I’ve been always interested in politics.
During my high school years, I was an activist at the student center.
Among other things, we organized huge camps,
of more than 600 students,
there wasn’t a single adult present,
so you can imagine how they were.
We came from years of a dictatorship, and we thought that
what was most important was to rebuild the community,
to open ourselves to other ways of relating to diverse people
and to take fear out of our bodies,
which is quite different to traditional politics.
One thing we learned is that this is very hard,
that we’re naturally not very well prepared
to deal with conflict, difference —
Thirty years later, I’m working at Facebook
and once again I’m in a camp,
but this time it’s a global camp of 2 billion people.
But it’s a sphere of interaction that could bring us together,
and it could be an incredible driving force to generate empathy.
However, the feeling is that we’re becoming more and more polarized.
Each country has their local version of polarization,
and many think that the Internet, and social networks in particular,
are one of its causes.
But is it like that?
Do social networks shut us out and push us away
from those who think differently from us?
Or can social networks be a meeting space
like the one we imagined in high school?
The advantage we have today is that we can try to understand this
by using the huge amount of data we generate online every day.
Just so you have an idea: every day users post on Facebook
an amount of text equivalent to seven Wikipedias.
Seven Wikipedias are generated on Facebook every day.
And with the photos that are uploaded in a week,
if we printed them, maybe, many of you have probably never seen a printed photo,
but we used to print them at some point in time —
With the photos of a week we could cover the City of Buenos Aires.
And I’m really privileged
to be immersed in this avalanche of data,
in the Computational Social Science group
at Facebook in Menlo Park, California.
And yes, you heard right, Social Sciences can be computational.
You need both to understand
what I call the “photo of our culture,”
we take every day on the Internet.
It’s worth noticing we do all of this
over anonymous data, okay?
It’s based on statistics about thousands or millions of cases
and they don’t tell us anything about one case in particular,
and they don’t predict anything about an individual person.
In order to give you sense of it, I’ll tell you something we did:
we looked at what happens to couple relationships on Facebook.
We started by counting the amount of interactions between two people
that are about to form a couple,
from 100 days before it’s formed to 100 days afterwards.
As you can see…
(Laughter)
the interactions keep increasing – boom!
The couple is formed, then interactions plunge,
we guess they’re spending more time face to face, right?
(Laughter)
In parallel, by doing text analysis,
we can see the words they use in these interaction
are every time more positive.
Something else we looked at is
how likely it is for a couple to last 5 years
based on how many months it’s been going on for.
For example, if the couple already lasted 6 months,
the probability of lasting 5 years goes up 40 percent
compared to the chances it had on its first day.
But obviously, couples break up,
some of them, and on the day we announce we’re newly single
our friends come to our help, or maybe jump at the chance —
our interactions online with our friends
on that day we announce our new relationship status, tripled.
I’m not sure why.
(Laughter)
Anyway, today I want to talk to you less about love,
and more about a lack of love.
Let me tell you a bit about what data tells us
about polarization in social networks,
and see if this global camp
we imagined at the high school is possible.
I’ll start with the good news.
On one hand, social networks make us closer,
they make the world a progressively smaller place.
Today, a Facebook user in Argentina
has in average more than 500 friends,
follows 250 pages,
going from celebrities to football clubs,
to politicians or mass media,
and is part of more than 30 groups
regarding all sort of topics.
Communication turned into many-to-many,
potentially so-many-to-so many,
and I say potentially everyone-to-everyone.
I’ll give you a specific number of how small the world has become.
In 1909, Guglielmo Marconi asked himself the following question:
if I want to call anyone in the world
using this “wireless telegraph,”
which he had just invented — a precursor of the radio —
but I can only call an acquaintance,
ask him to call an acquaintance of his,
and so on until I reach the person I want to call,
how many jumps will I have to go through?
Marconi obviously didn’t have the tools
to determine this though, just because,
Marconi guessed a number: 6.
And he planted this idea, that later on became very popular,
the “six degrees of separation,”
which you might’ve heard of.
And then movies came out about it and all.
There were some later attempts to determine this number,
but all very limited in scope.
Our team did the biggest-scale study to this date,
using the cloud, Facebook’s network of friendships.
But well, first we verified one thing:
that 99.9 percent of users, almost all of them,
are interconnected in the same friendship network,
in the same cloud, and not separated in different little clouds.
But of course, this thing we wanted to do involved calculating
2 billion by 2 billion distances.
It’s four billions of billions,
a 4 followed by 18 zeros, of distances,
and then you have to measure averages.
We said “we’re Facebook,”
we put it running on some hundreds of computers,
we waited for a week,
and after a week, we had that number computed for a thousand people.
A thousand, with no zeros behind.
Luckily, in computer science we have this thing called “approximation methods,”
and using these methods we managed to reach a number:
[3.5 degrees of separation]
This means that between you and Messi,
you and the Pope, you and a monk in the Tibet,
on average there’s a friend, a friend of a friend,
a friend of a friend of a friend,
and half a person more.
(Laughter)
(Applause)
It’s small. (Applause)
We’re all very close, very interconnected, almost cramped.
