david-nolan.txt

Transcription of David Nolan's talk "Idée Fixe"
given at GOTO Chicago on 16th May 2017

Taken from here:
https://www.youtube.com/watch?v=lzXHMy4ewtM


[Music]
if you're in any domain where
problem-solving is extremely important
like software then you're acutely aware
of this story about to tell about fixing
software bugs which is that you sit down
to fix a software bug and you're like oh
this going to take me 15 minutes and
then an hour later you haven't figured
it out two hours later you haven't
figured it out four hours you're like
why this seems like you know you have
you have this notion of where the
problem might be but somehow it's not
working
it may be eight hours pass and finally
you throw your hands out there like I
got to talk to somebody else about this
problem you walk over to your colleague
Steph can you tap them on the shoulder
like just you know give me one minute
let me just describe what I'm seeing you
start describing the problem maybe they
maybe they even like say something like
ask you a question and then like a bolt
of lightning right it just hits you why
didn't you see it
why couldn't you see it and it really
wasn't completely your fault right
there's something about ideas fixed
ideas which actively obscure our ability
to assess alternatives and you know
often you could usually the way to
escape and we know this is to have to
either step outside yourself describe
the problem yourself or to just get
another opinion but I don't really want
to talk this talk isn't about the fixed
idea at the level individual it's really
the problem of the fixed idea on the
software industry at large for a culture
that sort of prides itself in innovation
and invention there's a surprising
amount of fixed ideas a kind of rat's
nest a fixed ideas and I have some you
know you know armchair theories about
why that's the case and we'll talk about
that and also talk a bit about how one
might I mean I don't know if there's a
solution to the problem a fixed idea but
I think there's a way a self-awareness
we can have to know that we're actually
inside a web the fixed ideas and to keep
ourselves open so that when the
alternative presents itself we are ready
to it
and make real progress so a little bit
quickly about myself I'm a software
developer cognate ex cognate ex is a
software consultancy a part of the
reason why I'm so fascinated about the
fixed idea is that caught at cognate X
we we basically built the company on non
mainstream technologies list functional
programming and we convinced companies
to use this stuff right I think the last
time we check about 15 of the fortune 50
are one of our clients and we people pay
us to write lists and so how do we do
this right so we have to actively
convince people to dismantle whatever
sort of fixed notions they have about
how software development happens and
convince them that we have a a better
way one of this probably the fundamental
beliefs that we have a cognate text is
that there's a lot of incidental
complexity in the software industry I
mean you know definitely I'm not going
to talk about things at level of
business but certainly at the level of
managing the development process and the
development process itself there are
many fixed ideas and you know a lot of
our job is simply breaking those down
and showing people that there's a
different way one of the things we do a
cognitive effect and besides consulting
we also maintain you know a suite of
technologies that we use when we solve
problems the CTO is Ricky Ricky created
the programming language closure if you
haven't seen any of these talks or if
you don't anything about closure it's
okay I highly recommend it he's
definitely one of the definitely one of
the brilliant minds working in software
and he doesn't an incredible job
breaking down a lot of these fixed ideas
so closure is a functional list of
targets the JVM there's a cool neat
thing that rich did here right which is
that you know the calculus for choosing
technologies is rapidly changing it's
totally different from how it was when
you know this was the thing and they
were actively teaching it at MIT right
these days we want to solve problems
that are way harder and a lot of the
things that you decider you know there's
a certain set of libraries you want to
use and all this stuff so when Richard
key set out to create a new programming
language he's specifically
that not going to do a new interpreter a
new runtime a new compiler the JVM
exists it's huge it's not going anywhere
no matter how many posts people run in
hacker news the Java is dead it hasn't
happened yet it's not going to happen
the same is true for JavaScript
JavaScript simply isn't going anywhere
web browser is not going anywhere this
stuff is only getting more important if
you're following mobile development you
know you're probably aware that adoption
of react native is surging because to
date it's the only sort of like
reasonably excuse me efficient solution
for tackling sort of the heterogeneous
nature of the mobile landscape but
clojurescript is our answer there I'm
the lead developer of closure script I
work on the closure script compiler
which takes closure compiled to
JavaScript and the output that we
generated optimized for the Google
closure compiler which you may not have
heard of it's definitely it's probably
the best JavaScript of JavaScript
optimizing the piler in the world and we
directly generate output that
specialized for that the other thing
that we do at cognate X is we have a
