WEBVTT

00:00:01.354 --> 00:00:04.489
Technology has brought us so much:

00:00:04.489 --> 00:00:06.508
the moon landing, the Internet,

00:00:06.508 --> 00:00:09.133
the ability to sequence the human genome.

00:00:09.133 --> 00:00:12.857
But it also taps into a lot of our deepest fears,

00:00:12.857 --> 00:00:14.713
and about 30 years ago,

00:00:14.713 --> 00:00:17.266
the culture critic Neil Postman wrote a book

00:00:17.266 --> 00:00:19.381
called "Amusing Ourselves to Death,"

00:00:19.381 --> 00:00:22.140
which lays this out really brilliantly.

00:00:22.140 --> 00:00:23.790
And here's what he said,

00:00:23.790 --> 00:00:26.053
comparing the dystopian visions

00:00:26.053 --> 00:00:29.626
of George Orwell and Aldous Huxley.

00:00:29.626 --> 00:00:32.752
He said, Orwell feared we would become

00:00:32.752 --> 00:00:35.000
a captive culture.

00:00:35.000 --> 00:00:38.752
Huxley feared we would become a trivial culture.

00:00:38.752 --> 00:00:40.897
Orwell feared the truth would be

00:00:40.897 --> 00:00:42.820
concealed from us,

00:00:42.820 --> 00:00:45.010
and Huxley feared we would be drowned

00:00:45.010 --> 00:00:47.703
in a sea of irrelevance.

00:00:47.703 --> 00:00:49.873
In a nutshell, it's a choice between

00:00:49.873 --> 00:00:52.473
Big Brother watching you

00:00:52.473 --> 00:00:54.969
and you watching Big Brother.

00:00:54.969 --> 00:00:56.900
(Laughter)

NOTE Paragraph

00:00:56.900 --> 00:00:58.634
But it doesn't have to be this way.

00:00:58.634 --> 00:01:01.970
We are not passive consumers
of data and technology.

00:01:01.970 --> 00:01:04.373
We shape the role it plays in our lives

00:01:04.373 --> 00:01:06.503
and the way we make meaning from it,

00:01:06.503 --> 00:01:08.106
but to do that,

00:01:08.106 --> 00:01:11.619
we have to pay as much attention to how we think

00:01:11.619 --> 00:01:13.649
as how we code.

00:01:13.649 --> 00:01:16.747
We have to ask questions, and hard questions,

00:01:16.747 --> 00:01:18.616
to move past counting things

00:01:18.616 --> 00:01:21.218
to understanding them.

00:01:21.218 --> 00:01:23.664
We're constantly bombarded with stories

00:01:23.664 --> 00:01:26.140
about how much data there is in the world,

00:01:26.140 --> 00:01:27.720
but when it comes to big data

00:01:27.720 --> 00:01:30.316
and the challenges of interpreting it,

00:01:30.316 --> 00:01:32.404
size isn't everything.

00:01:32.404 --> 00:01:35.307
There's also the speed at which it moves,

00:01:35.307 --> 00:01:37.003
and the many varieties of data types,

00:01:37.003 --> 00:01:39.501
and here are just a few examples:

00:01:39.501 --> 00:01:41.699
images,

00:01:41.699 --> 00:01:45.706
text,

00:01:45.706 --> 00:01:47.801
video,

00:01:47.801 --> 00:01:49.631
audio.

00:01:49.631 --> 00:01:52.673
And what unites this disparate types of data

00:01:52.673 --> 00:01:54.894
is that they're created by people

00:01:54.894 --> 00:01:57.669
and they require context.

NOTE Paragraph

00:01:57.669 --> 00:02:00.114
Now, there's a group of data scientists

00:02:00.114 --> 00:02:02.419
out of the University of Illinois-Chicago,

00:02:02.419 --> 00:02:04.973
and they're called the Health Media Collaboratory,

00:02:04.973 --> 00:02:07.560
and they've been working with
the Centers for Disease Control

00:02:07.560 --> 00:02:09.065
to better understand

00:02:09.065 --> 00:02:11.913
how people talk about quitting smoking,

00:02:11.913 --> 00:02:14.593
how they talk about electronic cigarettes,

00:02:14.593 --> 00:02:16.578
and what they can do collectively

00:02:16.578 --> 00:02:18.562
to help them quit.

