[ted id=1091]
Mark Zuckerberg, a journalist was asking him a question about the news feed. And the journalist was asking him, “Why is this so important?” And Zuckerberg said, “A squirrel dying in your front yard may be more relevant to your interests right now than people dying in Africa.” And I want to talk about what a Web based on that idea of relevance might look like.
So when I was growing up in a really rural area in Maine, the Internet meant something very different to me. It meant a connection to the world. It meant something that would connect us all together. And I was sure that it was going to be great for democracy and for our society. But there’s this shift in how information is flowing online, and it’s invisible. And if we don’t pay attention to it, it could be a real problem. So I first noticed this in a place I spend a lot of time — my Facebook page. I’m progressive, politically — big surprise — but I’ve always gone out of my way to meet conservatives. I like hearing what they’re thinking about; I like seeing what they link to; I like learning a thing or two. And so I was surprised when I noticed one day that the conservatives had disappeared from my Facebook feed. And what it turned out was going on was that Facebook was looking at which links I clicked on, and it was noticing that, actually, I was clicking more on my liberal friends’ links than on my conservative friends’ links. And without consulting me about it, it had edited them out. They disappeared.
So Facebook isn’t the only place that’s doing this kind of invisible, algorithmic editing of the Web. Google’s doing it too. If I search for something, and you search for something, even right now at the very same time, we may get very different search results. Even if you’re logged out, one engineer told me, there are 57 signals that Google looks at — everything from what kind of computer you’re on to what kind of browser you’re using to where you’re located — that it uses to personally tailor your query results. Think about it for a second: there is no standard Google anymore. And you know, the funny thing about this is that it’s hard to see. You can’t see how different your search results are from anyone else’s.
But a couple of weeks ago, I asked a bunch of friends to Google “Egypt” and to send me screen shots of what they got. So here’s my friend Scott’s screen shot. And here’s my friend Daniel’s screen shot. When you put them side-by-side, you don’t even have to read the links to see how different these two pages are. But when you do read the links, it’s really quite remarkable. Daniel didn’t get anything about the protests in Egypt at all in his first page of Google results. Scott’s results were full of them. And this was the big story of the day at that time. That’s how different these results are becoming.
So it’s not just Google and Facebook either. This is something that’s sweeping the Web. There are a whole host of companies that are doing this kind of personalization. Yahoo News, the biggest news site on the Internet, is now personalized — different people get different things. Huffington Post, the Washington Post, the New York Times — all flirting with personalization in various ways. And this moves us very quickly toward a world in which the Internet is showing us what it thinks we want to see, but not necessarily what we need to see. As Eric Schmidt said, “It will be very hard for people to watch or consume something that has not in some sense been tailored for them.”
So I do think this is a problem. And I think, if you take all of these filters together, you take all these algorithms, you get what I call a filter bubble. And your filter bubble is your own personal unique universe of information that you live in online. And what’s in your filter bubble depends on who you are, and it depends on what you do. But the thing is that you don’t decide what gets in. And more importantly, you don’t actually see what gets edited out. So one of the problems with the filter bubble was discovered by some researchers at Netflix. And they were looking at the Netflix queues, and they noticed something kind of funny that a lot of us probably have noticed, which is there are some movies that just sort of zip right up and out to our houses. They enter the queue, they just zip right out. So “Iron Man” zips right out, and “Waiting for Superman” can wait for a really long time.
What they discovered was that in our Netflix queues there’s this epic struggle going on between our future aspirational selves and our more impulsive present selves. You know we all want to be someone who has watched “Rashomon,” but right now we want to watch “Ace Ventura” for the fourth time. (Laughter) So the best editing gives us a bit of both. It gives us a little bit of Justin Bieber and a little bit of Afghanistan. It gives us some information vegetables, it gives us some information dessert. And the challenge with these kinds of algorithmic filters, these personalized filters, is that, because they’re mainly looking at what you click on first, it can throw off that balance. And instead of a balanced information diet, you can end up surrounded by information junk food.
What this suggests is actually that we may have the story about the Internet wrong. In a broadcast society — this is how the founding mythology goes — in a broadcast society, there were these gatekeepers, the editors, and they controlled the flows of information. And along came the Internet and it swept them out of the way, and it allowed all of us to connect together, and it was awesome. But that’s not actually what’s happening right now. What we’re seeing is more of a passing of the torch from human gatekeepers to algorithmic ones. And the thing is that the algorithms don’t yet have the kind of embedded ethics that the editors did. So if algorithms are going to curate the world for us, if they’re going to decide what we get to see and what we don’t get to see, then we need to make sure that they’re not just keyed to relevance. We need to make sure that they also show us things that are uncomfortable or challenging or important — this is what TED does — other points of view.
And the thing is we’ve actually been here before as a society. In 1915, it’s not like newspapers were sweating a lot about their civic responsibilities. Then people noticed that they were doing something really important. That, in fact, you couldn’t have a functioning democracy if citizens didn’t get a good flow of information. That the newspapers were critical, because they were acting as the filter, and then journalistic ethics developed. It wasn’t perfect, but it got us through the last century. And so now, we’re kind of back in 1915 on the Web. And we need the new gatekeepers to encode that kind of responsibility into the code that they’re writing.
I know that there are a lot of people here from Facebook and from Google — Larry and Sergey — people who have helped build the Web as it is, and I’m grateful for that. But we really need you to make sure that these algorithms have encoded in them a sense of the public life, a sense of civic responsibility. We need you to make sure that they’re transparent enough that we can see what the rules are that determine what gets through our filters. And we need you to give us some control, so that we can decide what gets through and what doesn’t. Because I think we really need the Internet to be that thing that we all dreamed of it being. We need it to connect us all together. We need it to introduce us to new ideas and new people and different perspectives. And it’s not going to do that if it leaves us all isolated in a Web of one.