product called a tonic so this is route
what this is actually I think one of the
most interesting things we mostly
consult but see tonic is super cool
because there's you know even even if
you talk to like a functional programmer
that's making an industry whether
they're using Oh camel or using Erlang
or Haskell you know there's system
bottoms out in an up you know update
in-place database whether that's post
press or or you know my sequel whatever
right so they have this beautiful world
everything's functional it's that
there's business one global mutable
variable in their system which they
can't fix and so rich Hickey I think
about four years ago he had this idea
what if we got rid of update in place
databases what if we kept all the
properties we liked about post Koretz
and acid style databases and we simply
made it not be update in place and so we
took a lot of the theory around purely
functional data structures and sort of
the optimized version that he did for
closure and said
let's do this on disk let's have
thousand way trees over key value stores
and we can have a very efficient
relational immutable database and what
happens like what happens to your system
if you no longer have an update in place
database right what's going to happen
suddenly all these things that you
thought were hard or not worth doing now
become sort of trivial activities a lot
of the clients that interested in our
technology have been wanting to have a
database we're auditing this customers
profile what last year
I want to I want to run a query in the
database this person's profile on what
the profile looked like last year
against some facts that we know today so
doing queries across time on doing very
sophisticated auditing things without
having to build a lot of custom
infrastructure around logs or whatever
have you
satanic just give this to you for free
so client said you know that makes it if
it can convince them to get rid of their
fixed idea about update and place
databases people see that there's all
these types of things they can do that
they again they'd never ever considered
as a possibility I'm not going to say
anything more about cognitive X if any
of these things interest you you can
come find me we have a website check it
out there's a lot of cool stuff to see
so where does the what you know my
theory about why the fixed idea is so
problematic in the tech industry sort of
stems from a very pessimistic talk that
CP snow gave in 1959 at Cambridge in
fact I would say my talk is a as an
update on his talk my talk again is not
going to end on necessarily happy note
so that's why you know the beers there
you can go join your cause in the beer
afterwards but CP snow he he was a
scientist and an author and so he was
acutely aware of the value of the Arts
and the value of the sciences and he
really lived with lamenting the growing
rift between art and science in the 20th
century that these two cultures which
historically one which they weren't that
different in spirit you know science is
looking for some objective truth right
in the universe that's out there
art is looking for some truth in the
subjective reality of the human
experience right so they're both
investigative disciplines and their
methodologies just happened to be a
little bit different but what he was
saying was that he was witnessing this
riff and in fact it was there was a
growing disinterest it's not outright
antagonism between these two things
whenever you criticize what's happening
you make people angry so a lot of
scientists were angry that he brought
this up the idea said you know it's not
happening or they're like who gives a
shit and I don't think that attitudes a
good one and it's not and it's something
that I think that you see in definitely
in many corners of the tech industry I
myself feel sensitive this to my
background is not in computer science I
don't have a computer science degree I
just happen to hack on computers when I
was young and then when I often got a
film degree and somehow loop back into
computers because computers are cool and
I like them and I found I found working
with them very rewarding but I witnessed
this divide firsthand in the American
education system and again I'm going to
rail a bit on that but I'm sure you can
find this problem in education system in
the world and one problem is that as the
u.s. moves to a system in which they
were more concerned about results right
what is the result like how do we how do
we want to find whether school is
working or not and the culture which is
obsessed with metrics and quantification
then you know you were going to do is
that you were going to focus on the math
and sciences and we're going to be
focusing on you know what is just the
result but coming out of a particular
school and this has an effect on the
education system and on the experience
of individual when I was a teenager in
Texas my experience was that not not one
time in my entire life
all the time I was a nice girl for
failures were there ever a time that
another advanced placement student in
any of my math science classes or the
ever safety in our class so I live in
our class for four years every people my
across our classes were either
quote-unquote regular students or quote
unquote remedial student so with the
three months are seeing about that is
that there's an absolute metric either
neither your either your maintain in
math and sciences either you know your
your your maintaining a 4.0 4.0 GPA and
go to the best schools or not right it's
