
he didn't have to organize the  cooperation the
cooperation happened  maybe
even in a better way than he ever  dreamed although he had picked
us and  and it happened
because
this that and  the other well so one of the things
was  Taylor set up was
every single person
  who was at Park has to be totally
enthusiastic about the next person to  come in
it was a single black ball
and the reason that this
took forever  because
you know some people would know  this person  some
people would only know the  reputation person
would have to come in  and talk to everybody that would take
  days and it was
incredibly unwieldy so
  we grumbled however it worked perfectly  because
there were never any rivalries
  such as you would get when you just
plonk a new person who's really good  that
was the only kind they wanted there  so
there's no getting out to swords and  testing
out any of this stuff everybody  who's completely committed
to the new  member of the marriage
before it ever  happened and so so
by going through the  this enormous overhead
in the beginning  of the thing it
paid off many many times  over the decade
that Park was most  productive Taylor
had a lot of things  like that so let me
too much your time you should interview  him if I
would love to actually maybe  the nice thing is that you
Licklider I think was a little bit more  intuitive
but Taylor was a big fan of  lick
lighters and so he went
to town on  how Licklider did his thing and
applied  this when he was the
ARPA funder and  here's
a guy who funded the ARPANET
and  when Taylor
came to park and was in
the  position of setting up this lab of his  own
he was determined to try
every single one of these things that he  thought was a gem of a
principle to work  and he was willing to talk
was trying to do it wasn't anything  covert I
will make a point of that um so  just sort
of a wrap now maybe you and I  can continue some time down in LA
if you  were going
to if I was going to edit  this and tell the guys at Fujitsu
what  we should do as an institution to  innovate
more better different effect  well I think first
the there in Silicon  Valley now
  they should use the Silicon Valley  meanings
of the word innovate and invent  which
are I think made up by Regis  McKenna years
ago but so so innovation
  is taking an idea into the marketplace
  in Silicon Valley terms an
invention is  what we did at Xerox PARC which
is  dabbling getting
much closer to the word  new than news and both
of them are real
  art forms with real process
but they  have to get that clear because you  really
want to know whether you're a lot  of problems and companies is
the  confusion between those two things or  and
then they have to decide what
is the  cost of doing business for each of those
  I'll leave out in an innovation
here  just focus on invention so
so years ago
so Lickliter
had this idea that you  couldn't think of a good goal
while  you're behind the beltway in Washington
  anybody had the vision and he could say
  it in a sentence it was interactive
  computers as intellectual amplifiers
everybody on the planet pervasively  network worldwide
that was it and when  everybody
asked him what he was doing he  was just say that sentence and they'd
say well what about goes well we can't  think about goals here
you gonna operate he says well I'm gonna  I'm gonna fund people
rather than  projects and
so find really smart people
  who were interested in this vision and
  so so the Warriors
said well
you
know  this isn't that going to produce a lot
  of failures  I
think etc
etcetera he says well we're  playing we're not playing
golf we're  playing baseball some
ty Cobbs lifetime  batting average is 367
so like lawyers  say if we can bat if you look at what
  we're funding if we can bat 350
on the  whole portfolio
we will change the  entire
world so that's what happened and
so the weary er said well what about the  650
what about the 65 percent of failure
he says why it's the cost of doing  business and research
this is the number  one thing that companies
today do not  understand
they have to put in their  turns because
the way they think about  things generally they're not
romantics  the bottom line people
they're willing  to put a lot of money
into advertising  which is rather ephemeral
as far as ROI  but they think they
understand it and  part of their problems they do not
  understand research and invention and
so  they want to tighter rein on
it but in  fact what the old-time funders
did was  not to confuse responsibility
in control  Lickliter said
I'm responsible but I  can't be in control and
you know we have  to run this
stochastic lis  like like
baseball and
you know if we  get the best players and
we bet  reasonably over 300
we're just gonna  nail it and this
reasonable thing because if you look at  the return it's
astronomical like a good  return
on investment is like what 15%  17% I
mean the return on investment from  Xerox
from the
laser printer alone at  Xerox PARC was 20,000
percent was  zillions of
you know sharks didn't  understand anything except the laser
  printer and it paid for park hundreds of  times
over and yet Xerox
is worried  about all the things that weren't going
  well they're forgetting because
they  have confused making money
