
[Music]
so if you’ve had the feeling over the
last several months that society is
being torn apart I’d like to offer you a
simple explanation it actually is now
you know we as humans divide ourselves
into all sorts of interesting categories
naturally one way that we do that is by
organizing ourselves into cities and
this actually is a visualization of the
GDP of the United States split in half
and what you can see here is that the
most populated areas mostly on the
coasts account for 50% of the GDP of the
country while the entire rest of the
country accounts for the rest and this
is actually reflected in how the country
voted back in November you can see that
the bright blue areas there are areas
that voted for Clinton and the areas in
red where are people who voted for Trump
and it was 2623 counties that voted for
Trump and 489 for Clinton but yet
Clinton got a majority of the vote so
you know we see that this kind of rural
divided rural urban divide happens in
all sorts of political contexts around
the world we saw it and brexit as well
and so I’ve been digging into this to
try to understand it a little bit better
and so I went back to the year 2000 in
the United States and tried to
understand at what point people flipped
over from voting from Republican to
Democrat and it turns out that if you
organize all of the counties in the
country by their population density and
then sort and group those you can see a
clear crossover point and it turns out
to be about a thousand people per square
mile
is when people flip their behavior from
voting for a right-wing candidate to a
left-wing candidate now we can see this
again reflected in 2004 you can see the
gray compared to 2000 and it sort of
gets more extreme as we go over time but
the crossover point still remains about
the same about a thousand people per
square mile so you can see that going
all the way up into 2012 and 2016 and in
2016 indeed the division is greater in
that in that year in last year’s
election than it was in any previous
year so I started to wonder well what
might be driving
some of this so I did another kind of
visualization graph or what I do is took
data from OpenStreetMap and looked at
how much road mileage is necessary per
person to support a specific population
density and it turns out that when you
get down to about a thousand people per
square mile you end up kind of getting a
flattening and the amount of road
infrastructure that’s required and that
of course also applies to sewers and
other types of infrastructure it’s kind
of the same as the physics of the you
know circulatory system of an elephant
versus a mouse that sort of thing it’s a
similar kind of curve so you know that
that sort of explains its some extent
why we might be attenuated around this
urban-rural divide and the idea that
like people in cities are probably
interested in different things in terms
of their tax dollar spending you know
things like public transportation or
housing or health care versus people in
remote areas who might be more
interested in roads and gasoline and
cars
so I’ve also over the last few years
been looking at other ways that we sort
and divide ourselves this is a
visualization of my hometown Baltimore
Maryland in the United States you may
know it from the wire it’s our little
bit of marketing that we do for the city
but what this is is a visualization of
all the twitter users in the city that i
could find at a specific point and each
dot is a twitter user and each line is a
relationship between them a friend
follow relationship this is not a
geographic map just a visualization of
twitter users but what we find is that
people sort themselves very neatly into
kind of categories that are quite
clearly identifiable so on the right
where it says geeks and TEDx that’s kind
of where I’m at I’m a geek and I do TEDx
so I’m kind of in the second e’en geeks
there and then on the far left is mostly
younger mostly african-american kids who
are into hip-hop and stuff like that and
what’s kind of startling about this map
is it really shows the racial
segregation that exists in the city
people on the far left have almost no
idea that the people on the far I even
exist when I’ve shown this map to people
on the city they kind of scratch their
head and go wow this reshapes my mental
model of how I thought the city worked
so when you see this kind of segregation
you can also kind of spot the same
pattern in guess which city this is this
is
Louis Missouri in the United States and
if you recall you know we had a lot of
violence in both st. Louis first in st.