just that this is that close hard and
simple but if you're doing art you know
why does the funny thing art is just
objective we already sort of notice and
so if you get into our to get into
drawing you quickly discover most of
your drawings aren't any good they're
just not any good and and the MDM and
the packs enjoyment you make a lot of
crappy drawings and once in a blue moon
we make a good one every now and then
you make a good one and that keeps you
in third right so we're very different
attitude towards failure if you if you
get into you guys then if you are sort
of in the - Sciences arts and sciences
and with really weird right what's weird
is that if you stay in the exercise you
excel in a really great popular class
where the best screw the pen gives the
real world but there are no right
answers there no right answers remember
why we're nobody
the Socceroos gurus the punches the crap
that was a really explained you country
could we come to a conference and be
like they were doing this crap kind of
cure you should really so so there's a
river disconnect or disconnect between
reality and the way that we train people
to get into the software industry so I
think I think there's certainly a
problem there we'll talk a bit more
about that in a second but there's also
a problem which in that I think you know
science is beautiful but there's a sort
of a positivist positivism it's a
science and it's natural right science
is constantly discovering things it's
really incredible I changed my sides a
bit today to make it like a
geographically relevant this is a image
of the Chicago World's Fair 1896
I think 42 countries are present many
many new tech startups where they're
showing off their new things I think
this is the first time anybody ever saw
Oh like a escalator or sliding walkway I
guess lots of new inching electronic
devices I think Tesla was there with the
presentation but really exciting right
the Industrial Revolution but this is
sort of you know the problem is that
okay that's cool there is a youth tow
peon thing in science but it always has
to be counterbalanced by the reality of
the cost of that thing right people only
think I mean it's easy to get caught up
in the dream and forget those all those
are dark side so 10 years later somebody
wrote about the the dark effects of the
Industrial Revolution in Chicago and
that was Upton Sinclair when you write
about the jungle right this was written
ten years after the World's Fair and it
was about the horrible conditions which
immigrants work in the industrial sort
of meat processing district in Chicago
which actually led to a whole slew of
regular regulatory changes around food
processing and I think this I think is
extremely important as good as it may
feel everything is actually not awesome
that there are a lot of downsides to any
choices that we make and we'll get to
that more in a bit but
you know so how do we how do we maintain
a balance of perspective this is
actually harder than it sounds so Alan
Kay sort of the Godfather of Audrey
computing and a bunch of other cool
things this readeth many great quotes I
like this one a change in perspective is
worth 80 IQ points and I think it's
worth way more than that but the change
the change in your perspective is
incredibly difficult and I don't have an
easy answer to this problem he gave a
talk called I think a couple years ago
called
the future doesn't have to be
incremental he's talking about
innovation versus invention but he has
this slide that we're really stuck with
me which is that you know we're wired a
certain way right humans are wired
around fixed ideas the fact of the
matter is 80 percent of us me standing
in front stage I'm included in this 80
percent of us you know we get our ideas
by a social validation right if somebody
tells me something I'm not even going to
think about it right if somebody says
this is good and I should do it and
somebody else is is good I should do it
and something else is it I mean like
well I guess I have to do it right what
when I adopt the idea is awesome because
other people
validate the idea they vet it and I sort
of take that idea on through a social
process without actually necessarily
doing my own investigation to the same
um this has power right because I don't
have to reinvent the universe to get
something done right there is power and
value to to the way that humans work in
this way but there's of course the dark
side I mean you could say all the
terrible things that have happened in
society right the wars and the horrible
things and you know how long it took to
realize that equal rights is a good idea
I mean it was because 80% of the
population we're totally fine with it
not being that way right it's just the
way that we are so this is something to
be aware of like you know that we we in
the tech industry we fit this equation
exactly the same way there is no
difference between how the rest of the
world works and how we work our bias is
to you know whatever 80% of other
developers say that's probably what
we're going to do but I'm interested in
you know can it you know what what are
the
one of the ways and processes in which
we can at least if there's a good
positive change that we can make how can
we shift how can we shift people to the
lower right quadrant of this diagram
right I think there's a way and a lot of
that is learning to keep our our minds
open you can do this sort of it so you
can start an individual level like if
you do math and science if you're
probably familiar with this book how to
solve that's really great I mean one of
the best things the first thing is first
you have to understand the problem Alan
Kay has a great quip about this too
which is that you know if you go through
the math and science and you're in
computer science you're probably a
pretty good problem solver right we're
actually I think we know how to make
problem solvers how to how to get good
problem solvers but he points out that
we're not that good at finding you know
teaching people how to look for problems
right how do we find problems meaning
often the problems are right in front of
our faces but we can't see them again
because some fixed idea is obstructing
that a different book about how to solve
a certain type of problem which I highly
recommend if you've never looked at it
is there's a book by Betty Edwards
called drawing on the right side of the
brain when I was starting out learning
how to draw or teaching myself to draw
somebody recommend this book and I read
and I really loved it and what's
fascinating about this book is that
people have this fixed ideal IQ often
they're like like in this room you
probably many of you got computer
science degrees and you probably would
say if somebody said I just not good at
math I don't have a talent for math
you'd go well yeah there maybe there is
a thing called talent some people get it
really quick but you'd probably say
anybody can learn math there's nothing
special about it maybe we should use a
different teaching methodology with you
but there's no innate thing that
prevents people from learning
mathematics and you know Betty Edwards
did the same thing for drawing right you
know a lot of people say this they say
this like I just don't have a knack for
drawing I don't have a talent for
drawing and in this book she shows this
drawing of Stravinsky by Pablo Picasso
and she asked her students to draw it
somebody who'd not
dribble drawing they don't feel caught
in a drawing and of course they look at
this and what they'll do is they'll draw
a circle for the face and withdrawal
circles for the eyes they'll go outline
for the mouth and they'll draw a
rectangle for the body and a rectangle
for the arm recognized and they'll be
like ah see
I'm terrible and then she'll ask them to
do something very simple she'll say take
the drawing turn it upside down and
don't try to see anything at all
whatever lines you see just draw the
line and then she even asked you to do
even more simple exercise divide the
drawing into 16 squares or equal sides
only draw the lines you know for each
individual square and without fail
almost anybody can do this pass because
it's trivial and then she says turn the
drawing right-side up it more or less
looks like Picasso's original drawing so
what is the problem what was the problem
the problem was your mind you have a
fixed idea about already your mind has a
programmed idea about oh I have to draw
a circle for a face right that is
already in your mind your brain
intercepts it and you're just going to
draw a circle when really all you had to
do was turn to paint off for a second
just think about what your eye sees and
when your hands are that it's mostly
about motor coordination most artists
they just they have the neck that they
had was I'm gonna you know I'm going to
remove my mind from interpreting what
I'm seeing and I'm going to just see it
as it is and draw it as it is and yes it
takes a certain amount of leaps but she
points out that it's not special right
it's not a special quality right is
something that can be taught there's
another great book about drawing by
nicolay DS and he has another amazing
point which is that you know you see
these people they say they can't draw
what does this you know sailor or place
a person plays golf and then what you do
say okay there's an object that they
know and you say well draw me a sailboat
and you tell this is a sailor and they
can draw it perfectly
all the portrait proportions will be
correct if you ask a person who plays
golf jongmyo golf club all
proportions will be correct and nick
lady says well the reason is because you
think that drawing is only about seeing
and drawing is actually about all of
your senses all right you need to engage
all of your senses and that's how you
that's when you know you can draw a
thing my point about this will come back
I often think that that the answer for
our problems is often right in front of
our faces it's right in front of our
faces if we simply pay attention to it
actually pay attention to it there's
another person who has a great point
about this just to drive at home just a
little bit more enough I'll point out an
example from science I think somebody
sort of knew this innately there's a
book called La Vie modem flaw at the
French book by this author George Perec
Donald Knuth said this is the greatest
French novel of the 20th century so you
should go read it but georges perec had
this idea he had this idea of the inn
for ordinary and his and this idea
pervades his novel and it's going back
to what I said how artists try to
perceive the world right things that are
there they try to see at see it as it
really is and not what the what your
mind may want to interpret it as through
fixed ideas force yourself to write down
what is of no interest what is most
obvious most common most colorless make
an inventory of your pockets of your bag
question your teaspoons and you might
think that's ridiculous
why why why would you do this this seems
pointless ordinary things is there any
truth to completely ordinary things and
yeah this is this is from surely must be
joking mr. Feinman
this is from his um sort of memoirs I
guess and there's a story of how he had
fallen out of love with physics because
you know his great talent had been
mostly spent for the early part of his
career building the atomic bomb so great
talent spent you know creating new ways
to kill lots of people at one time so
one can imagine he lost his interest in
physics is his deep interest in physics
but he comes back to it and he talks
about how he came back to it how he was
able to regain his love and he tells his
story about watching a student in the
cornell university cafeteria he watched
the students through all
places the air and he saw that it
wobbled and that the Insignia is
spinning around and that there was a
relationship between the wobble on the
spinning around it's like you know I
used to look at water coming out of a
faucet and that was exciting I would sit
down and write the equations for that it
was likely he took that as inspiration
he started writing down what he thought
the relationship between the wobble and
the spinning the Insignia and a
colleague came by and said what the hell
are you doing
he's like I'm trying to understand the
wobble of plates and the person was like
what possible use could that be
and of course Richard Feynman it turns
out that had deep implications for
physics and you it kicked off the
research that got him the Nobel Prize in
Physics it started by observing place
wobbling I think we you know people have
this notion that great ideas come from
you know great things but the greatest
ideas are actually often right there
it's in the most ordinary location and
that was sort of the philosophy sub half
of the talk and now I want to talk about
something concrete something that I
experienced directly so I was able I was
lucky enough to experience oh you can
change 80% of people's minds and if
you're paying attention yourself and you
know some prepping has been put in place
to allow it to happen so I'm going to
talk specifically about you my
experiences you I work so if you're
interested you work you do front-end
stuff you're you're you're excited about
it then you're probably aware of Xerox
PARC and the work that was done around
the sort of small talk and the Alto and
all the software they built around small
talk but you know this project similar
to the Fineman thing ah
Alan Kay's goal his goal was not to
reinvent computing or to create
objective programming languages or you
know do all that stuff
his goal was I want to make a dynamic
medium for children right
you wanted to make a computer that a
child can understand and that's what
kicked off the personal computing
revolution then Apple comes along and
says ok wow this is all this stuff this
technology that you made that was
supposedly just for kids this is going
to change everything and we're going to
do a different thing we're going to
figure out how can we take hardware that
cost eighty five thousand dollars and
people can't pay for that we're going to
figure out it stick into this tiny box
that we can sell for a couple thousand
dollars alright so this is a great
moment in computing wind right there are
no fixed ideas yet right Xerox PARC
they're doing their thing they're just
experimenting Apple comes along and says
oh my god they have all this stuff oh we
can't recreate everything they did like
you know there's all this virtual
machine technology all this dynamic
compilation we're going to focus on the
interactivity and the GUI and the you
know the bitmap screen and stick to that
and make this a thing you know the
future of computing and that's that's
that time that was an amazing time I
love that time I love reading books
about that time it's really cool but
what's fascinating is how quickly we go
from you know really thinking about what
matters right Xerox PARC really think
about matters Apple seeing their
research distilling what really matters
if you want to make a product high level
version of this research right to a
whole bunch of fixed ideas on computing
industry so when I came into doing
professional software development 20
years later people were doing things the
same way as they were 25 years ago
so the first Xerox PARC memo on
model-view-controller as a pattern for
doing UI development was a memo right so
I did it and I took it as like you know
gospel truth MVC you I have to be the
way you do it because if you went to a
bookstore and you looked at UI UI stuff
would be shelf whole shelves I'm doing
design have to do it this way it's the
only way and then you know you do it for
five years they're like okay I get it I
get it it's useful
it works there are things I don't like I
see that it's easy to make a mess maybe
people thought about this and you go
back and you're like what did they say
when they created this and you find out
it was a memo there was not tablets that
came down the mountain somebody wrote an
interdepartmental memo and that was just
like hey this is kind of cool this is
probably not that good we'll fix it
later
and somehow 20 years later I'm stuck
with that memo right how did how did
this happen right it's this sort of
predilection for fixed ideas and some of
this I think is part of this is just
like you know we want to be right you
know this is a challenge right you're in
a business you got to make money you
want to be right you're an engineer you
want to deliver a result you want to be
right but we have to always balance that
with the fact that like there is no
right right there's no right there's no
there's no formula there's no proof
that's ever going to tell you what the
right way to the uiui program is it's
not going to exist it's never going to
exist so a funny thing happened about
three years ago a technology called
react appeared it's a much derision
because Facebook came on to its stage
and they said we solved your problem and
they showed a bunch of XML literals and
JavaScript files of people were like you
know it gave them the finger so it sort
of laying any language for about six
months nobody really really cared about
it I also was like add that seems a bit
weird but a good friend of mine Brandon
Bloom like took a closer look I said
actually there's some interesting stuff
here you should sit down and talk to
somebody about this like talk to
somebody has developed it and you know
screw all the market speak on the page
figure out what this thing actually is
so I was at the New York Times at the
time he arranged a meeting with me and
him we had coffee I spent about an hour
because I had been doing UI NBC
programming long enough to know know
what I didn't like
I was excited about functional
approaches being an alternative and so
react seems functional they had an
Augmented veneer and I was like is there
truly a functional mindset in the design
of the system and over coffee I came to
the conclusion yes there was it turned
out the original creator was a standard
ml okay
Kamel person in you know they were doing
NBC at Facebook doing backbone and they
sort of hated the spaghetti massive
state they'd constructed and so they did
react as a four-month prototype that
turned out that this idea had legs and
it was at least as fast as the backbone
thing even though it sounded crazy so he
had already gone through that I talked
to one of the other developers and I
said okay if that's what they went
through then maybe they're on to
something and then I went home and I
messed around with it and I wrote this
blog post called the future of
JavaScript MVC frameworks and this was
the most popular blog post I'd ever read
like 10 times more proper than all the
posts combined I think they were like
600 uploads on Hacker News like 150,000
page reads like complete page reads in
like a week and this sort of kicked off
in many ways the sort of the
reassessment of react what was funny is
that I wrote this for the closure script
community right because I was doing
closure script on the side and I was
like we need a functional solution to
the UI problem functional programming we
like it but if we can't solve the UI
problem I mean what real benefit are we
getting from our approach and it turned
out I made a naive demo and
clojurescript
and I showed that the naive demo and
closure script is faster than the naive
demo in JavaScript with react and it's
faster than of course you know the
typical backbone type thing and now fast
forward three years later react has
Eclipse ember angular I mean everything
it clips everything so this is one of
those rare instances where you see a sea
change right so the JavaScript developer
that's sitting down to write a react app
they are they are a functional
programmer right when I wrote this blog
post I said all this dividing up your
state into all these objects in your in
your UI that's a bunch of bullshit put
it into one single thing you when you
write web services what do you do you
have all these web services they all
talk to one thing the global database
why should you have a different design
pattern just because you're programming
me while it doesn't make any sense
me maybe you read about in a book and
somebody thought that made sense but I
say no it doesn't make any sense none of
your other systems look like that what
if you just did your UI is the same way
you do all your the system six months
people thought this was too crazy but
within another six months Redux became
the de facto way to do react apps Redux
is a single atom app State design over
react so that's interesting that's food
for thought
breaking through the occasional time you
can break through fix ideas so I'm just
going to end this I have a just few more
slides but before I move on
one thing that was introduced
interesting to me about this is that
just because something is good doesn't
mean okay you found a good thing you
remove all critical thought what was
fascinating to me was that within two
years of Reax introduction people went
from reactor school to being reactive
the only way right the cargo cult thing
starts so quickly and I was invited to
the very first Facebook conference and I
pointed out that actually react as great
as it is did throughout the baby with
the bathwater with respect to some
interesting features that are possible
in traditional and bdc designs that are
not possible in react or really annoying
to do in react so hopefully people don't
just keep on keep on this trend right
react is not the end react is just the
beginning of yet some other improved
things and hopefully where I'm stuck on
this for another 30 years so I want to
end on a final point about the fixed
idea which I think has deeper
repercussions if you take what I said
earlier about the problem of the fixed
idea and the fact that fixed ideas are
very prevalent in our industry then it
should give you pause because there
we're moving into a certain kind of new
future which this is more of a problem
than is today I think it's already a
huge problem it'll be even more of at