So when I was growing up in a really rural area in Maine, the Internet meant something very different to me. It meant a connection to the world. It meant something that would connect us all together. And I was sure that it was going to be great for democracy and for our society. But there’s this shift in how information is flowing online, and it’s invisible. And if we don’t pay attention to it, it could be a real problem. So I first noticed this in a place I spend a lot of time — my Facebook page. I’m progressive, politically — big surprise — but I’ve always gone out of my way to meet conservatives. I like hearing what they’re thinking about; I like seeing what they link to; I like learning a thing or two. And so I was surprised when I noticed one day that the conservatives had disappeared from my Facebook feed. And what it turned out was going on was that Facebook was looking at which links I clicked on, and it was noticing that, actually, I was clicking more on my liberal friends’ links than on my conservative friends’ links. And without consulting me about it, it had edited them out. They disappeared.
So Facebook isn’t the only place that’s doing this kind of invisible, algorithmic editing of the Web. Google’s doing it too. If I search for something, and you search for something, even right now at the very same time, we may get very different search results. Even if you’re logged out, one engineer told me, there are 57 signals that Google looks at — everything from what kind of computer you’re on to what kind of browser you’re using to where you’re located — that it uses to personally tailor your query results. Think about it for a second: there is no standard Google anymore. And you know, the funny thing about this is that it’s hard to see. You can’t see how different your search results are from anyone else’s.
But a couple of weeks ago, I asked a bunch of friends to Google “Egypt” and to send me screen shots of what they got. So here’s my friend Scott’s screen shot. And here’s my friend Daniel’s screen shot. When you put them side-by-side, you don’t even have to read the links to see how different these two pages are. But when you do read the links, it’s really quite remarkable. Daniel didn’t get anything about the protests in Egypt at all in his first page of Google results. Scott’s results were full of them. And this was the big story of the day at that time. That’s how different these results are becoming.
So it’s not just Google and Facebook either. This is something that’s sweeping the Web. There are a whole host of companies that are doing this kind of personalization. Yahoo News, the biggest news site on the Internet, is now personalized — different people get different things. Huffington Post, the Washington Post, the New York Times — all flirting with personalization in various ways. And this moves us very quickly toward a world in which the Internet is showing us what it thinks we want to see, but not necessarily what we need to see. As Eric Schmidt said, “It will be very hard for people to watch or consume something that has not in some sense been tailored for them.”
So I do think this is a problem. And I think, if you take all of these filters together, you take all these algorithms, you get what I call a filter bubble. And your filter bubble is your own personal unique universe of information that you live in online. And what’s in your filter bubble depends on who you are, and it depends on what you do. But the thing is that you don’t decide what gets in. And more importantly, you don’t actually see what gets edited out. So one of the problems with the filter bubble was discovered by some researchers at Netflix. And they were looking at the Netflix queues, and they noticed something kind of funny that a lot of us probably have noticed, which is there are some movies that just sort of zip right up and out to our houses. They enter the queue, they just zip right out. So “Iron Man” zips right out, and “Waiting for Superman” can wait for a really long time.
What they discovered was that in our Netflix queues there’s this epic struggle going on between our future aspirational selves and our more impulsive present selves. You know we all want to be someone who has watched “Rashomon,” but right now we want to watch “Ace Ventura” for the fourth time. (Laughter) So the best editing gives us a bit of both. It gives us a little bit of Justin Bieber and a little bit of Afghanistan. It gives us some information vegetables, it gives us some information dessert. And the challenge with these kinds of algorithmic filters, these personalized filters, is that, because they’re mainly looking at what you click on first, it can throw off that balance. And instead of a balanced information diet, you can end up surrounded by information junk food.
What this suggests is actually that we may have the story about the Internet wrong. In a broadcast society — this is how the founding mythology goes — in a broadcast society, there were these gatekeepers, the editors, and they controlled the flows of information. And along came the Internet and it swept them out of the way, and it allowed all of us to connect together, and it was awesome. But that’s not actually what’s happening right now. What we’re seeing is more of a passing of the torch from human gatekeepers to algorithmic ones. And the thing is that the algorithms don’t yet have the kind of embedded ethics that the editors did. So if algorithms are going to curate the world for us, if they’re going to decide what we get to see and what we don’t get to see, then we need to make sure that they’re not just keyed to relevance. We need to make sure that they also show us things that are uncomfortable or challenging or important — this is what TED does — other points of view.
And the thing is we’ve actually been here before as a society. In 1915, it’s not like newspapers were sweating a lot about their civic responsibilities. Then people noticed that they were doing something really important. That, in fact, you couldn’t have a functioning democracy if citizens didn’t get a good flow of information. That the newspapers were critical, because they were acting as the filter, and then journalistic ethics developed. It wasn’t perfect, but it got us through the last century. And so now, we’re kind of back in 1915 on the Web. And we need the new gatekeepers to encode that kind of responsibility into the code that they’re writing.
I know that there are a lot of people here from Facebook and from Google — Larry and Sergey — people who have helped build the Web as it is, and I’m grateful for that. But we really need you to make sure that these algorithms have encoded in them a sense of the public life, a sense of civic responsibility. We need you to make sure that they’re transparent enough that we can see what the rules are that determine what gets through our filters. And we need you to give us some control, so that we can decide what gets through and what doesn’t. Because I think we really need the Internet to be that thing that we all dreamed of it being. We need it to connect us all together. We need it to introduce us to new ideas and new people and different perspectives. And it’s not going to do that if it leaves us all isolated in a Web of one.
