How to pronounce "algorithms"

Translator: Joseph Geni Reviewer: Morton Bast

Let me tell you, it has been a fantastic month for deception.

And I'm not even talking about the American presidential race. (Laughter)

We have a high-profile journalist caught for plagiarism,

a young superstar writer whose book involves

so many made up quotes that they've pulled it from the shelves;

a New York Times exposé on fake book reviews.

It's been fantastic.

Now, of course, not all deception hits the news.

Much of the deception is everyday. In fact, a lot of research

shows that we all lie once or twice a day, as Dave suggested.

So it's about 6:30 now, suggests that most of us should have lied.

Let's take a look at Winnipeg. How many of you,

in the last 24 hours -- think back -- have told a little fib,

or a big one? How many have told a little lie out there?

All right, good. These are all the liars.

Make sure you pay attention to them. (Laughter)

No, that looked good, it was about two thirds of you.

The other third didn't lie, or perhaps forgot,

or you're lying to me about your lying, which is very,

very devious. (Laughter) This fits with a lot of the research,

which suggests that lying is very pervasive.

It's this pervasiveness, combined with the centrality

to what it means to be a human, the fact that we can

tell the truth or make something up,

that has fascinated people throughout history.

Here we have Diogenes with his lantern.

Does anybody know what he was looking for?

A single honest man, and he died without finding one

back in Greece. And we have Confucius in the East

who was really concerned with sincerity,

not only that you walked the walk or talked the talk,

but that you believed in what you were doing.

You believed in your principles.

Now my first professional encounter with deception

is a little bit later than these guys, a couple thousand years.

I was a customs officer for Canada back in the mid-'90s.

Yeah. I was defending Canada's borders.

You may think that's a weapon right there. In fact,

that's a stamp. I used a stamp to defend Canada's borders. (Laughter)

Very Canadian of me. I learned a lot about deception

while doing my duty here in customs,

one of which was that most of what I thought I knew about deception was wrong,

and I'll tell you about some of that tonight.

But even since just 1995, '96, the way we communicate

has been completely transformed. We email, we text,

we skype, we Facebook. It's insane.

Almost every aspect of human communication's been changed,

and of course that's had an impact on deception.

Let me tell you a little bit about a couple of new deceptions

we've been tracking and documenting.

They're called the Butler, the Sock Puppet

and the Chinese Water Army.

It sounds a little bit like a weird book,

but actually they're all new types of lies.

Let's start with the Butlers. Here's an example of one:

"On my way." Anybody ever written, "On my way?"

Then you've also lied. (Laughter)

We're never on our way. We're thinking about going on our way.

Here's another one: "Sorry I didn't respond to you earlier.

My battery was dead." Your battery wasn't dead.

You weren't in a dead zone.

You just didn't want to respond to that person that time.

Here's the last one: You're talking to somebody,

and you say, "Sorry, got work, gotta go."

But really, you're just bored. You want to talk to somebody else.

Each of these is about a relationship,

and this is a 24/7 connected world. Once you get my cell phone number,

you can literally be in touch with me 24 hours a day.

And so these lies are being used by people

to create a buffer, like the butler used to do,

between us and the connections to everybody else.

But they're very special. They use ambiguity

that comes from using technology. You don't know

where I am or what I'm doing or who I'm with.

And they're aimed at protecting the relationships.

These aren't just people being jerks. These are people

that are saying, look, I don't want to talk to you now,

or I didn't want to talk to you then, but I still care about you.

Our relationship is still important.

Now, the Sock Puppet, on the other hand,

is a totally different animal. The sock puppet isn't

about ambiguity, per se. It's about identity.

Let me give you a very recent example,

as in, like, last week.

Here's R.J. Ellory, best-seller author in Britain.

Here's one of his bestselling books.

Here's a reviewer online, on Amazon.

My favorite, by Nicodemus Jones, is,

"Whatever else it might do, it will touch your soul."

And of course, you might suspect

that Nicodemus Jones is R.J. Ellory.

He wrote very, very positive reviews about himself. Surprise, surprise.

Now this Sock Puppet stuff isn't actually that new.

Walt Whitman also did this back in the day,

before there was Internet technology. Sock Puppet

becomes interesting when we get to scale,

which is the domain of the Chinese Water Army.

Chinese Water Army refers to thousands of people

in China that are paid small amounts of money

to produce content. It could be reviews. It could be

propaganda. The government hires these people,

companies hire them, all over the place.

In North America, we call this Astroturfing,

and Astroturfing is very common now. There's a lot of concerns about it.

We see this especially with product reviews, book reviews,

everything from hotels to whether that toaster is a good toaster or not.

Now, looking at these three reviews, or these three types of deception,

you might think, wow, the Internet is really making us

a deceptive species, especially when you think about

the Astroturfing, where we can see deception brought up to scale.

But actually, what I've been finding is very different from that.

Now, let's put aside the online anonymous sex chatrooms,

which I'm sure none of you have been in.