Louis and then in Baltimore and I
predicted that we would probably see
some violence in Baltimore were there to
be a catalyzing incident and indeed we
have the same pattern there we’ve got
kind of startup geeky people on the
right TEDx and then on the far left it’s
mostly younger black folks listening to
rap music and there’s very little
overlap between these worlds and indeed
you can see similar patterns here in
Norway but you guys are a little bit
more cohesive than saying mr. Baltimore
you can see groups that clearly identify
around football and sports maybe not
surprisingly a lot of places innovation
tech and education the Norway diaspora
and Europe there’s sort of Norwegians
all scattered throughout Europe that are
plugged into this and you can see them
clustered there arts and culture
politics and governance activism and
commentary a lot of writers here in
Norway I found and then kind of
mainstream media in the centre so the
point of kind of showing you these
divisions is like you know we as humans
we have there’s a principle called
homophily which is basically the birds
of a feather kind of flock together and
we group into little subgroups kind of
naturally but what if people started to
take advantage of those divisions and
maybe even try to make them worse let’s
take a look at this
it’s a battle for social media supremacy
that could define this year’s US
presidential election Donald Trump’s
campaign used a london-based data
analytics company to help it by Dong
strategy since June a very sophisticated
data analysis operation had been
underway this involves a British data
science firm in Cambridge analytical
communication is fundamentally changing
but if you’re gonna spend a billion
dollars you want to be really ruthless
about making sure that you’re focusing
on the right people daily we’re updating
our modelling with this data to show an
accurate picture where the electorate
stands that day today in the United
States we have somewhere close to four
or five thousand data points on every
individual a query rated our polling and
allowed us to see that a lot of these
Midwestern Rust Belt states were one to
three points closer than what most
people thought
[Music]
how did you know what the voter was this
year was not going to be what happened
in 2012 clearly other groups weren’t as
dynamic as you the ability to sort of
shift the way you do what you did I
think what we do differently is the
combination of pulling a data science
but these sorts of new technologies
couple of social media incredibly
powerful you did something that a lot of
people who get paid a lot of money to do
were not able to see and do which is you
got this right
Matt Donald Trump yet he should be
kissing you today yeah so Cambridge
analytic oh I love this quote there are
no longer any experts except Cambridge
gentlemen ago they were Trump’s digital
team and figured out how to win I mean
this is from their website this is
propaganda that they put out and I use
that word for a specific purpose because
this is an actual propaganda firm you
may not know that Cambridge analytic is
a subsidiary of a firm called strategic
communication labs or SCL on that
company has a 20-year background in
running Psychological operations or
psyops in military contexts around the
world including in Afghanistan and other
recent Wars so let’s take a little look
at what Cambridge analytically is doing
in order to try to tip elections both in
their US and brexit and in Kenya and in
other places where they’ve been involved
so what they do is they try to derive an
ocean score this is a personality score
for every adult that they can sort of
cram into their database and ocean is
another another word for is the Big Five
personality score and it stands for
openness concierge conscientiousness
extraversion agreeableness and
neuroticism and depending on how each
individual kind of score is on each of
these axes you can create targeted
advertising that will modify your
behavior now that’s an important point
we’re not talking about necessarily
changing your opinion about what
candidate to vote for we’re talking
about modifying your behavior and that
might include getting you to argue with
people it might include getting you not
to vote and a whole variety of other
potential behaviors so if we take a look
at some other materials that SCO has put
out over time this is a slide that was
included in a
talking about work that they were doing
in Afghanistan and sort of complicated
mumbo-jumbo on here but the key thing to
kind of recognize with this is this one
element here taa target audience
analysis and what they’re doing there is
basically saying there’s a slice of the
population that we would like to get to
behave in a certain way how do we send
them messaging how do we give them
information that’s going to get them to
behave the way that we want them to
behave okay so this is the technology
that’s being deployed in elections in
open democracies so this is another kind
of a dashboard from an SEL piece that
shows how they’re addressing able to
address with very fine kind of
granularity different aspects of a young