We could be swimming in a sea of empathy.
But, well, we tend to gather with similar people, right?
From the same neighborhood, same social circle,
that share our same interests,
political ideas, or even musical taste.
But here networks can also help us.
In a study we did in the US
among people who declared themselves as right-wing or left-wing,
we saw that, in average,
20 percent of each person’s friends are of the opposite ideology.
This different people tend to be people we don’t see every day,
with whom we don’t interact that often.
Much before the Internet, sociologists called these people “weak bonds”
and what they saw is that, in a close circle of people, of ideas,
these weak bonds are what give us diversity.
And in social networks we stumble upon them every day.
But well, we are feeling that something isn’t helping us.
So what’s going on?
I identify at least three things
we should pay attention to:
algorithms, the quality of the news,
and our attitude towards those who think differently.
Firstly, all of our Internet experiences
are ruled by algorithms, a series of rules and criteria
to give us what we want.
When we open almost any social network,
the content is presented to us organized, personalized.
The algorithm tries to show us first
what it thinks is most interesting, relevant, and attractive to us.
And what many people are scared of, including myself,
is that the algorithm hides diversity,
that it starts to give us more
of what we already agree with and think
and hides us the rest.
To see if that’s true, as the algorithm is very complex,
we studied particular cases.
Going back to the US, to the left- and right- wing friends I told you about,
we saw that in average, for a right-wing person,
his left-wing friends produce 35 percent
of the articles in their network,
the articles he could potentially see.
And what we saw is that the algorithm
was showing almost all of it, 34 percent.
And the person clicked a more or less proportionally frequent amount of times.
So, in this article the team published in the Science journal,
what we saw is that, for starters,
the algorithm isn’t hiding diversity.
But well, it’s a very complex topic,
and it’s an ever-evolving topic,
so it’s something we need to be paying attention to
for a long time, and at all times.
The second issue, the second problem, is that there’s too much trash content
in the Internet,
from fake news to sensationalist headlines,
trying to catch our attention,
which we often call “clickbait”.
For example, this fake news about the Pope supporting Trump
was shared millions of times.
And our cognitive bias is to let them pass by,
smile a bit, or even worse,
press the share button on bad-quality content
that reinforces our views.
However, we are in fact outraged when it’s something that represents opposing ideas.
And this is a serious problem, because all the bad-quality content,
these fake news, tend to reinforce our opinion,
to lock us even more inside ourselves,
and they contaminate the network.
This leads us to the third problem,
which is our attitude towards those who think differently.
Even when a friend shares something that is real,
true, fact-checked, non-sensationalist,
but that goes against our ideas,
we tend to dismiss it or even block it.
In the weeks around Colombian peace referendum,
a very polarizing event,
the amount of unfriendings almost doubled
the same amount in, for example, the weeks around Christmas.
In short, we’re very close,
but it seems like we don’t make the most of the diversity of social networks.
There’s a risk of generating a vicious circle
of feedback between technology and our behavior:
we tend to reject or hide what we don’t agree with,
and we share and look at what reinforces our ideas.
As tech companies, it’s clear
we have a responsibility and a role to play
in order to reduce this polarization.
In fact, a couple of months ago Facebook changed its mission,
from simply connecting the world, to bringing it closer together.
There’s the idea, for example, of encouraging the creation of more groups
that generate a sense of community, give them more visibility,
and help the administrators of those groups to manage them.
Last year, by means of some changes
we got 50 percent more people engaged in at least one group
that they see as a meaningful group,
that awakes in them a sense of community.
Another line of action is to partner
with media, universities,
and fact-checking organizations,
as Chequeado in Argentina,
or Snopes, POLITICO, or Politifact, in the US
to give people more control
over what they consume and reduce misinformation.
Last year, we did an experiment
in the United States and Europe
where news came with a small icon saying they had been disputed,
by any of these organizations,
and a link to see the relevant discussion.
Another line of action is trying to offer related news
that give us diverse perspectives over the same topic.
And eventually, to detect those sensationalist headers
and rank them lower.
I think a key question is
what are we, the users, going to do
to break this vicious circle
and this feedback between technology and behavior.
We have to understand the role of the customization algorithms,
we have to learn how to filter,
but above all to take responsibility
for what we do to ourselves
and what we do to our network,
because every action we take in social networks
influences others.
Each trash website that gets likes becomes more prestigious.
We have to start by some basic steps.
If you find an article, check the source,
especially, much, much more especially,
if it’s something that reinforces your ideas.
If you find fact-checked and well-developed content, share it.
Now, if it’s one of those memes that make fun of the opposing side,
those who agree with you will laugh
and the rest will be offended,
let’s spare ourselves of this, let’s not share it, and put a stop to it.
As you can see, I’m not giving up on believing it’s still possible
to make a camp of 2 billion people.
But we’re at a turning point.
Tech companies on one side
and we, the users, on the other side,
have to work out the way
for this feedback between technology and behavior
to become a virtuous circle,
that generates communities of empathy
instead of information bubbles,
and that brings us closer instead of bringing us apart
and polarize us more and more.
Thank you very much.
(Applause)
”
Be First to Comment