Press "Enter" to skip to content

“The Science of Networks” | Rowland Kao | TEDxUniversityofGlasgow


I’m here to talk to you about the

science of networks now if you’ve heard

about networks there’s a pretty good

chance of what you’ll think about is

maybe the movie the social network you

might think about Facebook you might

think about Twitter or Instagram if

you’re a little bit older you might

think about another movie and play

called six degrees of separation

when Will Smith was younger and did

funny or more interesting movies my

comment networks from another

perspective okay so I look at infectious

diseases now I did my my undergrad in my

graduate work in physics but I soon got

really interested in how you could use

some of the techniques in physics to

understand how diseases pass through

populations okay a lot of you look

pretty young to remember this but in

2001 there was a big epidemic of

something called foot-and-mouth disease

now foot mouth disease is probably the

most infectious disease we know of and

when it came into the country it would

result in over six million livestock

being slaughtered cost us over 8 billion

pounds to control farmers today are

still traumatized by what happened to

them now one of the things about the

epidemic is it was spread initially by

sheep moving all around the country ok

so that’s what all those arrows and that

map behind me are and you can ask the

question well what does sheep moving

around Britain have to do with things

like Facebook or Twitter okay the thing

is if you view a population as a series

of individuals with connections between

them and information flowing between

them then it doesn’t matter what that

information is in a way if it’s diseases

it’s a bacteria or a virus it could be

pandemic flu or a bola virus

transmitting around the world if it’s

something like Facebook or Twitter

it’s information of another sort it

could be news it could be gossip it

could be fake news now there’s a few

commonalities to these things which help

us to understand what it is about making

a disease spread from place to place in

the same way it helps us understand when

does a piece of gossip sort of go like

wildfire also that everybody knows about

it one of those things is some

we call super spreaders so there’s some

individuals who are just really really

active if you think about gossip it’s

the individual always has their ears

grunted always know what everybody’s

doing and once they know it once they

know it then they tell everybody if it’s

disease over here for this system that

I’m talking about its markets because

markets bring in lots of animals and

then they spread them out again now

because they’re more important books for

transmitting information and forgetting

that information that makes them really

really important out of proportion to

their actual activity two times is

active four times as important now the

other thing is that what you can see on

this map is that there are different

areas the country which are infected and

that infection in different parts of the

country are cars bug the jumps by

individual groups of animals and these

are what we call small world links okay

and they refer to the that what often

happens when you talk to somebody a

party say oh isn’t it a small world we

both know the same person now later on

you’re going to be introduced to Steve

Backshall he’s not here at the moment

I’ve never met him however it just so

happens I used to live in a village

called Claire north of Glasgow some

friends who kids go to school with my

kids or news – Nigel Pope and Jackie

savoury happened to bring us wildlife

documentaries including the highlands

documentary which some of you wildlife

documentary which some of you would have

seen and they know Steve and so they

represent in a social sense those small

world links now what do those networks

look like that I actually deal with so

what you see here is what looks like a

big mess what it is is a bunch of farms

it’s all the farms that own sheep in

Scotland and all the movements of sheep

between them in one month June of 2014

and what that represents is a population

of over two thousand in individual farms

which over 3,000 individual events among

them so this information is recorded

from day to day now to a person like me

who loves looking at data because I sort

of almost voyeuristic interesting and

seeing what people do this is like candy

okay you can look at this for ages just

to try to figure out what people are

doing and the really big datasets that

we have

things like moving of capital around

Britain we know where every single cow

is for every single day of its life over

its entire lifespan so that’s 25 million

animals over roughly speaking 50 years

we’re talking terabytes of data and all

this information produces what appears

to be enormous complexity now I’m going

to thin this Knepper out a bit by saying

okay well let’s not take a look let’s

take a week and what you should be able

to see by comparing the two things is

that you got broadly those same

properties which I showed you before

with foot-and-mouth stuff so you get

these these groups these clusters of

nodes so those are the big groups of

circles those represents farms moving

things in and out of markets and then

you’ve got these things that join them

up those are those small world links and

so what you can see is the pattern of

the disease as it transmits in 2001 is

replicated in the pattern of movements

and that pattern is consistent so

underlying the complexity there are

often simple rules and to try to

convince you that those rules apply to

more than just sheep in Britain this is

my rather pathetic Twitter network these

are my Twitter followers all 50 out of

them they’re grouped as a network is

people who I follow and people who

follow me there actually are more of