unmarried males population in
Afghanistan and of course that would be
an important population to be able to
kind of manipulate in a war context so
how did they take this and apply it to
kind of elections and everyday stuff
that we do in our democracies what
they’re doing is using what’s called
Facebook dark posts now you’ve all seen
Facebook advertising where you know
you’ll see an ad for a product or
something dark posts are a little bit
different they’re an advertisement but
it’s sent specifically to a very tiny
targeted micro audience and no one else
can see those except you and the
advertiser so as political speech it’s
completely unregulated no one knows what
even messages have been sent out and
indeed yesterday if you saw the news
facebook under testimony to Congress
admitted that they had received over a
hundred thousand dollars in payments
from Russian based companies to
influence the US election that doesn’t
even include the sort of work that
Cambridge analytic acclaims to have been
doing that’s just Russian companies
trying to influence Americans so what
does this look like in practice well it
allows you to create different kinds of
messaging that you can send to different
people depending on you know sort of
what their inclinations are so for
somebody that’s driven by fear you can
say Islamic radicals are killing the
innocent or maybe somebody that’s going
to respond to tradition and patriotism
will respond to patriotic symbols well
somebody else might be more convinced by
a sort of a science logic reason based
argument so it doesn’t work well erin
banks one of the chief funders of brexit
made claims that they created a one
online followers that reached upwards of
25 millions on most week the engagement
level was huge and our FB page still
grows the 3k a week the key thing here
is that all it took was six hundred and
fifty thousand votes to sway the result
in the brexit election and that’s zero
point nine seven percent of the UK
population likewise in the u.s. Trump
lost by almost three million votes if
you just looked at the popular vote but
by targeting specific counties in
specific states he was able to get the
70 thousand vote margin that was
required in order to win the electoral
college by 306 votes to 232 votes so
I’ve been using some of this same
methodology to try to study the news and
to get a sense of what’s going on with
this whole Trump Russia situation and
SEL and Cambridge analytic and
everything else and it’s it’s definitely
overwhelming and I’m not going to go
into the details of this map here
because you can’t really read it but you
know the idea here is that we can
actually study what the relationships
between all of these different entities
and get some sense that kind of where
things are going so on the left side of
this graph is actually a lot of
transactions involving Azerbaijani and
Kazakhstan e financial transactions that
Trump’s been engaged in and it’s very
likely that that’s where the
investigation is gonna go next and it’s
very likely that he may be taken down by
this so a lot of people ask me you know
what’s going to happen and I think that
at this stage you know we’ve got a
tremendous conflict that’s been set up
clearly you know there’s a lot of forces
that are sort of moving in on Trump and
he’s feeling threatened and you know
whether you agree with right-wing or
left-wing politics isn’t the point we
just sort of have an attack on the
system right now and whenever a conflict
like this is resolved there’s a good
chance that it could be dangerous and
messy so I think we need to be able to
expect that in addition to that I think
we should probably expect to see kind of
a continued assault on consensus reality
the things that we all believe to be
true will be tempted to fight with each
other will be prompted to fight with
each other and we should resist those
temptations wherever we can and indeed
you know we’ve already seen divisions in
the right and left with the emergence of
neo-nazi groups on the right and then
increasing arguments on the left about
whether it’s appropriate to use violence
to fight such neo-nazi groups in
addition to the other question that
people ask me a lot is what should we be
doing about this and you know I think
clearly we need to protect our
vulnerable democracies they are
extremely vulnerable to attack as a
technology and we need to be aware of
that but a lot of the solutions that are
being proposed really amount to kind of
blaming the victim and saying like you
know you need to be better educated you
need to have browser plug-ins that are
lurk people to spurious information and
while that might be a reasonable
approach it’s kind of the same as saying
that our democracies are being attacked
because they’re wearing a short skirt
and that’s not OK we need to deal with
the aggressors in this case and that
includes Russia it includes SCL and
Cambridge analytical and also the Mercer
family that’s been funding Cambridge
analytic equai tab it so you know I
think we really have to you know start
to focus on how do we attack the people
who are would be attacking our
democracies you know during peacetime
and some people have even asked me like
you know are we gonna be at war soon and
I actually think we’re kind of at war
now and you know whether it’s a new kind
of cool war or a continuation of the
Cold War I can’t say but I do believe
that this war will be resolved the way
that other wars have been in the past
openness will assert itself over
nationalism and fear and indeed it must
because that is our one great prayer and
true hope for a prosperous shared future
thank you
[Applause]
so if you’ve had the feeling over the
last several months that society is
being torn apart I’d like to offer you a
simple explanation it actually is now
you know we as humans divide ourselves
into all sorts of interesting categories
naturally one way that we do that is by
organizing ourselves into cities and
this actually is a visualization of the
GDP of the United States split in half
and what you can see here is that the
most populated areas mostly on the
coasts account for 50% of the GDP of the
country while the entire rest of the
country accounts for the rest and this
is actually reflected in how the country
voted back in November you can see that
the bright blue areas there are areas
that voted for Clinton and the areas in
red where are people who voted for Trump
and it was 2623 counties that voted for
Trump and 489 for Clinton but yet
Clinton got a majority of the vote so
you know we see that this kind of rural
divided rural urban divide happens in
all sorts of political contexts around
the world we saw it and brexit as well
and so I’ve been digging into this to
try to understand it a little bit better
and so I went back to the year 2000 in
the United States and tried to
understand at what point people flipped
over from voting from Republican to
Democrat and it turns out that if you
organize all of the counties in the
country by their population density and
then sort and group those you can see a
clear crossover point and it turns out
to be about a thousand people per square
mile
is when people flip their behavior from
voting for a right-wing candidate to a
left-wing candidate now we can see this
again reflected in 2004 you can see the
gray compared to 2000 and it sort of
gets more extreme as we go over time but
the crossover point still remains about
the same about a thousand people per
square mile so you can see that going
all the way up into 2012 and 2016 and in
2016 indeed the division is greater in
that in that year in last year’s
election than it was in any previous
year so I started to wonder well what
might be driving
some of this so I did another kind of
visualization graph or what I do is took
data from OpenStreetMap and looked at
how much road mileage is necessary per
person to support a specific population
density and it turns out that when you
get down to about a thousand people per
square mile you end up kind of getting a
flattening and the amount of road
infrastructure that’s required and that
of course also applies to sewers and
other types of infrastructure it’s kind
of the same as the physics of the you
know circulatory system of an elephant
versus a mouse that sort of thing it’s a
similar kind of curve so you know that
that sort of explains its some extent
why we might be attenuated around this
urban-rural divide and the idea that
like people in cities are probably
interested in different things in terms
of their tax dollar spending you know
things like public transportation or
housing or health care versus people in
remote areas who might be more
interested in roads and gasoline and
cars
so I’ve also over the last few years
been looking at other ways that we sort
and divide ourselves this is a
visualization of my hometown Baltimore
Maryland in the United States you may
know it from the wire it’s our little
bit of marketing that we do for the city
but what this is is a visualization of
all the twitter users in the city that i
could find at a specific point and each
dot is a twitter user and each line is a
relationship between them a friend
follow relationship this is not a
geographic map just a visualization of
twitter users but what we find is that
people sort themselves very neatly into
kind of categories that are quite
clearly identifiable so on the right
where it says geeks and TEDx that’s kind
of where I’m at I’m a geek and I do TEDx
so I’m kind of in the second e’en geeks
there and then on the far left is mostly
younger mostly african-american kids who
are into hip-hop and stuff like that and
what’s kind of startling about this map
is it really shows the racial
segregation that exists in the city
people on the far left have almost no
idea that the people on the far I even
exist when I’ve shown this map to people
on the city they kind of scratch their
head and go wow this reshapes my mental
model of how I thought the city worked
so when you see this kind of segregation
you can also kind of spot the same
pattern in guess which city this is this
is
Louis Missouri in the United States and
if you recall you know we had a lot of
violence in both st. Louis first in st.
Louis and then in Baltimore and I
predicted that we would probably see
some violence in Baltimore were there to
be a catalyzing incident and indeed we
have the same pattern there we’ve got
kind of startup geeky people on the
right TEDx and then on the far left it’s
mostly younger black folks listening to
rap music and there’s very little
overlap between these worlds and indeed
you can see similar patterns here in
Norway but you guys are a little bit
more cohesive than saying mr. Baltimore
you can see groups that clearly identify
around football and sports maybe not
surprisingly a lot of places innovation
tech and education the Norway diaspora
and Europe there’s sort of Norwegians
all scattered throughout Europe that are
plugged into this and you can see them
clustered there arts and culture
politics and governance activism and
commentary a lot of writers here in
Norway I found and then kind of
mainstream media in the centre so the
point of kind of showing you these
divisions is like you know we as humans
we have there’s a principle called
homophily which is basically the birds
of a feather kind of flock together and
we group into little subgroups kind of
naturally but what if people started to
take advantage of those divisions and
maybe even try to make them worse let’s
take a look at this
it’s a battle for social media supremacy
that could define this year’s US
presidential election Donald Trump’s
campaign used a london-based data
analytics company to help it by Dong
strategy since June a very sophisticated
data analysis operation had been
underway this involves a British data
science firm in Cambridge analytical
communication is fundamentally changing
but if you’re gonna spend a billion
dollars you want to be really ruthless
about making sure that you’re focusing
on the right people daily we’re updating
our modelling with this data to show an
accurate picture where the electorate
stands that day today in the United
States we have somewhere close to four
or five thousand data points on every
individual a query rated our polling and
allowed us to see that a lot of these
Midwestern Rust Belt states were one to
three points closer than what most
people thought
[Music]
how did you know what the voter was this
year was not going to be what happened
in 2012 clearly other groups weren’t as
dynamic as you the ability to sort of
shift the way you do what you did I
think what we do differently is the
combination of pulling a data science
but these sorts of new technologies
couple of social media incredibly
powerful you did something that a lot of
people who get paid a lot of money to do
were not able to see and do which is you
got this right
Matt Donald Trump yet he should be
kissing you today yeah so Cambridge
analytic oh I love this quote there are
no longer any experts except Cambridge
gentlemen ago they were Trump’s digital
team and figured out how to win I mean
this is from their website this is
propaganda that they put out and I use
that word for a specific purpose because
this is an actual propaganda firm you
may not know that Cambridge analytic is
a subsidiary of a firm called strategic
communication labs or SCL on that
company has a 20-year background in
running Psychological operations or
psyops in military contexts around the
world including in Afghanistan and other
recent Wars so let’s take a little look
at what Cambridge analytically is doing
in order to try to tip elections both in
their US and brexit and in Kenya and in
other places where they’ve been involved
so what they do is they try to derive an
ocean score this is a personality score
for every adult that they can sort of
cram into their database and ocean is
another another word for is the Big Five
personality score and it stands for
openness concierge conscientiousness
extraversion agreeableness and
neuroticism and depending on how each
individual kind of score is on each of
these axes you can create targeted
advertising that will modify your
behavior now that’s an important point
we’re not talking about necessarily
changing your opinion about what
candidate to vote for we’re talking
about modifying your behavior and that
might include getting you to argue with
people it might include getting you not
to vote and a whole variety of other
potential behaviors so if we take a look
at some other materials that SCO has put
out over time this is a slide that was
included in a
talking about work that they were doing
in Afghanistan and sort of complicated
mumbo-jumbo on here but the key thing to
kind of recognize with this is this one
element here taa target audience
analysis and what they’re doing there is
basically saying there’s a slice of the
population that we would like to get to
behave in a certain way how do we send
them messaging how do we give them
information that’s going to get them to
behave the way that we want them to
behave okay so this is the technology
that’s being deployed in elections in
open democracies so this is another kind
of a dashboard from an SEL piece that
shows how they’re addressing able to
address with very fine kind of
granularity different aspects of a young
unmarried males population in
Afghanistan and of course that would be
an important population to be able to
kind of manipulate in a war context so
how did they take this and apply it to
kind of elections and everyday stuff
that we do in our democracies what