I can assure you there's deception there.

And let's put aside the Nigerian prince who's emailed you

about getting the 43 million out of the country. (Laughter)

Let's forget about that guy, too.

Let's focus on the conversations between our friends

and our family and our coworkers and our loved ones.

Those are the conversations that really matter.

What does technology do to deception with those folks?

Here's a couple of studies. One of the studies we do

are called diary studies, in which we ask people to record

all of their conversations and all of their lies for seven days,

and what we can do then is calculate how many lies took place

per conversation within a medium, and the finding

that we get that surprises people the most is that email

is the most honest of those three media.

And it really throws people for a loop because we think,

well, there's no nonverbal cues, so why don't you lie more?

The phone, in contrast, the most lies.

Again and again and again we see the phone is the device

that people lie on the most, and perhaps because of the Butler Lie ambiguities I was telling you about.

This tends to be very different from what people expect.

What about résumés? We did a study in which we had

people apply for a job, and they could apply for a job

either with a traditional paper résumé, or on LinkedIn,

which is a social networking site like Facebook,

but for professionals -- involves the same information as a résumé.

And what we found, to many people's surprise,

was that those LinkedIn résumés were more honest

on the things that mattered to employers, like your

responsibilities or your skills at your previous job.

How about Facebook itself?

You know, we always think that hey, there are these

idealized versions, people are just showing the best things

that happened in their lives. I've thought that many times.

My friends, no way they can be that cool and have good of a life.

Well, one study tested this by examining people's personalities.

They had four good friends of a person judge their personality.

Then they had strangers, many strangers,

judge the person's personality just from Facebook,

and what they found was those judgments of the personality

were pretty much identical, highly correlated,

meaning that Facebook profiles really do reflect our actual personality.

All right, well, what about online dating?

I mean, that's a pretty deceptive space.

I'm sure you all have "friends" that have used online dating. (Laughter)

And they would tell you about that guy that had no hair

when he came, or the woman that didn't look at all like her photo.

Well, we were really interested in it, and so what we did

is we brought people, online daters, into the lab,

and then we measured them. We got their height

up against the wall, we put them on a scale, got their weight --

ladies loved that -- and then we actually got their driver's license to get their age.

And what we found was very, very interesting.

Here's an example of the men and the height.

Along the bottom is how tall they said they were in their profile.

Along the y-axis, the vertical axis, is how tall they actually were.

That diagonal line is the truth line. If their dot's on it,

they were telling exactly the truth.

Now, as you see, most of the little dots are below the line.

What it means is all the guys were lying about their height.

In fact, they lied about their height about nine tenths of an inch,

what we say in the lab as "strong rounding up." (Laughter)

You get to 5'8" and one tenth, and boom! 5'9".

But what's really important here is, look at all those dots.

They are clustering pretty close to the truth. What we found

was 80 percent of our participants did indeed lie

on one of those dimensions, but they always lied by a little bit.

One of the reasons is pretty simple. If you go to a date,

a coffee date, and you're completely different than what you said,

game over. Right? So people lied frequently, but they lied

subtly, not too much. They were constrained.

Well, what explains all these studies? What explains the fact

that despite our intuitions, mine included,

a lot of online communication, technologically-mediated

communication, is more honest than face to face?

That really is strange. How do we explain this?

Well, to do that, one thing is we can look at the deception-detection literature.

It's a very old literature by now, it's coming up on 50 years.

It's been reviewed many times. There's been thousands of trials,

hundreds of studies, and there's some really compelling findings.

The first is, we're really bad at detecting deception,

really bad. Fifty-four percent accuracy on average when you have to tell

if somebody that just said a statement is lying or not.

That's really bad. Why is it so bad?

Well it has to do with Pinocchio's nose.

If I were to ask you guys, what do you rely on

when you're looking at somebody and you want to find out

if they're lying? What cue do you pay attention to?

Most of you would say that one of the cues you look at

is the eyes. The eyes are the window to the soul.

And you're not alone. Around the world, almost every culture,

one of the top cues is eyes. But the research

over the last 50 years says there's actually no reliable cue

to deception, which blew me away, and it's one of

the hard lessons that I learned when I was customs officer.

The eyes do not tell us whether somebody's lying or not.

Some situations, yes -- high stakes, maybe their pupils dilate,

their pitch goes up, their body movements change a little bit,

but not all the time, not for everybody, it's not reliable.

Strange. The other thing is that just because you can't see me

doesn't mean I'm going to lie. It's common sense,

but one important finding is that we lie for a reason.

We lie to protect ourselves or for our own gain

or for somebody else's gain.

So there are some pathological liars, but they make up

a tiny portion of the population. We lie for a reason.

Just because people can't see us doesn't mean

we're going to necessarily lie.

But I think there's actually something much more

interesting and fundamental going on here. The next big

thing for me, the next big idea, we can find by going

way back in history to the origins of language.

Most linguists agree that we started speaking somewhere

between 50,000 and 100,000 years ago. That's a long time ago.

A lot of humans have lived since then.

We've been talking, I guess, about fires and caves

and saber-toothed tigers. I don't know what they talked about,

but they were doing a lot of talking, and like I said,

there's a lot of humans evolving speaking,

about 100 billion people in fact.

What's important though is that writing only emerged

about 5,000 years ago. So what that means is that

all the people before there was any writing,

every word that they ever said, every utterance

disappeared. No trace. Evanescent. Gone.

So we've been evolving to talk in a way in which

there is no record. In fact, even the next big change

to writing was only 500 years ago now,

with the printing press, which is very recent in our past,

and literacy rates remained incredibly low right up until World War II,

so even the people of the last two millennia,

most of the words they ever said -- poof! -- disappeared.

Let's turn to now, the networked age.

How many of you have recorded something today?

Anybody do any writing today? Did anybody write a word?

It looks like almost every single person here recorded something.

In this room, right now, we've probably recorded more

than almost all of human pre-ancient history.

That is crazy. We're entering this amazing period

of flux in human evolution where we've evolved to speak

in a way in which our words disappear, but we're in

an environment where we're recording everything.

In fact, I think in the very near future, it's not just

what we write that will be recorded, everything we do

will be recorded.

What does that mean? What's the next big idea from that?

Well, as a social scientist, this is the most amazing thing

I have ever even dreamed of. Now, I can look at

all those words that used to, for millennia, disappear.

I can look at lies that before were said and then gone.

You remember those Astroturfing reviews that we were

talking about before? Well, when they write a fake review,

they have to post it somewhere, and it's left behind for us.

So one thing that we did, and I'll give you an example of

looking at the language, is we paid people

to write some fake reviews. One of these reviews is fake.

The person never was at the James Hotel.

The other review is real. The person stayed there.

Now, your task now is to decide

which review is fake?

I'll give you a moment to read through them.

But I want everybody to raise their hand at some point.

Remember, I study deception. I can tell if you don't raise your hand.

All right, how many of you believe that A is the fake?

All right. Very good. About half.

And how many of you think that B is?

All right. Slightly more for B.

Excellent. Here's the answer.

B is a fake. Well done second group. You dominated the first group. (Laughter)

You're actually a little bit unusual. Every time we demonstrate this,

it's usually about a 50-50 split, which fits

with the research, 54 percent. Maybe people here

in Winnipeg are more suspicious and better at figuring it out.

Those cold, hard winters, I love it.

All right, so why do I care about this?

Well, what I can do now with my colleagues in computer science

is we can create computer algorithms that can analyze

the linguistic traces of deception.

Let me highlight a couple of things here

in the fake review. The first is that liars tend to think

about narrative. They make up a story:

Who? And what happened? And that's what happened here.

Our fake reviewers talked about who they were with

and what they were doing. They also used the first person singular, I,

way more than the people that actually stayed there.

They were inserting themselves into the hotel review,

kind of trying to convince you they were there.

In contrast, the people that wrote the reviews that were actually there,

their bodies actually entered the physical space,

they talked a lot more about spatial information.

They said how big the bathroom was, or they said,

you know, here's how far shopping is from the hotel.

Now, you guys did pretty well. Most people perform at chance at this task.

Our computer algorithm is very accurate, much more accurate

than humans can be, and it's not going to be accurate all the time.

This isn't a deception-detection machine to tell

if your girlfriend's lying to you on text messaging.

We believe that every lie now, every type of lie --

fake hotel reviews, fake shoe reviews,

your girlfriend cheating on you with text messaging --

those are all different lies. They're going to have

different patterns of language. But because everything's

recorded now, we can look at all of those kinds of lies.

Now, as I said, as a social scientist, this is wonderful.

It's transformational. We're going to be able to learn

so much more about human thought and expression,

about everything from love to attitudes,

because everything is being recorded now, but

what does it mean for the average citizen?

What does it mean for us in our lives?

Well, let's forget deception for a bit. One of the big ideas,

I believe, is that we're leaving these huge traces behind.

My outbox for email is massive,

and I never look at it. I write all the time,

but I never look at my record, at my trace.

And I think we're going to see a lot more of that,

where we can reflect on who we are by looking at

what we wrote, what we said, what we did.

Now, if we bring it back to deception, there's a couple

of take-away things here.

First, lying online can be very dangerous, right?

Not only are you leaving a record for yourself on your machine,

but you're leaving a record on the person that you were lying to,

and you're also leaving them around for me to analyze

with some computer algorithms.

So by all means, go ahead and do that, that's good.

But when it comes to lying and what we want to do

with our lives, I think we can go back to

Diogenes and Confucius. And they were less concerned

about whether to lie or not to lie, and more concerned about

being true to the self, and I think this is really important.

Now, when you are about to say or do something,

we can think, do I want this to be part of my legacy,

part of my personal record?

Because in the digital age we live in now,

in the networked age, we are all leaving a record.

Thank you so much for your time,

and good luck with your record. (Applause)