home so if I'd shown this talk about a
year and a half ago you wouldn't even
know what this is what I'm showing you
right here unless you're from China
Korea Japan and even then you might not
know these are two old Korean men in the
park playing this game called go it's
about four thousand years
chances are you now know about it
because of this company probably never
heard of called Google they had a
company a sub company called deep mind
that's doing a lot of experimentation
into deep learning and reinforcement
learning and they had a very interesting
project called alphago which was secret
for about a year
as long maybe two years and then they
tried it on a professional and they
realized that they had something big
people have thought this moment was
about twenty years further out they
didn't think it was as close but gup-d
might not they were very close because
of the good result against a Chinese
professional and they decided to do the
match of the century
basically the NGO equivalent of deep
mind sorry deep blue cats progress was
deep blue and so as a undergo my mom's
Korean I kind of go as a child I didn't
have plans like 15 years ago so I'm an
amateur player I like it it's fun and so
I was able to follow along basically
what's going on
with the help of the commentary of
professional commentators what was
really cool so this is the cool part
about this AI stuff what was really cool
is that if there's a bright side to this
to the AI thing and I don't not talking
about general I mean like the type of AI
systems that we're probably going to
build next year or the year after that I
actually am currently on a cognate at a
AI consulting thing this has some
interesting points for the type of
system you might build which is that the
professional commentators while they
were watching the alphago games they
were constantly surprised by its moves
because alphago played moves that humans
wouldn't play because even after 4000
years of optimization people had
developed very fixed ideas of what the
right move was there's all these
proverbs that have been passed down
passed down passed down about how you
should do a thing and alphago because of
the way it was trained it didn't have
these biases it didn't have these fixed
ideas and in fact it more or less kicked
off immediately kicked off a revolution
in the professional community about
patterns so many patterns that people
would have never considered before they
now take very seriously
I highlight this game because this was
the game the only game that he said Oh
won and he played moves 78 and this is
where alphago made a mistake because
alphago said there was only a 1 in
10,000 chance that this move would be
played so you would use the probability
to figure out should we examine the
variations for this move because it had
a finite amount of time to consider
things so it would give more weight to
things which looked more likely so it
didn't look at this and it made the
wrong move and he said Oh went on to win
the commentators were like you know that
you know they said this was like the
divine move and go there's this notion
of the move that you get to play once in
a lifetime that changes the flow of the
game and actually this is a good chance
that this is the last time a human
professional will ever be able to beat a
system like alphago so this was a divine
move this move a move this good probably
won't ever be played ever again not
between a human and a computer
technically what's fascinating is
technically this move doesn't work after
extensive analysis humans have
determined that this move shouldn't have
work it's just that alphago didn't look
at it it didn't look at it it didn't see
it as a viable alternative right so the
new version of alphago that's going to
be presented next month and wujin in
China playing against the top players
and the Chinese players in the world it
has a new algorithm an adversarial
algorithm in which that if alphago has a
few moves chat it says are very low
probability there's an anti alphago that
will double-check
they'll double-check that work yeah
right but this goes back this goes back
goes back when I said in the very
beginning when you're looking at bugs
right this notion of you need an
alternative opinion right diversity of
viewpoints is not it's not a nice thing
to have it's an imperative right how can
you know that you're making the right
thing if you don't have a counterbalance
so while this is a great story and I
think deep mind is very they're very
rigorous and their approach and I think
their choices are based on both research
and insight the thing that we should be
afraid of
is that most of us might not have that
level of insight right most of us will
probably will not do the right thing and
the systems that we build will simply be
a mirror of all the bad fixed ideas that
we have and we're simply going to put
those bad fixed ideas into the systems
that we build so this is a future that I
think we should be a keen ly afraid of
so again I you know not a happy note but
I hope that you have your arms is a bit
more critical more of a critical eye
change is important but in order to have
change you have to have an open mind and
you have to be able to to believe that
you know another world is possible
right but the world that we currently
live in with respect to how software
development is done is not the only one
and surely there are better ones that's
all I have is you have any questions or
interesting things to talk to me about
hit me up over a beer thank you
[Applause]