Return to Video

The moral bias behind your search results

  • 0:01 - 0:04
    So whenever I visit a school
    and talk to students,
  • 0:04 - 0:06
    I always ask them the same thing:
  • 0:07 - 0:08
    Why do you Google?
  • 0:09 - 0:12
    Why is Google the search engine
    of choice for you?
  • 0:13 - 0:15
    Strangely enough, I always get
    the same three answers.
  • 0:15 - 0:17
    One, "Because it works,"
  • 0:17 - 0:20
    which is a great answer;
    that's why I Google, too.
  • 0:20 - 0:22
    Two, somebody will say,
  • 0:22 - 0:25
    "I really don't know of any alternatives."
  • 0:26 - 0:29
    It's not an equally great answer
    and my reply to that is usually,
  • 0:29 - 0:31
    "Try to Google the word 'search engine,'
  • 0:31 - 0:33
    you may find a couple
    of interesting alternatives."
  • 0:33 - 0:35
    And last but not least, thirdly,
  • 0:35 - 0:39
    inevitably, one student will raise
    her or his hand and say,
  • 0:39 - 0:44
    "With Google, I'm certain to always get
    the best, unbiased search result."
  • 0:45 - 0:52
    Certain to always get the best,
    unbiased search result.
  • 0:53 - 0:55
    Now, as a man of the humanities,
  • 0:56 - 0:58
    albeit a digital humanities man,
  • 0:58 - 0:59
    that just makes my skin curl,
  • 0:59 - 1:04
    even if I, too, realize that that trust,
    that idea of the unbiased search result
  • 1:04 - 1:08
    is a cornerstone in our collective love
    for and appreciation of Google.
  • 1:09 - 1:13
    I will show you why that, philosophically,
    is almost an impossibility.
  • 1:13 - 1:16
    But let me first elaborate,
    just a little bit, on a basic principle
  • 1:16 - 1:19
    behind each search query
    that we sometimes seem to forget.
  • 1:20 - 1:22
    So whenever you set out
    to Google something,
  • 1:22 - 1:26
    start by asking yourself this:
    "Am I looking for an isolated fact?"
  • 1:26 - 1:29
    What is the capital of France?
  • 1:30 - 1:32
    What are the building blocks
    of a water molecule?
  • 1:32 - 1:34
    Great -- Google away.
  • 1:34 - 1:37
    There's not a group of scientists
    who are this close to proving
  • 1:37 - 1:39
    that it's actually London and H30.
  • 1:39 - 1:42
    You don't see a big conspiracy
    among those things.
  • 1:42 - 1:43
    We agree, on a global scale,
  • 1:43 - 1:46
    what the answers are
    to these isolated facts.
  • 1:46 - 1:52
    But if you complicate your question
    just a little bit and ask something like,
  • 1:52 - 1:54
    "Why is there
    an Israeli-Palestine conflict?"
  • 1:55 - 1:58
    You're not exactly looking
    for a singular fact anymore,
  • 1:58 - 1:59
    you're looking for knowledge,
  • 1:59 - 2:02
    which is something way more
    complicated and delicate.
  • 2:03 - 2:04
    And to get to knowledge,
  • 2:04 - 2:07
    you have to bring 10 or 20
    or 100 facts to the table
  • 2:07 - 2:10
    and acknowledge them and say,
    "Yes, these are all true."
  • 2:10 - 2:12
    But because of who I am,
  • 2:12 - 2:14
    young or old, black or white,
    gay or straight,
  • 2:14 - 2:16
    I will value them differently.
  • 2:16 - 2:18
    And I will say, "Yes, this is true,
  • 2:18 - 2:20
    but this is more important
    to me than that."
  • 2:20 - 2:22
    And this is where it becomes interesting,
  • 2:22 - 2:24
    because this is where we become human.
  • 2:24 - 2:27
    This is when we start
    to argue, to form society.
  • 2:27 - 2:30
    And to really get somewhere,
    we need to filter all our facts here --
  • 2:30 - 2:33
    through friends and neighbors
    and parents and children
  • 2:33 - 2:35
    and coworkers and newspapers
    and magazines,
  • 2:35 - 2:38
    to finally be grounded in real knowledge,
  • 2:38 - 2:42
    which is something that a search engine
    is a poor help to achieve.
  • 2:43 - 2:50
    So, I promised you an example
    just to show you why it's so hard
  • 2:50 - 2:53
    to get to the point of true, clean,
    objective knowledge --
  • 2:53 - 2:55
    as food for thought.
  • 2:55 - 2:58
    I will conduct a couple of simple
    queries, search queries.
  • 2:58 - 3:03
    We'll start with "Michelle Obama,"
  • 3:03 - 3:04
    the First Lady of the United States.
  • 3:04 - 3:06
    And we'll click for pictures.
  • 3:07 - 3:09
    It works really well, as you can see.
  • 3:09 - 3:12
    It's a perfect search
    result, more or less.
  • 3:12 - 3:15
    It's just her in the picture,
    not even the President.
  • 3:16 - 3:17
    How does this work?
  • 3:18 - 3:19
    Quite simple.
  • 3:19 - 3:22
    Google uses a lot of smartness
    to achieve this, but quite simply,
  • 3:22 - 3:25
    they look at two things
    more than anything.
  • 3:25 - 3:30
    First, what does it say in the caption
    under the picture on each website?
  • 3:30 - 3:32
    Does it say "Michelle Obama"
    under the picture?
  • 3:32 - 3:34
    Pretty good indication
    it's actually her on there.
  • 3:34 - 3:37
    Second, Google looks at the picture file,
  • 3:37 - 3:40
    the name of the file as such
    uploaded to the website.
  • 3:40 - 3:42
    Again, is it called "MichelleObama.jpeg"?
  • 3:43 - 3:46
    Pretty good indication it's not
    Clint Eastwood in the picture.
  • 3:46 - 3:50
    So, you've got those two and you get
    a search result like this -- almost.
  • 3:50 - 3:57
    Now, in 2009, Michelle Obama
    was the victim of a racist campaign,
  • 3:57 - 4:01
    where people set out to insult her
    through her search results.
  • 4:01 - 4:04
    There was a picture distributed
    widely over the Internet
  • 4:04 - 4:07
    where her face was distorted
    to look like a monkey.
  • 4:07 - 4:10
    And that picture was published all over.
  • 4:10 - 4:14
    And people published it
    very, very purposefully,
  • 4:14 - 4:16
    to get it up there in the search results.
  • 4:16 - 4:19
    They made sure to write
    "Michelle Obama" in the caption
  • 4:19 - 4:23
    and they made sure to upload the picture
    as "MichelleObama.jpeg" or the like.
  • 4:23 - 4:25
    You get why -- to manipulate
    the search result.
  • 4:25 - 4:27
    And it worked, too.
  • 4:27 - 4:29
    So when you picture-Googled
    for "Michelle Obama" in 2009,
  • 4:29 - 4:33
    that distorted monkey picture
    showed up among the first results.
  • 4:33 - 4:36
    Now, the results are self-cleansing,
  • 4:36 - 4:38
    and that's sort of the beauty of it,
  • 4:38 - 4:42
    because Google measures relevance
    every hour every day.
  • 4:42 - 4:44
    However, Google didn't settle
    for that this time,
  • 4:44 - 4:47
    they just thought, "That's racist
    and it's a bad search result
  • 4:48 - 4:51
    and we're going to go back
    and clean that up manually.
  • 4:51 - 4:54
    We are going to write
    some code and fix it,"
  • 4:54 - 4:55
    which they did.
  • 4:55 - 4:59
    And I don't think anyone in this room
    thinks that was a bad idea.
  • 5:00 - 5:01
    Me, neither.
  • 5:03 - 5:06
    But then, a couple of years go by,
  • 5:06 - 5:09
    and the world's most-Googled Anders,
  • 5:09 - 5:11
    Anders Behring Breivik,
  • 5:11 - 5:13
    did what he did.
  • 5:13 - 5:15
    This is July 22 in 2011,
  • 5:15 - 5:18
    and a terrible day in Norwegian history.
  • 5:18 - 5:21
    This man, a terrorist, blew up
    a couple of government buildings
  • 5:21 - 5:24
    walking distance from where we are
    right now in Oslo, Norway
  • 5:24 - 5:26
    and then he traveled
    to the island of Utøya
  • 5:26 - 5:29
    and shot and killed a group of kids.
  • 5:29 - 5:31
    Almost 80 people died that day.
  • 5:32 - 5:37
    And a lot of people would describe
    this act of terror as two steps,
  • 5:37 - 5:40
    that he did two things: he blew up
    the buildings and he shot those kids.
  • 5:40 - 5:42
    It's not true.
  • 5:42 - 5:44
    It was three steps.
  • 5:44 - 5:47
    He blew up those buildings,
    he shot those kids,
  • 5:47 - 5:50
    and he sat down and waited
    for the world to Google him.
  • 5:51 - 5:54
    And he prepared
    all three steps equally well.
  • 5:55 - 5:57
    And if there was somebody
    who immediately understood this,
  • 5:57 - 5:59
    it was a Swedish web developer,
  • 5:59 - 6:03
    a search engine optimization expert
    in Stockholm, named Nikke Lindqvist.
  • 6:03 - 6:04
    He's also a very political guy
  • 6:04 - 6:07
    and he was right out there
    in social media, on his blog and Facebook.
  • 6:07 - 6:09
    And he told everybody,
  • 6:09 - 6:11
    "If there's something that
    this guy wants right now,
  • 6:11 - 6:14
    it's to control the image of himself.
  • 6:15 - 6:17
    Let's see if we can distort that.
  • 6:17 - 6:21
    Let's see if we, in the civilized world,
    can protest against what he did
  • 6:21 - 6:25
    through insulting him
    in his search results."
  • 6:25 - 6:26
    And how?
  • 6:27 - 6:29
    He told all of his readers the following,
  • 6:29 - 6:31
    "Go out there on the Internet,
  • 6:31 - 6:34
    find pictures of dog poop on sidewalks --
  • 6:35 - 6:37
    find pictures of dog poop on sidewalks --
  • 6:37 - 6:40
    publish them in your feeds,
    on your websites, on your blogs.
  • 6:40 - 6:43
    Make sure to write the terrorist's
    name in the caption,
  • 6:43 - 6:48
    make sure to name
    the picture file "Breivik.jpeg."
  • 6:48 - 6:52
    Let's teach Google that that's
    the face of the terrorist."
  • 6:54 - 6:55
    And it worked.
  • 6:56 - 6:59
    Two years after that campaign
    against Michelle Obama,
  • 6:59 - 7:02
    this manipulation campaign
    against Anders Behring Breivik worked.
  • 7:02 - 7:07
    If you picture-Googled for him weeks after
    the July 22 events from Sweden,
  • 7:07 - 7:11
    you'd see that picture of dog poop
    high up in the search results,
  • 7:11 - 7:12
    as a little protest.
  • 7:13 - 7:18
    Strangely enough, Google
    didn't intervene this time.
  • 7:18 - 7:23
    They did not step in and manually
    clean those search results up.
  • 7:24 - 7:26
    So the million-dollar question is:
  • 7:26 - 7:29
    is there anything different
    between these two happenings here?
  • 7:29 - 7:32
    Is there anything different between
    what happened to Michelle Obama
  • 7:32 - 7:34
    and what happened
    to Anders Behring Breivik?
  • 7:34 - 7:36
    Of course not.
  • 7:37 - 7:38
    It's the exact same thing,
  • 7:38 - 7:41
    yet Google intervened in one case
    and not in the other.
  • 7:41 - 7:42
    Why?
  • 7:43 - 7:47
    Because Michelle Obama
    is an honorable person, that's why,
  • 7:47 - 7:50
    and Anders Behring Breivik
    is a despicable person.
  • 7:50 - 7:52
    See what happens there?
  • 7:52 - 7:55
    An evaluation of a person takes place
  • 7:55 - 7:59
    and there's only one
    power-player in the world
  • 7:59 - 8:01
    with the authority to say who's who.
  • 8:02 - 8:04
    "We like you, we dislike you.
  • 8:04 - 8:06
    We believe in you,
    we don't believe in you.
  • 8:06 - 8:08
    You're right, you're wrong.
    You're true, you're false.
  • 8:08 - 8:10
    You're Obama, and you're Breivik."
  • 8:11 - 8:13
    That's power if I ever saw it.
  • 8:15 - 8:19
    So I'm asking you to remember
    that behind every algorithm
  • 8:19 - 8:21
    is always a person,
  • 8:21 - 8:23
    a person with a set of personal beliefs
  • 8:23 - 8:26
    that no code can ever
    completely eradicate.
  • 8:26 - 8:28
    And my message goes
    out not only to Google,
  • 8:28 - 8:31
    but to all believers in the faith
    of code around the world.
  • 8:31 - 8:34
    You need to identify
    your own personal bias.
  • 8:34 - 8:36
    You need to understand that you are human
  • 8:36 - 8:39
    and take responsibility accordingly.
  • 8:40 - 8:43
    And I say this because I believe
    we've reached a point in time
  • 8:43 - 8:44
    when it's absolutely imperative
  • 8:44 - 8:48
    that we tie those bonds
    together again, tighter:
  • 8:48 - 8:50
    the humanities and the technology.
  • 8:50 - 8:52
    Tighter than ever.
  • 8:52 - 8:56
    And, if nothing else, to remind us
    that that wonderfully seductive idea
  • 8:56 - 8:58
    of the unbiased, clean search result
  • 8:58 - 9:01
    is, and is likely to remain, a myth.
  • 9:02 - 9:03
    Thank you for your time.
  • 9:03 - 9:06
    (Applause)
Title:
The moral bias behind your search results
Speaker:
Andreas Ekstrøm
Description:

more » « less
Video Language:
English
Team:
closed TED
Project:
TEDTalks
Duration:
09:18

English subtitles

Revisions Compare revisions