— source ted.com
[youtube https://www.youtube.com/watch?v=6_sim_Wc3mY&rel=0]
Eli Pariser
anyone who believes in indefinite growth
on a physical finite panic is either mad
or an economist we don’t want to focus
politics on a notion that involves the
rejection of principles around which the
large majority of our fellow citizens
live we are not as endlessly manipulable
and its predictable as you would think I
thought that the internet was going to
be this new connective medium but more
recently I’ve been wondering whether
that’s actually the case I’ve been
concerned that the Internet may not be
as connective as we thought and the
first place that I began to have an
inkling that something weird was going
on was one morning when I logged on to
my Facebook page I had just on this
morning gone to the trouble of seeking
out friends who disagreed with me I
really wanted to get outside of my own
sort of narrow parochial political point
of view I wanted to be hearing what
people who had more conservative views
were posting and see what they were
liking and be talking to them on on
Facebook and on this particular morning
I logged on and they weren’t there and
it was strange it was like Facebook was
editing them out it was like they had
disappeared and as a matter of fact it
was a Facebook was looking at which
links I was clicking on it was looking
at who I was having conversations with
and what I was liking and it had noticed
that I was clicking more on the links
that I agreed with than on the links
that I disagreed with and as a result it
had edited them out and they disappeared
Facebook at this point is kind of a big
deal it’s a medium that one in 11 human
beings now uses an increasing proportion
of people go there to find their news 50
percent of the traffic driven to many
news websites now comes from Facebook
and if Facebook is subtly distorting
what we see and what we don’t see that’s
going to make a big consequence for the
six
700 million people who use it on a daily
basis and so I began to wonder you know
sort of why why is Facebook doing this
and as I began to learn more about it
why is this technology creeping all
across the web and I think the best
place to start in answering that
question is a statistic that Eric
Schmidt from Google likes to talk about
he says if you start at the beginning of
human history and you were to record
every conversation every book every
piece of art and you were to track all
of that data it would be about five
exabytes of data that’s about 8 million
80 gigabyte iPod touches and the problem
says Eric Schmidt is that that same
amount of data poured online in the last
two days so there’s this immense torrent
of data that we have to deal with it’s
growing every day it comes from our cell
phones it comes from our email it comes
from our status updates and a number of
companies have realized that if you can
provide a way to sort through that data
and pull out the little bits of gold and
this sort of in the torrent you can
actually make some good money and so
they began to investigate well how do
you do that what’s the best mechanism
for for figuring out what actually
matters to people and the technique that
they came up with at its most simple
we’re all familiar with now from Amazon
and from Netflix it’s if you like this
you’ll like that and if you like this
you’ll like that works in two basic ways
one is that it looks for people like you
people who seem to have some of the same
preferences than that you do and uses
that to project out what your other
preferences might be and it looks at
sort of groups of products or movies
that group together you know how do
these you know which which movies if if
someone likes this movie they’ll also
tend to like that movie the interesting
thing about this is you know to do it
you have to create you have to collect
on a norm
of data the numbers are just kind of
staggering the top 50 websites now each
put on average of 67 cookies and
tracking pieces of code on your computer
every time that you visit and that’s so
that they can collect this massive
amount of data that’s needed to crunch
this but the interesting thing is you
don’t necessarily need data on the topic
you’re trying to represent in order to
make a guess about what someone might
like you can actually start to infer
from seemingly unrelated data who
someone is and so for example if you
know that someone prefers milk with
their meal over wine then you know that
they’re much more likely to be
politically conservative and if you know
that they rarely frequent fast food
outfits you know that they’re much more
likely to be liberal in this immense
amount of data we begin to reveal things
that we never thought that we were
revealing we begin to be able to draw
relationships between pieces of data
that seem not to have much to do with
each other what’s interesting about this
is you don’t need that much data in
order to begin to be able to make these
extrapolations and well five data points
seem pretty crude it’s not a very good
approximation of who you are nonetheless
with just five data points the company
hunch says that they can predict your
consumer preferences about 80% of the
time correctly in fact even if they
don’t have your five data points if they
know that you’re friends with two people
and you have their five data points then
you can still predict their preferences
correctly in other words it doesn’t even
necessarily matter how much data you
give away about yourself your friends
may be doing that work for you and the
company that has actually sort of gone
the farthest with this kind of
statistical technique is Google the way
that Google works as this idea
PageRank PageRank is sort of getting the
web to vote on which page is the most
authoritative response to a query but
PageRank is an increasingly small part
of the Google algorithm increasingly
Google uses this personal data to give
each of us a different view of Google
search results in fact as of December
for 2009 there is no standard Google
anymore we each get different results
based on what we’ve queried before based
on what we’ve clicked on so I actually
asked a bunch of friends to Google Egypt
and send me screenshots of what they got
here’s my friend Scott’s screenshot and
here’s my friend Daniels screenshot
they’re both white men they’re both
around the same age they have similar
political views they both live in New
York and yet when you put these two
search results side-by-side you don’t
even have to read the links to see how
different these two views of the world
are but when you do put them
side-by-side the difference is really
quite striking Scott got all sorts of
information about the crisis in Egypt
about the protests there about what was
going on politically but Daniel didn’t
in fact know we’re on the first page of
results was there anything about the
political upheaval in Egypt he just got
information about going to see the
pyramids
so we’re drifting apart online we’re
seeing different pictures of the world
and the reason is that there’s sort of
this obsession right now in Silicon
Valley with this idea of relevance
relevance is sort of the name for the
solution to that information overload
and when they talk about relevance they
mean something very particular because
these are companies that generally live
or die on the basis of ad sales and ad
sales happen when people click more and
so relevance means the things that you
click the most increasingly
we see different world’s online
increasingly the internet is showing us
what it thinks we want to see not
necessarily the world as it is
increasingly we’re sort of surrounded by
this membrane of personalised filters
and the thing is that this sort of
creates a filter bubble it creates a
personal universe of unique information
that makes it through all of these
filters now you don’t necessarily choose
what gets into your filter bubble these
algorithms are doing the work for you
and therefore importantly you don’t
really know what is outside of it you
don’t really know what you’re missing so
I want to talk about three problems with
living in a in the filter bubble world
the first is something I call the
distortion problem it’s that when you
don’t know what the editorial viewpoint
is through which you’re viewing the
world it’s very hard to see how
distorted your view is for the
scientists in the room it’s kind of like
you can’t extrapolate from a small
sample what the full set looks like and
as opposed to traditional media where
when you turn on a politically slanted
news broadcast or you pick up a magazine
you kind of have a sense of on what
basis information is being edited in and
edited out here you don’t really have a
sense you don’t know who Facebook or
Google thinks you are you don’t know on
what basis they’re deciding to show you
things and therefore you don’t know what
you might be missing it’s an unknown
unknown stuff that’s important but not
necessarily sort of sexy or highly
clickable drops out of you this is the
like button and the like button is the
primary way that people transmit
information across Facebook it’s very
heavily weighted in the algorithm that
Facebook uses to decide what to show you
and what not to show you and the like
button has a very particular valence
it’s not a neutral word like it’s a
positive word and so it’s easy to click
like on I ran a marathon but it’s hard
to click like on a woman about to be
stoned to death in Iran and so certain
kinds of information can travel very
quickly on Facebook and other kinds of
information drop out of sight entirely
the second problem is what the
anthropologist Dana Boyd called the
psychological equivalent of obesity and
the people who sort of did some of the
great research in this area were looking
at the movie rental website Netflix they
were looking at why certain movies move
faster through the Netflix queue than
others they noticed that movies like
Iron Man just sort of zip right through
the queue they get added to the list of
movies you want to watch and almost
immediately there at your house you’re
watching them and other movies like
Waiting for Superman this sort of
education documentary can just spend a
long time dawdling around waiting to get
out to your home
so what’s going on here there are two
very distinct clusters of movies that
emerge you have your want movies at the
short end the movie is where you come
home after a long day you just want to
turn something on and be entertained and
you can probably guess what the
categories of the movies are that are at
the other end of the spectrum right this
is documentary films holocaust movies
and almost all of French cinema and what
they realized was that there was sort of
an interesting internal tug-of-war here
going on between two selves between a
sort of more compulsive more immediate
present self and a more aspirational
future self what the filter bubble tends
to do is to sort of push that balance in
one direction the traditional media give
us a balance you get some Justin Bieber
and some Afghanistan you get some
information vegetables and some
information dessert but because the
filter bubble is just focused on what
are you going to click most what are you
going to click next there are very
different dynamics at play and instead
of getting a balanced information diet
you can find yourself surrounded by
information junk food in fact you can
live in a world in which information
junk food appears to be the only thing
that exists and the final issue is a
matter of control because that’s really
what’s at stake here that when these
algorithms are cranked all the way up
they limit what we believe to be the
options what we believe to be the
different points of view and thus they
limit what we choose they limit our
sense of freedom when I’m googling my
dentist phone number I’m trying to
remember what the capital of Switzerland
is it makes sense to have one result but
it doesn’t make sense when you’re
googling Obama or climate change then
you want a bigger view of the world so
what this all suggested to me is that I
had my idea of the internet a little bit
wrong you know I had sort of bought this
bismuth ology that said that in the you
know sort of 20th century Society
information flows were guarded by
keepers these were the editors and the
producers and because there was a
limited ability to get information out
of the out to the public they got to
choose what people saw and didn’t see
and that was a problem because they were
often elites they often had their own
parochial interests it meant that the
public only saw what they wanted them to
see and so it was a great thing when the
internet swept them out of the way all
of a sudden everyone could talk to
everyone there was this great
decentralization and a sort of new
democratized society emerged but that’s
not really what’s going on it’s not the
case that we live in a gatekeeper ‘less
society now there’s a new set of
gatekeepers and they’re not people their
code and the thing is that the code does
many of the same things that those
twentieth-century gatekeepers did it
decides what we see and we don’t see it
makes priority decisions about which
pieces of information are important but
it doesn’t have the sense of embedded
ethics that had their best the 20th
century gatekeepers did it’s worth
noting that we’ve sort of been here
before it’s not like newspapers in the
1800s had any sense of civic
responsibility either they just put on
the front page whatever would sell the
most copies but around the end of World
War one there was this shift that
occurred and people began to realize
that really you couldn’t have a
functioning democracy if you didn’t have
informed citizens that knew about what
the topics they had to make choices on
and that these newspapers were right at
the center of that equation that they
had a responsibility to make sure that
citizens had good information on which
to base their decisions and that’s where
journalistic ethics came from and I
think on the web were sort of back in
1921 and we need these new institutions
to recognize their responsibilities here
and to step up as not just commercial
institutions but really critical parts
of a functioning democracy we need them
to give us more control so that we can
decide when these filters are at work
and when they’re not we need to give
them transparent them to give us
transparency so that we can see how this
is all working what results are
personalized and on what basis and we
need to make sure that they
have these sort of ethics embedded in
maybe next to the like button on
Facebook there should be an important
button that allows people to send that
signal to their friends because I think
we really actually need the internet to
be that thing that connects us to new
ideas and to new people and to new ways
of thinking and we need it not just
because it makes life a more satisfying
endeavor but because actually I don’t
think we can solve many of the big
problems that are in front of us in any
other way you know if you think about
poverty or you think about climate
change or you think about terrorism
these are not problems that can be
solved by someone sitting in a room
somewhere they’re problems that require
solutions that bring people together
across vast differences from that from
different disciplines and different
continents thinking in different ways
and talking to each other and convincing
each other about a common solution and I
don’t think that that can happen if
we’re all stuck in a web of one you
mentioned that the code that the
computer has no ethic it doesn’t have a
sense of embedded ethics that previous
gatekeepers have whether they were in
whether they were editors or whether
they were agents or whomever but those
people who were creating the code
themselves have an ethic and I was
wondering if you could explore maybe the
that ethic and how it may have evolved
how it may have changed from the early
days of the web when we did have this
this openness and the potential to reach
out and touch lots and lots of
information to now where the code that’s
blocking us from all of that information
is about personalization about relevance
what’s changed there to be clear I’m not
saying that the code has no embedded
ethics it says that I’m saying that you
know it doesn’t have a sort of
civic-minded ethics in it I had this
argument with someone at Google where I
said so how do you think ethically about
your sort of your editorial or
curatorial responsibilities as you’re
making these personalization decisions
and he said oh no there’s no ethics here
it’s just about giving people what they
want
and I said well okay but if I’m a 9/11
conspiracy theorist and I Google nine
your job to give me the conspiracy
website that I’m most likely to click on
or is it your job to actually give me
you know the Popular Mechanics article
that debunks that which one is relevant
which one is what I really want I really
need initially I think because the
internet was sort of this fringe
phenomenon that really no one could
quite figure out what it was or what how
to make money on it there was this real
ethos of kind of openness and
connectedness that anyone you know that
is sort of a pure meritocracy in some
sense where any idea could move and and
get a massive adoption and I think
essentially what’s happened is that you
know as people have figured out how to
make really large amounts of money
online that sort of commercial mentality
has has seeped in and so when I was
arguing about this with Facebook I was
talking to an engineer there and he said
well you know let me just be blunt what
you want us to do is think about all of
these complicated and important problems
and social issues and whatever what we
want to do is sit in a room and think of
clever ways of making people spend more
minutes per day on Facebook and that’s I
think sort of one of the one of the
disconnects what’s interesting there is
this conflict this juxtaposition between
the technologically driven solutions and
the human beings who are actually
interacting online I mean ultimately
what this this notion of the the web 2.0
phenomenon the social media phenomenon
which has been represented by things
like Facebook and blogs and those types
of opinion networks the friend networks
the the social graph as some people have
described it and that’s that’s us as
social beings and so in many ways it’s
as if this notion of creating
technological solutions becomes a
depersonalization of us as as human
beings and it seems to have worked quite
well for things like Facebook and and
you know things like blogs so why should
they change well you know I do think in
order to do personalization well and to
be clear that’s sort of more what I
think needs to happen than just to sort
of roll back time and go a different way
I think you know somehow people will
want to use these filters but they’ll
want to actually
you know use these tools rather than
having the tools use them and the
question is how do you do that
when it comes to why these companies
would have an interest in this you know
I think if you look at the way that
Facebook is starting to lose interests
in some countries you know there’s this
first downtick of use in facebook since
it really started I wonder whether
that’s because it’s sort of over
optimized over personalized so that on
any given day the content appears sort
of compulsively compelling you know but
when you look back on a year of Facebook
use it’s very hard to say well what
value did I actually get out of that it
would did it add any meaning or learning
to my life actually what might be in
Facebook’s interests or Google’s
interest is in the long run to provide
sort of people with a sense not just
that they’re getting sort of what’s most
clickable but they’re actually
discovering and learning things they’re
the the real trend is data driven
development so taking as much
information as it can from you and then
creating this relevance creating this
personalization now what’s interesting
though is that as you you made the quote
or you you reference the quote from Eric
Schmidt you know Google wants to give
you what you don’t yet know what you
want he’s also talked a lot about this
notion of a serendipity engine he says
he wants Google to to produce things in
front of you that you didn’t actually
realize that you wanted so in importing
a little bit of randomness but what is
interesting about that is how he’s
thinking about that multimodal e so he’s
thinking about not not just from using
search but from using Google Docs and
from using the Android platform which is
the mobile phone technology so he’s
looking at at getting more and more
information based upon you know and even
wider breadth and and where do you see
the problems emerging from there how do
you think that’s going to affect you
know what our actual notion of
serendipity is and what the
technologically-driven notion of
serendipity is yeah so this is a hot
topic given all of this sort of
personalization is how do you you know
manufacture those moments where you feel
like you’re learning something you feel
like something new is coming up
and I wonder whether at least in the
short term that’s not sort of
fundamentally at odds with what with the
current way that they’re thinking about
a lot of these tools so Google really is
pretty focused on how do we get the top
link to be the link that you most want
to click which is just kind of a
different problem from yet which-which
and the reason they want to do that is
so that they can also manufacture that
kind of relevance in the ads right next
to it and that may be actually the
length of you click as these ads they
these two problems kind of go together
when you’re actually trying to provide
that sort of sense of discovery it’s a
different mode I was having this
conversation with someone at Google who
said oh the problem is you’re talking
about Google as an information mapping
tool and we really think of it as an
information extraction engine which is
if you have a you know sort of a blank
in your mind I’m trying to remember how
many people were at the last supper or
whatever you know you can google it and
find out and and for that
personalization works really well but
for this more exploratory thing it’s in
a way it’s just not the problem that
they’re focused on and I think it may
take someone else sort of coming up with
some clever ways of generating that and
that feeding back to Google who do you
think needs to be embedded in technology
companies who should be involved in the
development of of a technology that
allows you to outreach beyond the filter
bubble peter thiel who’s one of the big
investors in facebook and developed
PayPal and it’s kind of a fairly major
figure in Silicon Valley is that is
currently running this kind of it’s not
a contest but basically he’ll offer
people a hundred thousand dollars to
drop out of college and do a start-up
because he feels like college is a waste
of time and why not just go do your
startup and I think that’s sort of
precisely wrong like the problems with
Facebook and the way that it thinks
about privacy and the way that it thinks
about identity come directly from the
fact that it’s run by someone who has
spent a lot of time thinking about
engineering and not
a lot of time thinking about people and
so we need people who actually can bring
that more humanistic perspective to
these projects and actually you know
sort of if you’re going to do
personalization actually have a clear
idea of what a person it
you
on a physical finite panic is either mad
or an economist we don’t want to focus
politics on a notion that involves the
rejection of principles around which the
large majority of our fellow citizens
live we are not as endlessly manipulable
and its predictable as you would think I
thought that the internet was going to
be this new connective medium but more
recently I’ve been wondering whether
that’s actually the case I’ve been
concerned that the Internet may not be
as connective as we thought and the
first place that I began to have an
inkling that something weird was going
on was one morning when I logged on to
my Facebook page I had just on this
morning gone to the trouble of seeking
out friends who disagreed with me I
really wanted to get outside of my own
sort of narrow parochial political point
of view I wanted to be hearing what
people who had more conservative views
were posting and see what they were
liking and be talking to them on on
Facebook and on this particular morning
I logged on and they weren’t there and
it was strange it was like Facebook was
editing them out it was like they had
disappeared and as a matter of fact it
was a Facebook was looking at which
links I was clicking on it was looking
at who I was having conversations with
and what I was liking and it had noticed
that I was clicking more on the links
that I agreed with than on the links
that I disagreed with and as a result it
had edited them out and they disappeared
Facebook at this point is kind of a big
deal it’s a medium that one in 11 human
beings now uses an increasing proportion
of people go there to find their news 50
percent of the traffic driven to many
news websites now comes from Facebook
and if Facebook is subtly distorting
what we see and what we don’t see that’s
going to make a big consequence for the
six
700 million people who use it on a daily
basis and so I began to wonder you know
sort of why why is Facebook doing this
and as I began to learn more about it
why is this technology creeping all
across the web and I think the best
place to start in answering that
question is a statistic that Eric
Schmidt from Google likes to talk about
he says if you start at the beginning of
human history and you were to record
every conversation every book every
piece of art and you were to track all
of that data it would be about five
exabytes of data that’s about 8 million
80 gigabyte iPod touches and the problem
says Eric Schmidt is that that same
amount of data poured online in the last
two days so there’s this immense torrent
of data that we have to deal with it’s
growing every day it comes from our cell
phones it comes from our email it comes
from our status updates and a number of
companies have realized that if you can
provide a way to sort through that data
and pull out the little bits of gold and
this sort of in the torrent you can
actually make some good money and so
they began to investigate well how do
you do that what’s the best mechanism
for for figuring out what actually
matters to people and the technique that
they came up with at its most simple
we’re all familiar with now from Amazon
and from Netflix it’s if you like this
you’ll like that and if you like this
you’ll like that works in two basic ways
one is that it looks for people like you
people who seem to have some of the same
preferences than that you do and uses
that to project out what your other
preferences might be and it looks at
sort of groups of products or movies
that group together you know how do
these you know which which movies if if
someone likes this movie they’ll also
tend to like that movie the interesting
thing about this is you know to do it
you have to create you have to collect
on a norm
of data the numbers are just kind of
staggering the top 50 websites now each
put on average of 67 cookies and
tracking pieces of code on your computer
every time that you visit and that’s so
that they can collect this massive
amount of data that’s needed to crunch
this but the interesting thing is you
don’t necessarily need data on the topic
you’re trying to represent in order to
make a guess about what someone might
like you can actually start to infer
from seemingly unrelated data who
someone is and so for example if you
know that someone prefers milk with
their meal over wine then you know that
they’re much more likely to be
politically conservative and if you know
that they rarely frequent fast food
outfits you know that they’re much more
likely to be liberal in this immense
amount of data we begin to reveal things
that we never thought that we were
revealing we begin to be able to draw
relationships between pieces of data
that seem not to have much to do with
each other what’s interesting about this
is you don’t need that much data in
order to begin to be able to make these
extrapolations and well five data points
seem pretty crude it’s not a very good
approximation of who you are nonetheless
with just five data points the company
hunch says that they can predict your
consumer preferences about 80% of the
time correctly in fact even if they
don’t have your five data points if they
know that you’re friends with two people
and you have their five data points then
you can still predict their preferences
correctly in other words it doesn’t even
necessarily matter how much data you
give away about yourself your friends
may be doing that work for you and the
company that has actually sort of gone
the farthest with this kind of
statistical technique is Google the way
that Google works as this idea
PageRank PageRank is sort of getting the
web to vote on which page is the most
authoritative response to a query but
PageRank is an increasingly small part
of the Google algorithm increasingly
Google uses this personal data to give
each of us a different view of Google
search results in fact as of December
for 2009 there is no standard Google
anymore we each get different results
based on what we’ve queried before based
on what we’ve clicked on so I actually
asked a bunch of friends to Google Egypt
and send me screenshots of what they got
here’s my friend Scott’s screenshot and
here’s my friend Daniels screenshot
they’re both white men they’re both
around the same age they have similar
political views they both live in New
York and yet when you put these two
search results side-by-side you don’t
even have to read the links to see how
different these two views of the world
are but when you do put them
side-by-side the difference is really
quite striking Scott got all sorts of
information about the crisis in Egypt
about the protests there about what was
going on politically but Daniel didn’t
in fact know we’re on the first page of
results was there anything about the
political upheaval in Egypt he just got
information about going to see the
pyramids
so we’re drifting apart online we’re
seeing different pictures of the world
and the reason is that there’s sort of
this obsession right now in Silicon
Valley with this idea of relevance
relevance is sort of the name for the
solution to that information overload
and when they talk about relevance they
mean something very particular because
these are companies that generally live
or die on the basis of ad sales and ad
sales happen when people click more and
so relevance means the things that you
click the most increasingly
we see different world’s online
increasingly the internet is showing us
what it thinks we want to see not
necessarily the world as it is
increasingly we’re sort of surrounded by
this membrane of personalised filters
and the thing is that this sort of
creates a filter bubble it creates a
personal universe of unique information
that makes it through all of these
filters now you don’t necessarily choose
what gets into your filter bubble these
algorithms are doing the work for you
and therefore importantly you don’t
really know what is outside of it you
don’t really know what you’re missing so
I want to talk about three problems with
living in a in the filter bubble world
the first is something I call the
distortion problem it’s that when you
don’t know what the editorial viewpoint
is through which you’re viewing the
world it’s very hard to see how
distorted your view is for the
scientists in the room it’s kind of like
you can’t extrapolate from a small
sample what the full set looks like and
as opposed to traditional media where
when you turn on a politically slanted
news broadcast or you pick up a magazine
you kind of have a sense of on what
basis information is being edited in and
edited out here you don’t really have a
sense you don’t know who Facebook or
Google thinks you are you don’t know on
what basis they’re deciding to show you
things and therefore you don’t know what
you might be missing it’s an unknown
unknown stuff that’s important but not
necessarily sort of sexy or highly
clickable drops out of you this is the
like button and the like button is the
primary way that people transmit
information across Facebook it’s very
heavily weighted in the algorithm that
Facebook uses to decide what to show you
and what not to show you and the like
button has a very particular valence
it’s not a neutral word like it’s a
positive word and so it’s easy to click
like on I ran a marathon but it’s hard
to click like on a woman about to be
stoned to death in Iran and so certain
kinds of information can travel very
quickly on Facebook and other kinds of
information drop out of sight entirely
the second problem is what the
anthropologist Dana Boyd called the
psychological equivalent of obesity and
the people who sort of did some of the
great research in this area were looking
at the movie rental website Netflix they
were looking at why certain movies move
faster through the Netflix queue than
others they noticed that movies like
Iron Man just sort of zip right through
the queue they get added to the list of
movies you want to watch and almost
immediately there at your house you’re
watching them and other movies like
Waiting for Superman this sort of
education documentary can just spend a
long time dawdling around waiting to get
out to your home
so what’s going on here there are two
very distinct clusters of movies that
emerge you have your want movies at the
short end the movie is where you come
home after a long day you just want to
turn something on and be entertained and
you can probably guess what the
categories of the movies are that are at
the other end of the spectrum right this
is documentary films holocaust movies
and almost all of French cinema and what
they realized was that there was sort of
an interesting internal tug-of-war here
going on between two selves between a
sort of more compulsive more immediate
present self and a more aspirational
future self what the filter bubble tends
to do is to sort of push that balance in
one direction the traditional media give
us a balance you get some Justin Bieber
and some Afghanistan you get some
information vegetables and some
information dessert but because the
filter bubble is just focused on what
are you going to click most what are you
going to click next there are very
different dynamics at play and instead
of getting a balanced information diet
you can find yourself surrounded by
information junk food in fact you can
live in a world in which information
junk food appears to be the only thing
that exists and the final issue is a
matter of control because that’s really
what’s at stake here that when these
algorithms are cranked all the way up
they limit what we believe to be the
options what we believe to be the
different points of view and thus they
limit what we choose they limit our
sense of freedom when I’m googling my
dentist phone number I’m trying to
remember what the capital of Switzerland
is it makes sense to have one result but
it doesn’t make sense when you’re
googling Obama or climate change then
you want a bigger view of the world so
what this all suggested to me is that I
had my idea of the internet a little bit
wrong you know I had sort of bought this
bismuth ology that said that in the you
know sort of 20th century Society
information flows were guarded by
keepers these were the editors and the
producers and because there was a
limited ability to get information out
of the out to the public they got to
choose what people saw and didn’t see
and that was a problem because they were
often elites they often had their own
parochial interests it meant that the
public only saw what they wanted them to
see and so it was a great thing when the
internet swept them out of the way all
of a sudden everyone could talk to
everyone there was this great
decentralization and a sort of new
democratized society emerged but that’s
not really what’s going on it’s not the
case that we live in a gatekeeper ‘less
society now there’s a new set of
gatekeepers and they’re not people their
code and the thing is that the code does
many of the same things that those
twentieth-century gatekeepers did it
decides what we see and we don’t see it
makes priority decisions about which
pieces of information are important but
it doesn’t have the sense of embedded
ethics that had their best the 20th
century gatekeepers did it’s worth
noting that we’ve sort of been here
before it’s not like newspapers in the
1800s had any sense of civic
responsibility either they just put on
the front page whatever would sell the
most copies but around the end of World
War one there was this shift that
occurred and people began to realize
that really you couldn’t have a
functioning democracy if you didn’t have
informed citizens that knew about what
the topics they had to make choices on
and that these newspapers were right at
the center of that equation that they
had a responsibility to make sure that
citizens had good information on which
to base their decisions and that’s where
journalistic ethics came from and I
think on the web were sort of back in
1921 and we need these new institutions
to recognize their responsibilities here
and to step up as not just commercial
institutions but really critical parts
of a functioning democracy we need them
to give us more control so that we can
decide when these filters are at work
and when they’re not we need to give
them transparent them to give us
transparency so that we can see how this
is all working what results are
personalized and on what basis and we
need to make sure that they
have these sort of ethics embedded in
maybe next to the like button on
Facebook there should be an important
button that allows people to send that
signal to their friends because I think
we really actually need the internet to
be that thing that connects us to new
ideas and to new people and to new ways
of thinking and we need it not just
because it makes life a more satisfying
endeavor but because actually I don’t
think we can solve many of the big
problems that are in front of us in any
other way you know if you think about
poverty or you think about climate
change or you think about terrorism
these are not problems that can be
solved by someone sitting in a room
somewhere they’re problems that require
solutions that bring people together
across vast differences from that from
different disciplines and different
continents thinking in different ways
and talking to each other and convincing
each other about a common solution and I
don’t think that that can happen if
we’re all stuck in a web of one you
mentioned that the code that the
computer has no ethic it doesn’t have a
sense of embedded ethics that previous
gatekeepers have whether they were in
whether they were editors or whether
they were agents or whomever but those