they’re doing is using what’s called
Facebook dark posts now you’ve all seen
Facebook advertising where you know
you’ll see an ad for a product or
something dark posts are a little bit
different they’re an advertisement but
it’s sent specifically to a very tiny
targeted micro audience and no one else
can see those except you and the
advertiser so as political speech it’s
completely unregulated no one knows what
even messages have been sent out and
indeed yesterday if you saw the news
facebook under testimony to Congress
admitted that they had received over a
hundred thousand dollars in payments
from Russian based companies to
influence the US election that doesn’t
even include the sort of work that
Cambridge analytic acclaims to have been
doing that’s just Russian companies
trying to influence Americans so what
does this look like in practice well it
allows you to create different kinds of
messaging that you can send to different
people depending on you know sort of
what their inclinations are so for
somebody that’s driven by fear you can
say Islamic radicals are killing the
innocent or maybe somebody that’s going
to respond to tradition and patriotism
will respond to patriotic symbols well
somebody else might be more convinced by
a sort of a science logic reason based
argument so it doesn’t work well erin
banks one of the chief funders of brexit
made claims that they created a one
online followers that reached upwards of
25 millions on most week the engagement
level was huge and our FB page still
grows the 3k a week the key thing here
is that all it took was six hundred and
fifty thousand votes to sway the result
in the brexit election and that’s zero
point nine seven percent of the UK
population likewise in the u.s. Trump
lost by almost three million votes if
you just looked at the popular vote but
by targeting specific counties in
specific states he was able to get the
70 thousand vote margin that was
required in order to win the electoral
college by 306 votes to 232 votes so
I’ve been using some of this same
methodology to try to study the news and
to get a sense of what’s going on with
this whole Trump Russia situation and
SEL and Cambridge analytic and
everything else and it’s it’s definitely
overwhelming and I’m not going to go
into the details of this map here
because you can’t really read it but you
know the idea here is that we can
actually study what the relationships
between all of these different entities
and get some sense that kind of where
things are going so on the left side of
this graph is actually a lot of
transactions involving Azerbaijani and
Kazakhstan e financial transactions that
Trump’s been engaged in and it’s very
likely that that’s where the
investigation is gonna go next and it’s
very likely that he may be taken down by
this so a lot of people ask me you know
what’s going to happen and I think that
at this stage you know we’ve got a
tremendous conflict that’s been set up
clearly you know there’s a lot of forces
that are sort of moving in on Trump and
he’s feeling threatened and you know
whether you agree with right-wing or
left-wing politics isn’t the point we
just sort of have an attack on the
system right now and whenever a conflict
like this is resolved there’s a good
chance that it could be dangerous and
messy so I think we need to be able to
expect that in addition to that I think
we should probably expect to see kind of
a continued assault on consensus reality
the things that we all believe to be
true will be tempted to fight with each
other will be prompted to fight with
each other and we should resist those
temptations wherever we can and indeed
you know we’ve already seen divisions in
the right and left with the emergence of
neo-nazi groups on the right and then
increasing arguments on the left about
whether it’s appropriate to use violence
to fight such neo-nazi groups in
addition to the other question that
people ask me a lot is what should we be
doing about this and you know I think
clearly we need to protect our
vulnerable democracies they are
extremely vulnerable to attack as a
technology and we need to be aware of
that but a lot of the solutions that are
being proposed really amount to kind of
blaming the victim and saying like you
know you need to be better educated you
need to have browser plug-ins that are
lurk people to spurious information and
while that might be a reasonable
approach it’s kind of the same as saying
that our democracies are being attacked
because they’re wearing a short skirt
and that’s not OK we need to deal with
the aggressors in this case and that
includes Russia it includes SCL and
Cambridge analytical and also the Mercer
family that’s been funding Cambridge
analytic equai tab it so you know I
think we really have to you know start
to focus on how do we attack the people
who are would be attacking our
democracies you know during peacetime
and some people have even asked me like
you know are we gonna be at war soon and
I actually think we’re kind of at war
now and you know whether it’s a new kind
of cool war or a continuation of the
Cold War I can’t say but I do believe
that this war will be resolved the way
that other wars have been in the past
openness will assert itself over
nationalism and fear and indeed it must
because that is our one great prayer and
true hope for a prosperous shared future
thank you
[Applause]
Please follow and like us: