WEBVTT

00:00:04.440 --> 00:00:06.965
So whenever I visit a school
and talk to students,

00:00:06.965 --> 00:00:09.353
I always ask them the same thing:

00:00:10.186 --> 00:00:11.874
Why do you Google?

00:00:11.874 --> 00:00:15.477
Why is Google the search engine
of choice for you?

00:00:16.158 --> 00:00:18.771
Strangely enough, I always get
the same three answers.

00:00:18.771 --> 00:00:20.717
One, "Because it works,"

00:00:20.717 --> 00:00:23.487
which is a great answer;
that's why I Google, too.

00:00:23.487 --> 00:00:25.771
Two, somebody will say,

00:00:25.771 --> 00:00:28.533
"I really don't know of any alternatives."

00:00:28.921 --> 00:00:32.018
It's not an equally great answer
and my reply to that is usually,

00:00:32.018 --> 00:00:34.037
"Try to Google the word 'search engine,'

00:00:34.037 --> 00:00:36.660
you may find a couple
of interesting alternatives."

00:00:36.660 --> 00:00:38.635
And last but not least, thirdly,

00:00:38.635 --> 00:00:42.660
inevitably, one student will raise
her or his hand and say,

00:00:42.684 --> 00:00:47.713
"With Google, I'm certain to always get
the best, unbiased search result."

00:00:48.643 --> 00:00:54.610
Certain to always get the best,
unbiased search result.

00:00:56.550 --> 00:00:58.866
Now, as a man of the humanities,

00:00:58.866 --> 00:01:01.111
albeit a digital humanities man,

00:01:01.111 --> 00:01:02.883
that just makes my skin curl,

00:01:02.883 --> 00:01:07.773
even if I, too, realize that that trust,
that idea of the unbiased search result

00:01:07.773 --> 00:01:11.737
is a cornerstone in our collective love
for and appreciation of Google.

00:01:12.062 --> 00:01:16.118
I will show you why that, philosophically,
is almost an impossibility.

00:01:16.118 --> 00:01:19.361
But let me first elaborate,
just a little bit, on a basic principle

00:01:19.361 --> 00:01:22.758
behind each search query
that we sometimes seem to forget.

00:01:23.253 --> 00:01:25.391
So whenever you set out
to Google something,

00:01:25.391 --> 00:01:29.359
start by asking yourself this:
"Am I looking for an isolated fact?"

00:01:29.845 --> 00:01:32.525
What is the capital of France?

00:01:32.960 --> 00:01:35.288
What are the building blocks
of a water molecule?

00:01:35.288 --> 00:01:37.743
Great -- Google away.

00:01:38.051 --> 00:01:41.048
There's not a group of scientists
who are this close to proving

00:01:41.048 --> 00:01:42.928
that it's actually London and H30.

00:01:42.928 --> 00:01:45.279
You don't see a big conspiracy
among those things.

00:01:45.279 --> 00:01:46.951
We agree, on a global scale,

00:01:46.951 --> 00:01:49.621
what the answers are
to these isolated facts.

00:01:49.621 --> 00:01:54.801
But if you complicate your question
just a little bit and ask something like,

00:01:54.801 --> 00:01:57.718
"Why is there
an Israeli-Palestine conflict?"

00:01:58.461 --> 00:02:01.113
You're not exactly looking
for a singular fact anymore,

00:02:01.113 --> 00:02:02.973
you're looking for knowledge,

00:02:02.973 --> 00:02:05.775
which is something way more
complicated and delicate.

00:02:06.016 --> 00:02:07.602
And to get to knowledge,

00:02:07.602 --> 00:02:10.628
you have to bring 10 or 20
or 100 facts to the table

00:02:10.628 --> 00:02:13.476
and acknowledge them and say,
"Yes, these are all true."

00:02:13.476 --> 00:02:15.203
But because of who I am,

00:02:15.203 --> 00:02:17.687
young or old, black or white,
gay or straight,

00:02:17.687 --> 00:02:19.306
I will value them differently.

00:02:19.306 --> 00:02:21.018
And I will say, "Yes, this is true,

00:02:21.018 --> 00:02:23.119
but this is more important
to me than that."

00:02:23.119 --> 00:02:25.262
And this is where it becomes interesting,

00:02:25.262 --> 00:02:27.399
because this is where we become human.

00:02:27.399 --> 00:02:30.327
This is when we start
to argue, to form society.

00:02:30.327 --> 00:02:33.461
And to really get somewhere,
we need to filter all our facts here,

00:02:33.461 --> 00:02:36.256
through friends and neighbors
and parents and children

00:02:36.256 --> 00:02:38.471
and coworkers and newspapers
and magazines,

00:02:38.471 --> 00:02:41.381
to finally be grounded in real knowledge,

00:02:41.381 --> 00:02:45.799
which is something that a search engine
is a poor help to achieve.

00:02:46.673 --> 00:02:52.477
So, I promised you an example
just to show you why it's so hard

00:02:52.981 --> 00:02:56.588
to get to the point of true, clean,
objective knowledge --

00:02:56.588 --> 00:02:58.098
as food for thought.

00:02:58.098 --> 00:03:01.798
I will conduct a couple of simple
queries, search queries.

00:03:01.798 --> 00:03:05.487
We'll start with "Michelle Obama,"

00:03:06.332 --> 00:03:08.341
the First Lady of the United States.

00:03:08.365 --> 00:03:10.094
And we'll click for pictures.

00:03:10.643 --> 00:03:13.279
It works really well, as you can see.

00:03:13.303 --> 00:03:16.331
It's a perfect search
result, more or less.

00:03:16.355 --> 00:03:19.105
It's just her in the picture,
not even the President.

00:03:19.257 --> 00:03:20.977
How does this work?

00:03:21.317 --> 00:03:22.825
Quite simple.

00:03:22.825 --> 00:03:25.951
Google uses a lot of smartness
to achieve this, but quite simply,

00:03:25.951 --> 00:03:28.532
they look at two things
more than anything.

00:03:28.556 --> 00:03:33.712
First, what does it say in the caption
under the picture on each website?

00:03:33.736 --> 00:03:35.951
Does it say "Michelle Obama"
under the picture?

00:03:35.975 --> 00:03:38.331
Pretty good indication
it's actually her on there.

00:03:38.355 --> 00:03:40.741
Second, Google looks at the picture file,

00:03:40.765 --> 00:03:43.515
the name of the file as such
uploaded to the website.

00:03:43.515 --> 00:03:46.376
Again, is it called "MichelleObama.jpeg"?

00:03:46.376 --> 00:03:49.422
Pretty good indication it's not
Clint Eastwood in the picture.

00:03:49.422 --> 00:03:53.404
So, you've got those two and you get
a search result like this -- almost.

00:03:53.404 --> 00:03:59.457
Now, in 2009, Michelle Obama
was the victim of a racist campaign,

00:03:59.991 --> 00:04:04.317
where people set out to insult her
through her search results.

00:04:04.682 --> 00:04:07.578
There was a picture distributed
widely over the Internet

00:04:07.578 --> 00:04:10.358
where her face was distorted
to look like a monkey.

00:04:10.358 --> 00:04:13.344
And that picture was published all over.

00:04:13.680 --> 00:04:17.141
And people published it
very, very purposefully,

00:04:17.141 --> 00:04:19.179
to get it up there in the search results.

00:04:19.179 --> 00:04:22.345
They made sure to write
"Michelle Obama" in the caption

00:04:22.345 --> 00:04:26.309
and they made sure to upload the picture
as "MichelleObama.jpeg," or the like.

00:04:26.309 --> 00:04:28.837
You get why -- to manipulate
the search result.

00:04:28.837 --> 00:04:30.323
And it worked, too.

00:04:30.323 --> 00:04:33.119
So when you picture-Googled
for "Michelle Obama" in 2009,

00:04:33.119 --> 00:04:36.464
that distorted monkey picture
showed up among the first results.

00:04:36.464 --> 00:04:39.851
Now, the results are self-cleansing,

00:04:39.851 --> 00:04:41.678
and that's sort of the beauty of it,

00:04:41.678 --> 00:04:45.078
because Google measures relevance
every hour, every day.

00:04:45.078 --> 00:04:47.972
However, Google didn't settle
for that this time,

00:04:47.972 --> 00:04:50.903
they just thought, "That's racist
and it's a bad search result

00:04:50.903 --> 00:04:54.022
and we're going to go back
and clean that up manually.

00:04:54.343 --> 00:04:57.063
We are going to write
some code and fix it,"

00:04:57.063 --> 00:04:58.473
which they did.

00:04:59.115 --> 00:05:02.792
And I don't think anyone in this room
thinks that was a bad idea.

00:05:03.392 --> 00:05:04.599
Me neither.

00:05:06.367 --> 00:05:09.098
But then, a couple of years go by,

00:05:09.401 --> 00:05:12.282
and the world's most-Googled Anders,

00:05:12.282 --> 00:05:14.648
Anders Behring Breivik,

00:05:14.648 --> 00:05:16.219
did what he did.

00:05:16.443 --> 00:05:18.518
This is July 22 in 2011,

00:05:18.518 --> 00:05:21.172
and a terrible day in Norwegian history.

00:05:21.172 --> 00:05:25.011
This man, a terrorist, blew up
a couple of government buildings

00:05:25.011 --> 00:05:27.814
walking distance from where we are
right now in Oslo, Norway

00:05:27.814 --> 00:05:29.864
and then he traveled
to the island of Utøya

00:05:29.864 --> 00:05:32.239
and shot and killed a group of kids.

00:05:32.487 --> 00:05:34.486
Almost 80 people died that day.

00:05:35.966 --> 00:05:40.621
And a lot of people would describe
this act of terror as two steps,

00:05:40.621 --> 00:05:44.073
that he did two things: he blew up
the buildings and he shot those kids.

00:05:44.073 --> 00:05:45.580
It's not true.

00:05:45.890 --> 00:05:47.851
It was three steps.

00:05:47.851 --> 00:05:50.382
He blew up those buildings,
he shot those kids,

00:05:50.382 --> 00:05:54.035
and he sat down and waited
for the world to Google him.

00:05:54.771 --> 00:05:57.425
And he prepared
all three steps equally well.

00:05:58.003 --> 00:06:00.730
And if there was somebody
who immediately understood this,

00:06:00.730 --> 00:06:02.525
it was a Swedish web developer,

00:06:02.525 --> 00:06:06.210
a search engine optimization expert
in Stockholm, named Nikke Lindqvist.

00:06:06.210 --> 00:06:07.952
He's also a very political guy

00:06:07.952 --> 00:06:11.219
and he was right out there
in social media, on his blog and Facebook.

00:06:11.219 --> 00:06:12.671
And he told everybody,

00:06:12.695 --> 00:06:15.150
"If there's something that
this guy wants right now,

00:06:15.174 --> 00:06:17.633
it's to control the image of himself.

00:06:18.291 --> 00:06:20.466
Let's see if we can distort that.

00:06:21.490 --> 00:06:25.467
Let's see if we, in the civilized world,
can protest against what he did

00:06:25.491 --> 00:06:28.808
through insulting him
in his search results."

00:06:28.832 --> 00:06:30.019
And how?

00:06:30.332 --> 00:06:32.540
He told all of his readers the following,

00:06:32.540 --> 00:06:34.451
"Go out there on the Internet,

00:06:34.451 --> 00:06:37.452
find pictures of dog poop on sidewalks --

00:06:38.384 --> 00:06:40.536
find pictures of dog poop on sidewalks --

00:06:40.536 --> 00:06:44.087
publish them in your feeds,
on your websites, on your blogs.

00:06:44.087 --> 00:06:47.037
Make sure to write the terrorist's
name in the caption,

00:06:47.037 --> 00:06:51.386
make sure to name
the picture file "Breivik.jpeg."

00:06:51.386 --> 00:06:55.195
Let's teach Google that that's
the face of the terrorist."

00:06:57.053 --> 00:06:58.584
And it worked.

00:06:59.612 --> 00:07:02.416
Two years after that campaign
against Michelle Obama,

00:07:02.416 --> 00:07:05.727
this manipulation campaign
against Anders Behring Breivik worked.

00:07:05.727 --> 00:07:10.183
If you picture-Googled for him weeks after
the July 22 events from Sweden,

00:07:10.183 --> 00:07:14.382
you'd see that picture of dog poop
high up in the search results,

00:07:14.382 --> 00:07:16.036
as a little protest.

00:07:17.008 --> 00:07:21.176
Strangely enough, Google
didn't intervene this time.

00:07:22.074 --> 00:07:26.449
They did not step in and manually
clean those search results up.

00:07:27.431 --> 00:07:29.360
So the million-dollar question,

00:07:29.360 --> 00:07:32.821
is there anything different
between these two happenings here?

00:07:32.821 --> 00:07:36.128
Is there anything different between
what happened to Michelle Obama

00:07:36.128 --> 00:07:38.378
and what happened
to Anders Behring Breivik?

00:07:38.402 --> 00:07:39.686
Of course not.

00:07:40.353 --> 00:07:42.022
It's the exact same thing,

00:07:42.022 --> 00:07:44.731
yet Google intervened in one case
and not in the other.

00:07:44.731 --> 00:07:46.139
Why?

00:07:47.076 --> 00:07:50.125
Because Michelle Obama
is an honorable person, that's why,

00:07:50.125 --> 00:07:53.228
and Anders Behring Breivik
is a despicable person.

00:07:53.707 --> 00:07:55.450
See what happens there?

00:07:55.450 --> 00:07:58.491
An evaluation of a person takes place

00:07:58.491 --> 00:08:01.611
and there's only one
power-player in the world

00:08:02.148 --> 00:08:04.997
with the authority to say who's who.

00:08:05.701 --> 00:08:07.623
"We like you, we dislike you.

00:08:07.647 --> 00:08:09.686
We believe in you,
we don't believe in you.

00:08:09.710 --> 00:08:12.257
You're right, you're wrong.
You're true, you're false.

00:08:12.281 --> 00:08:14.086
You're Obama, and you're Breivik."

00:08:14.412 --> 00:08:16.579
That's power if I ever saw it.

00:08:18.609 --> 00:08:22.532
So I'm asking you to remember
that behind every algorithm

00:08:22.532 --> 00:08:24.357
is always a person,

00:08:24.357 --> 00:08:26.911
a person with a set of personal beliefs

00:08:26.911 --> 00:08:29.397
that no code can ever
completely eradicate.

00:08:29.397 --> 00:08:31.935
And my message goes
out not only to Google,

00:08:31.935 --> 00:08:34.789
but to all believers in the faith
of code around the world.

00:08:34.789 --> 00:08:37.719
You need to identify
your own personal bias.

00:08:37.719 --> 00:08:40.056
You need to understand that you are human

00:08:40.080 --> 00:08:42.571
and take responsibility accordingly.

00:08:43.528 --> 00:08:46.459
And I say this because I believe
we've reached a point in time

00:08:46.459 --> 00:08:48.164
when it's absolutely imperative

00:08:48.164 --> 00:08:51.252
that we tie those bonds
together again, tighter:

00:08:51.252 --> 00:08:53.751
the humanities and the technology.

00:08:54.077 --> 00:08:56.020
Tighter than ever.

00:08:56.020 --> 00:08:59.438
And, if nothing else, to remind us
that that wonderfully seductive idea

00:08:59.438 --> 00:09:01.995
of the unbiased, clean search result

00:09:01.995 --> 00:09:04.674
is, and is likely to remain, a myth.

00:09:05.390 --> 00:09:06.894
Thank you for your time.

00:09:06.894 --> 00:09:08.726
(Applause)