people who were creating the code
themselves have an ethic and I was
wondering if you could explore maybe the
that ethic and how it may have evolved
how it may have changed from the early
days of the web when we did have this
this openness and the potential to reach
out and touch lots and lots of
information to now where the code that’s
blocking us from all of that information
is about personalization about relevance
what’s changed there to be clear I’m not
saying that the code has no embedded
ethics it says that I’m saying that you
know it doesn’t have a sort of
civic-minded ethics in it I had this
argument with someone at Google where I
said so how do you think ethically about
your sort of your editorial or
curatorial responsibilities as you’re
making these personalization decisions
and he said oh no there’s no ethics here
it’s just about giving people what they
want
and I said well okay but if I’m a 9/11
conspiracy theorist and I Google nine
your job to give me the conspiracy
website that I’m most likely to click on
or is it your job to actually give me
you know the Popular Mechanics article
that debunks that which one is relevant
which one is what I really want I really
need initially I think because the
internet was sort of this fringe
phenomenon that really no one could
quite figure out what it was or what how
to make money on it there was this real
ethos of kind of openness and
connectedness that anyone you know that
is sort of a pure meritocracy in some
sense where any idea could move and and
get a massive adoption and I think
essentially what’s happened is that you
know as people have figured out how to
make really large amounts of money
online that sort of commercial mentality
has has seeped in and so when I was
arguing about this with Facebook I was
talking to an engineer there and he said
well you know let me just be blunt what
you want us to do is think about all of
these complicated and important problems
and social issues and whatever what we
want to do is sit in a room and think of
clever ways of making people spend more
minutes per day on Facebook and that’s I
think sort of one of the one of the
disconnects what’s interesting there is
this conflict this juxtaposition between
the technologically driven solutions and
the human beings who are actually
interacting online I mean ultimately
what this this notion of the the web 2.0
phenomenon the social media phenomenon
which has been represented by things
like Facebook and blogs and those types
of opinion networks the friend networks
the the social graph as some people have
described it and that’s that’s us as
social beings and so in many ways it’s
as if this notion of creating
technological solutions becomes a
depersonalization of us as as human
beings and it seems to have worked quite
well for things like Facebook and and
you know things like blogs so why should
they change well you know I do think in
order to do personalization well and to
be clear that’s sort of more what I
think needs to happen than just to sort
of roll back time and go a different way
I think you know somehow people will
want to use these filters but they’ll
want to actually
you know use these tools rather than
having the tools use them and the
question is how do you do that
when it comes to why these companies
would have an interest in this you know
I think if you look at the way that
Facebook is starting to lose interests
in some countries you know there’s this
first downtick of use in facebook since
it really started I wonder whether
that’s because it’s sort of over
optimized over personalized so that on
any given day the content appears sort
of compulsively compelling you know but
when you look back on a year of Facebook
use it’s very hard to say well what
value did I actually get out of that it
would did it add any meaning or learning
to my life actually what might be in
Facebook’s interests or Google’s
interest is in the long run to provide
sort of people with a sense not just
that they’re getting sort of what’s most
clickable but they’re actually
discovering and learning things they’re
the the real trend is data driven
development so taking as much
information as it can from you and then
creating this relevance creating this
personalization now what’s interesting
though is that as you you made the quote
or you you reference the quote from Eric
Schmidt you know Google wants to give
you what you don’t yet know what you
want he’s also talked a lot about this
notion of a serendipity engine he says
he wants Google to to produce things in
front of you that you didn’t actually
realize that you wanted so in importing
a little bit of randomness but what is
interesting about that is how he’s
thinking about that multimodal e so he’s
thinking about not not just from using
search but from using Google Docs and
from using the Android platform which is
the mobile phone technology so he’s
looking at at getting more and more
information based upon you know and even
wider breadth and and where do you see
the problems emerging from there how do
you think that’s going to affect you
know what our actual notion of
serendipity is and what the
technologically-driven notion of
serendipity is yeah so this is a hot
topic given all of this sort of
personalization is how do you you know
manufacture those moments where you feel
like you’re learning something you feel
like something new is coming up
and I wonder whether at least in the
short term that’s not sort of
fundamentally at odds with what with the
current way that they’re thinking about
a lot of these tools so Google really is
pretty focused on how do we get the top
link to be the link that you most want
to click which is just kind of a
different problem from yet which-which
and the reason they want to do that is
so that they can also manufacture that
kind of relevance in the ads right next
to it and that may be actually the
length of you click as these ads they
these two problems kind of go together
when you’re actually trying to provide
that sort of sense of discovery it’s a
different mode I was having this
conversation with someone at Google who
said oh the problem is you’re talking
about Google as an information mapping
tool and we really think of it as an
information extraction engine which is
if you have a you know sort of a blank
in your mind I’m trying to remember how
many people were at the last supper or
whatever you know you can google it and
find out and and for that
personalization works really well but
for this more exploratory thing it’s in
a way it’s just not the problem that
they’re focused on and I think it may
take someone else sort of coming up with
some clever ways of generating that and
that feeding back to Google who do you
think needs to be embedded in technology
companies who should be involved in the
development of of a technology that
allows you to outreach beyond the filter
bubble peter thiel who’s one of the big
investors in facebook and developed
PayPal and it’s kind of a fairly major
figure in Silicon Valley is that is
currently running this kind of it’s not
a contest but basically he’ll offer
people a hundred thousand dollars to
drop out of college and do a start-up
because he feels like college is a waste
of time and why not just go do your
startup and I think that’s sort of
precisely wrong like the problems with
Facebook and the way that it thinks
about privacy and the way that it thinks
about identity come directly from the
fact that it’s run by someone who has
spent a lot of time thinking about
engineering and not
a lot of time thinking about people and
so we need people who actually can bring
that more humanistic perspective to
these projects and actually you know
sort of if you’re going to do
personalization actually have a clear
idea of what a person it
you
— source thersa.org