00:02:18.562 --> 00:02:20.575
The interesting thing is, if you want to understand

00:02:20.575 --> 00:02:22.791
how people talk about smoking,

00:02:22.791 --> 00:02:24.692
first you have to understand

00:02:24.692 --> 00:02:27.257
what they mean when they say "smoking."

00:02:27.257 --> 00:02:31.183
And on Twitter, there are four main categories:

00:02:31.183 --> 00:02:34.180
number one, smoking cigarettes;

00:02:34.180 --> 00:02:36.987
number two, smoking marijuana;

00:02:36.987 --> 00:02:39.630
number three, smoking ribs;

00:02:39.630 --> 00:02:43.183
and number four, smoking hot women.

00:02:43.183 --> 00:02:46.176
(Laughter)

NOTE Paragraph

00:02:46.176 --> 00:02:48.602
So then you have to think about, well,

00:02:48.602 --> 00:02:50.742
how do people talk about electronic cigarettes?

00:02:50.742 --> 00:02:52.767
And there are so many different ways

00:02:52.767 --> 00:02:55.366
that people do this, and you can see from the slide

00:02:55.366 --> 00:02:57.976
it's a complex kind of a query.

00:02:57.976 --> 00:03:01.200
And what it reminds us is that

00:03:01.200 --> 00:03:03.611
language is created by people,

00:03:03.611 --> 00:03:05.951
and people are messy and we're complex

00:03:05.951 --> 00:03:08.718
and we use metaphors and slang and jargon

00:03:08.718 --> 00:03:11.997
and we do this 24/7 in many, many languages,

00:03:11.997 --> 00:03:15.221
and then as soon as we figure it out, we change it up.

NOTE Paragraph

00:03:15.221 --> 00:03:20.339
So did these ads that the CDC put on,

00:03:20.339 --> 00:03:22.769
these television ads that featured a woman

00:03:22.769 --> 00:03:24.790
with a hole in her throat and that were very graphic

00:03:24.790 --> 00:03:26.694
and very disturbing,

00:03:26.694 --> 00:03:28.579
did they actually have an impact

00:03:28.579 --> 00:03:31.250
on whether people quit?

00:03:31.250 --> 00:03:34.557
And the Health Media Collaboratory
respected the limits of their data,

00:03:34.557 --> 00:03:36.562
but they were able to conclude

00:03:36.562 --> 00:03:39.874
that those advertisements —
and you may have seen them —

00:03:39.874 --> 00:03:42.465
that they had the effect of jolting people

00:03:42.465 --> 00:03:44.287
into a thought process

00:03:44.287 --> 00:03:47.954
that may have an impact on future behavior.

00:03:47.954 --> 00:03:51.845
And what I admire and 
appreciate about this project,

00:03:51.845 --> 00:03:53.334
aside from the fact, including the fact

00:03:53.334 --> 00:03:57.391
that it's based on real human need,

00:03:57.391 --> 00:04:00.237
is that it's a fantastic example of courage

00:04:00.237 --> 00:04:04.680
in the face of a sea of irrelevance.

NOTE Paragraph

00:04:04.680 --> 00:04:07.985
And so it's not just big data that causes

00:04:07.985 --> 00:04:10.586
challenges of interpretation, because let's face it,

00:04:10.586 --> 00:04:13.180
we human beings have a very rich history

00:04:13.180 --> 00:04:15.873
of taking any amount of data, no matter how small,

00:04:15.873 --> 00:04:17.490
and screwing it up.

00:04:17.490 --> 00:04:21.227
So many years ago, you may remember

00:04:21.227 --> 00:04:23.500
that former President Ronald Reagan

00:04:23.500 --> 00:04:25.491
was very criticized for making a statement

00:04:25.491 --> 00:04:28.501
that facts are stupid things.

00:04:28.501 --> 00:04:31.295
And it was a slip of the tongue, let's be fair.

00:04:31.295 --> 00:04:33.725
He actually meant to quote John Adams' defense

00:04:33.725 --> 00:04:36.476
of British soldiers in the Boston Massacre trials

00:04:36.476 --> 00:04:39.626
that facts are stubborn things.

00:04:39.626 --> 00:04:42.250
But I actually think there's

00:04:42.250 --> 00:04:45.668
a bit of accidental wisdom in what he said,

00:04:45.668 --> 00:04:48.444
because facts are stubborn things,

00:04:48.444 --> 00:04:51.367
but sometimes they're stupid, too.

NOTE Paragraph

00:04:51.367 --> 00:04:53.255
I want to tell you a personal story

00:04:53.255 --> 00:04:56.803
about why this matters a lot to me.

00:04:56.803 --> 00:04:59.240
I need to take a breath.

00:04:59.240 --> 00:05:01.994
My son Isaac, when he was two,

00:05:01.994 --> 00:05:04.411
was diagnosed with autism,

00:05:04.411 --> 00:05:06.572
and he was this happy, hilarious,

00:05:06.572 --> 00:05:08.607
loving, affectionate little guy,

00:05:08.607 --> 00:05:11.509
but the metrics on his developmental evaluations,

00:05:11.509 --> 00:05:13.579
which looked at things like 
the number of words —

00:05:13.579 --> 00:05:17.236
at that point, none —

00:05:17.236 --> 00:05:21.176
communicative gestures and minimal eye contact,

00:05:21.176 --> 00:05:23.179
put his developmental level

00:05:23.179 --> 00:05:27.140
at that of a nine-month-old baby.

00:05:27.140 --> 00:05:30.100
And the diagnosis was factually correct,

00:05:30.100 --> 00:05:33.309
but it didn't tell the whole story.

00:05:33.309 --> 00:05:34.710
And about a year and a half later,

00:05:34.710 --> 00:05:36.812
when he was almost four,

00:05:36.812 --> 00:05:39.175
I found him in front of the computer one day

00:05:39.175 --> 00:05:44.628
running a Google image search on women,

00:05:44.628 --> 00:05:48.244
spelled "w-i-m-e-n."

00:05:48.244 --> 00:05:50.984
And I did what any obsessed parent would do,

00:05:50.984 --> 00:05:52.885
which is immediately started 
hitting the "back" button

00:05:52.885 --> 00:05:56.248
to see what else he'd been searching for.

00:05:56.248 --> 00:05:58.419
And they were, in order: men,

00:05:58.419 --> 00:06:05.686
school, bus and computer.

00:06:05.686 --> 00:06:07.756
And I was stunned,

00:06:07.756 --> 00:06:09.758
because we didn't know that he could spell,

00:06:09.758 --> 00:06:11.524
much less read, and so I asked him,

00:06:11.524 --> 00:06:13.717
"Isaac, how did you do this?"

00:06:13.717 --> 00:06:16.395
And he looked at me very seriously and said,

00:06:16.395 --> 00:06:19.747
"Typed in the box."

NOTE Paragraph

00:06:19.747 --> 00:06:23.481
He was teaching himself to communicate,

00:06:23.481 --> 00:06:26.485
but we were looking in the wrong place,

00:06:26.485 --> 00:06:28.780
and this is what happens when assessments

00:06:28.780 --> 00:06:31.176
and analytics overvalue one metric —

00:06:31.176 --> 00:06:33.785
in this case, verbal communication —

00:06:33.785 --> 00:06:39.488
and undervalue others, such
as creative problem-solving.

00:06:39.488 --> 00:06:41.795
Communication was hard for Isaac,

00:06:41.795 --> 00:06:43.707
and so he found a workaround

00:06:43.707 --> 00:06:46.564
to find out what he needed to know.

00:06:46.564 --> 00:06:48.454
And when you think about it, it makes a lot of sense,

00:06:48.454 --> 00:06:50.535
because forming a question

00:06:50.535 --> 00:06:53.100
is a really complex process,

00:06:53.100 --> 00:06:55.622
but he could get himself a lot of the way there

00:06:55.622 --> 00:06:59.714
by putting a word in a search box.

NOTE Paragraph

00:06:59.714 --> 00:07:02.650
And so this little moment

00:07:02.650 --> 00:07:05.486
had a really profound impact on me

00:07:05.486 --> 00:07:06.795
and our family

00:07:06.795 --> 00:07:09.936
because it helped us change our frame of reference

00:07:09.936 --> 00:07:12.144
for what was going on with him,

00:07:12.144 --> 00:07:15.120
and worry a little bit less and appreciate

00:07:15.120 --> 00:07:17.302
his resourcefulness more.

NOTE Paragraph

00:07:17.302 --> 00:07:20.163
Facts are stupid things.

00:07:20.163 --> 00:07:22.560
And they're vulnerable to misuse,

00:07:22.560 --> 00:07:24.213
willful or otherwise.

00:07:24.213 --> 00:07:27.239
I have a friend, Emily Willingham, who's a scientist,

00:07:27.239 --> 00:07:30.040
and she wrote a piece for Forbes not long ago

00:07:30.040 --> 00:07:32.020
entitled "The 10 Weirdest Things

00:07:32.020 --> 00:07:33.830
Ever Linked to Autism."

00:07:33.830 --> 00:07:36.835
It's quite a list.

00:07:36.835 --> 00:07:40.367
The Internet, blamed for everything, right?

00:07:40.367 --> 00:07:44.124
And of course mothers, because.

00:07:44.124 --> 00:07:45.711
And actually, wait, there's more,

00:07:45.711 --> 00:07:49.141
there's a whole bunch in 
the "mother" category here.

00:07:49.141 --> 00:07:53.956
And you can see it's a pretty 
rich and interesting list.

00:07:53.956 --> 00:07:56.149
I'm a big fan of

00:07:56.149 --> 00:07:59.853
being pregnant near freeways, personally.

00:07:59.853 --> 00:08:01.392
The final one is interesting,

00:08:01.392 --> 00:08:04.395
because the term "refrigerator mother"

00:08:04.395 --> 00:08:07.000
was actually the original hypothesis

00:08:07.000 --> 00:08:08.431
for the cause of autism,

00:08:08.431 --> 00:08:11.166
and that meant somebody 
who was cold and unloving.

NOTE Paragraph

00:08:11.166 --> 00:08:12.728
And at this point, you might be thinking,

00:08:12.728 --> 00:08:14.385
"Okay, Susan, we get it,

00:08:14.385 --> 00:08:16.167
you can take data, you can 
make it mean anything."

00:08:16.167 --> 00:08:20.870
And this is true, it's absolutely true,

00:08:20.870 --> 00:08:26.480
but the challenge is that

00:08:26.480 --> 00:08:28.928
we have this opportunity

00:08:28.928 --> 00:08:31.212
to try to make meaning out of it ourselves,

00:08:31.212 --> 00:08:36.564
because frankly, data doesn't 
create meaning. We do.

00:08:36.564 --> 00:08:39.820
So as businesspeople, as consumers,

00:08:39.820 --> 00:08:42.359
as patients, as citizens,

00:08:42.359 --> 00:08:44.755
we have a responsibility, I think,

00:08:44.755 --> 00:08:46.949
to spend more time

00:08:46.949 --> 00:08:49.819
focusing on our critical thinking skills.

00:08:49.819 --> 00:08:50.897
Why?

00:08:50.897 --> 00:08:54.075
Because at this point in our history, as we've heard

00:08:54.075 --> 00:08:55.781
many times over,

00:08:55.781 --> 00:08:57.762
we can process exabytes of data

00:08:57.762 --> 00:08:59.915
at lightning speed,

00:08:59.915 --> 00:09:03.430
and we have the potential to make bad decisions

00:09:03.430 --> 00:09:05.264
far more quickly, efficiently,

00:09:05.264 --> 00:09:10.292
and with far greater impact than we did in the past.

00:09:10.292 --> 00:09:11.680
Great, right?

00:09:11.680 --> 00:09:14.710
And so what we need to do instead

00:09:14.710 --> 00:09:17.040
is spend a little bit more time

00:09:17.040 --> 00:09:19.786
on things like the humanities

00:09:19.786 --> 00:09:23.250
and sociology, and the social sciences,

00:09:23.250 --> 00:09:25.558
rhetoric, philosophy, ethics,

00:09:25.558 --> 00:09:28.414
because they give us context that is so important

00:09:28.414 --> 00:09:30.990
for big data, and because

00:09:30.990 --> 00:09:33.408
they help us become better critical thinkers.

00:09:33.408 --> 00:09:37.615
Because after all, if I can spot

00:09:37.615 --> 00:09:40.101
a problem in an argument, it doesn't much matter

00:09:40.101 --> 00:09:42.860
whether it's expressed in words or in numbers.

00:09:42.860 --> 00:09:45.579
And this means

00:09:45.579 --> 00:09:50.000
teaching ourselves to find 
those confirmation biases

00:09:50.000 --> 00:09:51.822
and false correlations

00:09:51.822 --> 00:09:53.960
and being able to spot a naked emotional appeal

00:09:53.960 --> 00:09:55.622
from 30 yards,

00:09:55.622 --> 00:09:58.144
because something that happens after something

00:09:58.144 --> 00:10:01.226
doesn't mean it happened 
because of it, necessarily,

00:10:01.226 --> 00:10:03.345
and if you'll let me geek out on you for a second,

00:10:03.345 --> 00:10:07.642
the Romans called this 
"post hoc ergo propter hoc,"

00:10:07.642 --> 00:10:10.938
after which therefore because of which.

NOTE Paragraph

00:10:10.938 --> 00:10:14.695
And it means questioning
disciplines like demographics.

00:10:14.695 --> 00:10:17.215
Why? Because they're based on assumptions

00:10:17.215 --> 00:10:19.521
about who we all are based on our gender

00:10:19.521 --> 00:10:20.983
and our age and where we live

00:10:20.983 --> 00:10:24.461
as opposed to data on what 
we actually think and do.

00:10:24.461 --> 00:10:26.124
And since we have this data,

00:10:26.124 --> 00:10:29.263
we need to treat it with appropriate privacy controls

00:10:29.263 --> 00:10:32.839
and consumer opt-in,

00:10:32.839 --> 00:10:35.832
and beyond that, we need to be clear

00:10:35.832 --> 00:10:37.935
about our hypotheses,

00:10:37.935 --> 00:10:40.531
the methodologies that we use,

00:10:40.531 --> 00:10:43.335
and our confidence in the result.

00:10:43.335 --> 00:10:45.809
As my high school algebra teacher used to say,

00:10:45.809 --> 00:10:47.340
show your math,

00:10:47.340 --> 00:10:50.781
because if I don't know what steps you took,

00:10:50.781 --> 00:10:52.772
I don't know what steps you didn't take,

00:10:52.772 --> 00:10:55.210
and if I don't know what questions you asked,

00:10:55.210 --> 00:10:58.407
I don't know what questions you didn't ask.

00:10:58.407 --> 00:10:59.930
And it means asking ourselves, really,

00:10:59.930 --> 00:11:01.409
the hardest question of all:

00:11:01.409 --> 00:11:04.909
Did the data really show us this,

00:11:04.909 --> 00:11:07.220
or does the result make us feel

00:11:07.220 --> 00:11:11.098
more successful and more comfortable?

NOTE Paragraph

00:11:11.098 --> 00:11:13.682
So the Health Media Collaboratory,

00:11:13.682 --> 00:11:15.381
at the end of their project, they were able

00:11:15.381 --> 00:11:18.789
to find that 87 percent of tweets

00:11:18.789 --> 00:11:20.933
about those very graphic and disturbing

00:11:20.933 --> 00:11:24.971
anti-smoking ads expressed fear,

00:11:24.971 --> 00:11:26.827
but did they conclude

00:11:26.827 --> 00:11:29.988
that they actually made people stop smoking?

00:11:29.988 --> 00:11:32.530
No. It's science, not magic.

NOTE Paragraph

00:11:32.530 --> 00:11:35.720
So if we are to unlock

00:11:35.720 --> 00:11:38.582
the power of data,

00:11:38.582 --> 00:11:42.030
we don't have to go blindly into

00:11:42.030 --> 00:11:45.466
Orwell's vision of a totalitarian future,

00:11:45.466 --> 00:11:48.583
or Huxley's vision of a trivial one,

00:11:48.583 --> 00:11:51.603
or some horrible cocktail of both.

00:11:51.603 --> 00:11:53.982
What we have to do

00:11:53.982 --> 00:11:56.700
is treat critical thinking with respect

00:11:56.700 --> 00:11:58.729
and be inspired by examples

00:11:58.729 --> 00:12:01.339
like the Health Media Collaboratory,

00:12:01.339 --> 00:12:03.667
and as they say in the superhero movies,

00:12:03.667 --> 00:12:05.489
let's use our powers for good.

NOTE Paragraph

00:12:05.489 --> 00:12:07.840
Thank you.

NOTE Paragraph

00:12:07.840 --> 00:12:10.174
(Applause)