with making  money safely
and so they
always say well we're in business to  make money and
I would say no you're not  you just want to make millions
you know  research people want to make trillions
  because we're creating new paradigms
  here we're creating new industries
so  that industry we create created only
  took about 20 years to pass
the Ottoman  worldwide automobile bid industry
and it  didn't come from incrementalism
so there
  so a lot of it is just not these
guys  not being able to call a spade a spade  they
have a nice view of themselves and  they
think of themselves as you know  hard-nosed business people but
in fact  they're playing it safe and
because of  their lack of try
anything that they don't think they  understand
and because they don't really  understand Science and Technology they
  have no idea what the process actually
is and so they tend to want to be the  third or fourth person on their
block to  do something they're hoping somebody
  else will do something and they have  zillions
of words  I'm not saying Fujitsu does this but I'm
  saying it's typical of companies - it's
like the thing I said about change  everybody talks
gonna change this we're gonna have a  better process and all
the stuff but in  the end when
it comes down to am I gonna  change
am I am
I gonna go against  something I'm a my going to risk my  identity
my house mortgage
and so what  we call
middle managers disease at Xerox  PARC was people
who had gotten to the  stage where they're wearing more about  their house mortgage
than what their  actual job was and
I was leaking into  their job so
I mean Taylor had a million  ways of getting around this
and part of  it
was the got himself fired
even after  our greatest successes and
couple years later he got the National  Medal of Technology
  so Xerox was just
  and the irony was of course that Xerox
was that kind of company itself in the  50s
but by the time we got to the early
  70s the original fireplugs
had died
Joe  Wilson had died these
guys were just  like us if you go back to look at
these  guys when IBM wouldn't
accept the  prototype of the 914
these guys were so  and
there's a whole bunch of interesting  funny stuff
with consulting companies in  IBM not wanting to
dip in and delay of a  year and a half
and something like this  and these guys are just so pissed off  that
they use their life insurance in
  their house their own house mortgages to
  get the loans to build the first
  factories for the 914 of
course they  became incredibly rich
because you
know  they had a there were no really no VCS
  that but these guys have risked every
damn thing they have because they knew  they had and they invented
geography  twice the
first first time around they  invented offset printing that
was done  in the early 50s so
these guys were just  fantastic guys
but you know a decade  later Xerox is the fastest
growing  company in the US and etc etc all of
sudden things were very different there  was a more of a caretaker
management in  there and they had a lot of words
but  one of the secrets to Xerox
PARC were  was a certain agreement
a hard agreement
  that Taylor had made with Xerox
before  agreeing to set up this lab that
prevented
Xerox from actually permuting  any
of the research there for the first  five years and
that is when we got most  of our stuff done and they
signed that  agreement thinking Taylor would never
  use it he had to use it several times in  that first
five you sit permuting you  mean messing
with basically in any way
  zero and
that was part of Taylor's sales  pitch too because
none of us wanted to  work for a company yeah we're all by
the  way I was the oldest person there
as an  actual researcher I was 30
  Taylor the older people
who were our  mentors Taylor hired as advisors there's
  an advisory board Butler Lampson was
27  Chuck
was probably 26 Peter deutsches 25  or 24
I'm gonna suck it out cuz you're  making me feel
bad well but I mean this  is Taylor's theory
right Taylor I paid  for all of our PhDs and
and he knew the  ones
who had really drunk the kool-aid  you
know that that group of people just
  burned and
we know we didn't get paid  that much but we
burned to do this and  the problems the ARPA funding
was going  away and so Taylor convinced
us that  Xerox was the place
we could finish up  the grand dream
there and he had
this  insulation and
you know as an
he was lucky hugely lucky  that there
was a downturn in business  and the
berkeley computer company  corporation that butler
and chuck  another two guys on the wall
there had
  set up to do a system that failed  because
of the financial thing and  taylor got the entire company which
is  about nine incredible people that's
when  I decided to stay cuz I was consulting  for
the Taylor at that time and I
going to go to Carnegie to try and do a  personal computer
with Gordon Bell and  man when those guys came in
I realized  oh now we can do anything these
guys can  do anything and so
I just called up  Carnegie and you know got out
of my  agreement there and thankfully
they  forgave me after a while
but I knew it  was just going to be
super special even  when we had
we only had about 12 people  here starting off but we
had 12 really  really good people