them but a lot of them are junk where

people just sort of well anyway there’s

there’s a lot of spam out there in the

Twitter world they’re grouped in four

groove but you can still see that same

pattern the groups which represent

different interest areas that I have are

all connected by bridging follows okay

so the same kind of structure you see

saw before look at the same thing

another way so this represents a series

of tweets retweets and mentions all

around the term bovine TB and the reason

being is because one of the disease’s I

look at is bovine tuberculosis in cattle

and badgers now some of you only know

this is a very controversial disease so

there actually is quite a bit of Twitter

traffic about it so in this short time

frame you can see that same pattern

individuals who are super spreaders lots

of things going back and forth with them

and bridging between different groups

the different groups who represent

things like government officials farmers

badger welfare groups okay and so what

you should be able to see from this is

that a network is a network but what

travels over the network largely depends

on the behavior different bits of

information are interesting to different

people for different reasons and one of

the really interesting things for me is

what happens when things like behavior

and disease operate over the same

network in other words what happens when

two networks collide so I did a study

with a friend of mine Jess Enright who

works in computer science where we

simulated a population of supposed

farmers so every individual that you see

there is a farmer and there are linked

to their next-door neighbors so these

are like farms spreading disease to each

other and in this simple simulation that

we ran we had four properties and that’s

it that’s the entire simulation we have

an infection risk so these farmers can

infect each other but they can also be

infected by some external factor and

let’s imagine that they’re say Badgers

infecting the farms without the farmers

doing anything themselves the farmers

have memory so they know what happened

to them before did they get disease

yesterday did they get the disease two

days ago etc etc and they applied some

sort of value system to it

so the question is how much do I value

having a disease-free farm and the last

thing they have is on the basis of that

combination of memory and value is they

make choices and the choices do I do

something to stop disease or do I not

and so what this graph represents is the

outcome of some of those simulations so

if you look across the bottom axis

glowing to the right what we say is well

what happens as you increase the amount

of risk that a farm is experiencing okay

so more and more badgers infected for

example well what you could see at the

beginning as you move from left to right

is that as that risk increases what you

would expect to happen is the amount of

disease increases but you reach a

critical point

okay what’s called a phase transition

where all of a sudden the disease risk

drops the reason being that farmers

start to change their behavior

and so you get kind of the

counterintuitive result that increasing

risk actually results in decreasing

disease now that’s a really good thing

right people reacting to circumstance

doing things for the good of everybody

and everybody benefits

but the really curious thing about it is

this now in that last simulation that I

was mentioning everybody more or less

have the same value system okay they

value Disease Control about the same

amount some randomness in it but we’re

pretty much the same now what this says

if we now look at the red line if only

one percent of individuals one percent

don’t care about disease at all what

happens is there is actually no benefit

at all to people being hearing about

biosecurity okay so one percent of

individuals can make all the difference

between controlling a disease and not

now the thing to keep in mind about this

is that these individuals actually don’t

disease spread very far they can only

spread it to their next-door neighbors

okay so that my friend just would put it

what they’re responsible for is not so

much the spreading of disease but what

they’re responsible is for the spreading

of despair okay farmers deciding that

just isn’t worth it to do some kind of

action because there it does no good now

that that contrast okay between

individuals doing what they very

rationally perceive as being the right

thing and an outcome that is worse for

everybody is actually something that’s

known not from social networks so much

but from something called game theory

and you may have heard from about game

theory in for example the movie A

Beautiful Mind which is about John Nash

who was one of the originators of the

whole field what the networks add to it

is they said that a small number of

influential individuals can change

everything even though most people don’t

even know that they exist now that’s an

example of where networks and small

number of individuals can cause

something bad to happen what I’m going

to argue about now is to say that

sometimes those few individuals can do

something good

now when I talk about difficult friends

I’m pretty sure most of you know what I

mean effectively they are that person

who you least want to see at your door

at the end of a long day either at uni

or at work asking to go up for a drink

okay but you’d go out anyway because

they’re your friends and those friends

could be your classmates those friends

could be people you’ll work with

it could be friends from your childhood

they could be your parents okay this is

simulated my daughter actually is pretty

good about having things anyway and the

thing is in today’s world where a lot of

our interactions are through social

media okay where we do things virtually

it is very easy to unfriend somebody to

make sure that you only follow the

people on Twitter you agree with to only

read the newspapers online that say the

things that you like to hear and that is

true whether you read breitbart news or

or listen to Fox News or a Guardian

reader it’s true about everybody we can

do that same thing about isolating

ourselves from individuals who do not

think like us and the reason why those

individuals are important I’m going to

try to show over here now this is my

Twitter network again now remember I

said these four groups are broadly

speaking groups of people who tend to

follow each other all right and broadly

speaking as well they refer to different

disciplines that are involved in the

work I do so I work in infectious

diseases population policy so I need to

talk to people or evolutionary

biologists statisticians mathematicians

people do networks veterinarians etc etc

etc but scientists often do not speak

the same language they speak different

languages mathematicians in particular

there were some I met upstairs are

completely different from everybody else

okay and unless you make the effort to

break through those language barriers

then you lose an awful lot in the

science especially in these areas where

all of them are important now the thing

about these links is because there are a

few a number they’re actually very

fragile lose them and again you lose

the communication the cooperation in the

network and I’m going to show you

another example or this has actually

been really really important okay so now

this graph over here this represents

American Senate voting patterns from the

time of dwight d eisenhower being

present which is in the early 50s

through de barackobama so there’s no

data yet on donald trump and i don’t

want to know what that looks like

so these US Senate voting patterns

represent the proportion of all votes

per year where Democrats although one

way and Republicans all vote the other

and so what you can see is all the way

pretty much until the first George Bush

that proportion of these polarized votes

actually stays fairly constant obviously

his fluctuations from the year to year

but we’re talking about a time when all

sorts of major incredibly controversial

events occurred we’re talking about

civil rights movements in the 60s we’re

talking about the Vietnam War we’re

talking about the oil crisis we’re

talking about all sorts of things which

should have been divisive and yet

because the American parties believe

that consensus and the good of the

countries is more important in the party

they tended to agree as much as they

could that has all changed over the last

few years and in fact by the time

Obama’s president the polarization is

absolutely incredible

now I haven’t looked at these data

myself but other people have and the

pattern is being driven by a relatively

few number of senators who are willing

to cross the floor okay they’re not many

of them there never have been but they

were always there but they have

virtually disappeared over the last

eight years now the net result of that

is an American government that became

largely dysfunctional over the last few

years the disillusionment of the

American people with the traditional

directions of power and the rise of

populist like Donald Trump now the thing

about Donald Trump is he’s actually not

that special I mean there have been

people like him all the time there

always will be Donald Trump’s in the

world okay what has changed is our

attitudes all right now some of you may

have heard of something

the Doomsday Clock okay the Doomsday

Clock was started in 1947 by a bunch of

nuclear scientists we wanted to have

some sort of graphical representation of

how closely were to an apocalypse

initially nuclear but now including

things like climate change for example

and just this past January they decided

that we were actually now closer to some

major disaster than we have been since

the 1950s now when things are bad okay

when when we have tough times this is a

time when we most want to retreat into

our comfort zone it’s a time when it’s

very easy to blame others Outsiders for

what our problems are and that’s true in

the American election but we certainly

cannot point fingers here can we with

results after the bread’s vote for

example so I remember people blaming

older people for why we were in this

problem saying why are they allowed to

vote because they’re not going to live

very long okay I mean this goes

completely against the principles of

democracy it’s precisely these times

where we need to reach out where we need

to reach out to those difficult friends

people who don’t necessarily agree with

not because they’re going to change

their minds but so we can understand the

way they think

now I’m gonna end this talk a slightly

different way from I originally intended

to I just found out last night that my

sister my older sister is in hospital

and she’s not doing very well and so

this is sort of a tribute to her she’s

one of those difficult people she does

not agree with me in any way politically

personally the way we raise our kids etc

etc we don’t talk that much but I know

she cares and I know that she always

tries to understand my point of view she

almost never succeeds okay and I almost

never succeed with her but the truth of

the matter is sometimes we do and in

that course of doing that I think I’ve

actually become a much better person so

the reason why I like networks the

reason I enjoy studying them because I

like the way they tell us about how the

disease is transmitted but also I like

the way they tell us about ourselves

they tell us about how we interact not

just as individuals there’s populations

how we interact as a group

and I hope you think so a little bit

more now yourself thank you very much [Applause]

Please follow and like us: