How to Find Love Online Horizon


How to Find Love Online

Similar Content

Browse content similar to How to Find Love Online. Check below for episodes and series from the same categories and more!

Transcript


LineFromTo

Internet dating is big business.

0:00:100:00:12

It's worth over 2 billion per year

0:00:120:00:15

and claims to generate one fifth of all current committed relationships.

0:00:150:00:19

Worldwide, 91 million people log on to dating sites and I'm one of them.

0:00:190:00:25

Since I started dating, it has changed a huge amount.

0:00:250:00:28

Really, now, almost everyone I know has tried online dating.

0:00:280:00:32

I've been dating in both London and New York

0:00:330:00:36

and so far I've yet to find The One.

0:00:360:00:38

I'm 37, all my friends are married, my brothers are married,

0:00:400:00:44

it's not fun being single any more.

0:00:440:00:46

Although I'm happily married now, I've done my share of online dating.

0:00:460:00:51

As a mathematician, I'm fascinated by the algorithms that dating sites

0:00:510:00:55

claim can find you love,

0:00:550:00:57

but there's little hard evidence that they actually deliver.

0:00:570:01:00

So, I'm going to put them to the test using Xand as my guinea pig.

0:01:000:01:05

I'm willing to gamble with Xand's heart and see if we can use

0:01:050:01:09

a little bit of maths to find him a girl that he really likes.

0:01:090:01:13

But can applying some science really help me find love?

0:01:130:01:16

-Hiya, how are you doing?

-Oh, really nice to meet you.

0:01:160:01:19

Are matching sites any better than just choosing yourself?

0:01:190:01:23

There is no way that these algorithms can do what

0:01:230:01:26

they're claiming to do.

0:01:260:01:28

Which picture should you use?

0:01:280:01:30

God, you're like the bloomin' hair police.

0:01:300:01:32

And what's the best way to write a profile?

0:01:320:01:35

Just being a nice guy is not necessarily the best pitch.

0:01:350:01:39

We look at the science of online dating and have some fun along the way.

0:01:390:01:43

You're going right there. SHE LAUGHS

0:01:430:01:46

That's where your genitals lit up.

0:01:460:01:48

MUSIC: Ooh La La by Goldfrapp

0:01:530:01:56

Horizon has invited 50 single ladies and gentlemen

0:02:000:02:04

to a dating event in London.

0:02:040:02:06

-No-one's snogging, but...

-No.

-So far.

0:02:060:02:10

-It's early yet.

-Early days. It's early days.

0:02:100:02:12

They haven't had enough to drink yet.

0:02:120:02:15

We are all taking part in an experiment to test the

0:02:150:02:18

mathematical matching systems used by many online dating sites.

0:02:180:02:22

There is that sort of, almost placebo effect of expectation.

0:02:230:02:27

Whether that does better or not then our actual extended algorithm,

0:02:270:02:32

we're about to find out.

0:02:320:02:34

The recent upsurge in online dating is a gold mine for researchers

0:02:340:02:37

studying human behaviour

0:02:370:02:39

and it's starting to produce some good, hard science

0:02:390:02:42

about the best techniques to help those like Xand looking for love.

0:02:420:02:46

But wait till she stands up, because she's super smoking.

0:02:460:02:49

I hate to admit it, but I really need some help.

0:02:530:02:56

So, I think the worst thing about online dating is having

0:03:010:03:04

to admit to yourself that you're single

0:03:040:03:07

and that you want to meet someone else. And at 37, I don't feel old,

0:03:070:03:11

but I'm probably in a bit more of a mess than I used to be.

0:03:110:03:14

I've been to eight weddings in the last two years and I feel like

0:03:150:03:19

always the bridesmaid, never the bride does almost literally apply to me.

0:03:190:03:24

I think I am getting a little grumpy at weddings now. I think

0:03:240:03:27

I slightly do resent...

0:03:270:03:29

..other people's happiness, which is not...

0:03:310:03:35

It's not a good position to be in.

0:03:350:03:38

First couple of divorces, though, so that made me feel better.

0:03:380:03:41

I found online dating a minefield.

0:03:470:03:50

There are thousands of different sites, from mobile apps

0:03:500:03:52

that hook you up with a simple swipe,

0:03:520:03:54

to internet dating with complex matching systems

0:03:540:03:57

that promise to find you the perfect partner.

0:03:570:04:00

But which is better?

0:04:000:04:02

I get that online dating is an efficient way of encountering other people...

0:04:020:04:07

-Yeah.

-..and probably, therefore, getting me dates,

0:04:070:04:09

but it makes a much bigger claim than that.

0:04:090:04:12

-They're saying, "We'll find you someone right for you."

-Mm.

0:04:120:04:14

Well, there's certainly a range of different websites that offer different things.

0:04:140:04:18

So, some of them are just effectively a catalogue of strangers, right?

0:04:180:04:21

Whereas others have these algorithms built in,

0:04:210:04:24

which is a series of calculations.

0:04:240:04:26

So, you take some input - maybe what you're looking for in a partner,

0:04:260:04:30

what prospective partners are looking for in their partner -

0:04:300:04:34

and you put it through a series of logical steps, like a recipe effectively,

0:04:340:04:38

and in the end you come up with an output, which is how good

0:04:380:04:41

the two of you would be matched as a potential couple.

0:04:410:04:44

-So, it's like a decision tree?

-Yeah, exactly.

0:04:440:04:47

MUSIC: Escape (The Pina Colada Song) By Rupert Holmes

0:04:470:04:50

At their simplest, algorithms work like a flow chart,

0:04:510:04:55

with different inputs or instructions that feed into an end result or output.

0:04:550:05:00

OK, but I've only ever used the simplest algorithm,

0:05:000:05:04

-which is proximity, age, sex.

-Yeah.

0:05:040:05:07

But I think that if we get you to fill in a questionnaire about

0:05:070:05:12

the type of things that you find appealing in a partner,

0:05:120:05:15

I think I could write an algorithm that would find a girl

0:05:150:05:18

who is better suited for you than if you just walked into a bar and picked someone at random.

0:05:180:05:23

I'm pretty sceptical about this, but, hey, let's give it a go.

0:05:230:05:27

-And this is actually applied maths.

-Yeah.

-Like, we'll use it to get me a date.

-Yeah, totally.

0:05:270:05:32

All right, but the challenge is not for you to do better than me going to a bar,

0:05:320:05:36

you're actually up against me choosing random people through swiping.

0:05:360:05:39

-That's the challenge.

-OK.

-You're on.

-Deal.

0:05:390:05:42

With mobile dating apps like Tinder, Zoosk and Happn

0:05:490:05:52

allowing users to select or reject partners with a simple

0:05:520:05:55

swipe left or right, online daters like me can feel a little jaded.

0:05:550:06:00

It sometimes feels really superficial...

0:06:000:06:03

It's very easy for people to just reject you outright.

0:06:030:06:06

So, you think, "Oh, why are they rejecting me? What's wrong with me?"

0:06:060:06:10

And traditional dating sites have their downsides, too.

0:06:100:06:14

You get a lot of creeps online who try to send you, like,

0:06:140:06:18

gross pictures of themselves.

0:06:180:06:21

The catfish. Someone who's pretending to be somebody else.

0:06:210:06:26

You're never quite sure who's telling lies or not.

0:06:260:06:30

Let's face it - online, anybody can be anything they want.

0:06:300:06:33

In order to find Xand a date, I'm going to build an experimental

0:06:360:06:40

dating website with the help of my colleague Tom Russell.

0:06:400:06:43

This is one user's set of responses.

0:06:440:06:47

Some commercial dating sites use multiple layers of algorithms.

0:06:470:06:51

eHarmony claims they assess psychological compatibility

0:06:510:06:55

and interpersonal chemistry.

0:06:550:06:57

Lovestruck uses a recommendation engine based on search activity,

0:06:570:07:01

a bit like Netflix or Amazon.

0:07:010:07:04

Show me the bit where the scores for each question goes.

0:07:040:07:08

We're using a scored questionnaire that is similar to OkCupid.

0:07:080:07:12

So, if we have three levels, then...

0:07:120:07:14

Matching algorithms are useful because they help daters sort through

0:07:140:07:18

the vast numbers of potential partners available via online dating.

0:07:180:07:23

One of the problems with online dating is the paradox of choice.

0:07:260:07:31

It felt slightly overwhelming.

0:07:310:07:33

Surprised at the number of people who were there.

0:07:330:07:35

You can sit there for hours ploughing through people from all over the country.

0:07:350:07:41

It becomes like fishing, I guess.

0:07:410:07:43

I find it's, like, almost like a drudgery.

0:07:430:07:45

The commitment to time, is exhausting, it's mentally exhausting.

0:07:450:07:49

Xand is convinced that he can choose himself a better date

0:07:510:07:54

than my algorithm, but with thousands of potential dates

0:07:540:07:57

in New York and London to choose from,

0:07:570:08:00

I think he needs some mathematical help.

0:08:000:08:03

So, what you really need is an effective search strategy to help

0:08:030:08:07

you find the perfect woman for you, without having to date

0:08:070:08:11

-every single one of them.

-OK.

-I've got one.

0:08:110:08:14

Optimal stopping theory, it's called.

0:08:140:08:16

That is not what I thought you were going to say. OK.

0:08:160:08:18

So, it was invented in 1875 by a chap called Arthur Cayley,

0:08:180:08:23

essentially to gamble better,

0:08:230:08:25

and two Harvard mathematicians worked out the best chance you can

0:08:250:08:29

give yourself of stopping on The One, the perfect woman for you,

0:08:290:08:33

is to spend the first 37% of your dates just, kind of, not taking them too seriously,

0:08:330:08:40

having a nice time, getting a bit of a feel for the marketplace and so on.

0:08:400:08:43

And then after that 37% period has passed,

0:08:430:08:46

you should then pick the next woman to come along that is better than

0:08:460:08:50

everybody that you've seen before.

0:08:500:08:53

And if you do that, you're maximising your chances,

0:08:530:08:55

mathematically, of finding the perfect woman for you.

0:08:550:08:59

And what will the chances be of the one after 37%,

0:08:590:09:03

that one being the right woman for me?

0:09:030:09:06

Well, OK. So, if you pick somebody completely at random

0:09:060:09:09

in your list of 100 and just chose that person to go on your date with,

0:09:090:09:12

your chance of getting the best person in your list

0:09:120:09:14

would be 1%, right? One in 100.

0:09:140:09:17

But this technique - just having that little rejection phase of 37% at the beginning -

0:09:170:09:22

means that you increase your chances all the way from 1% to 37%.

0:09:220:09:27

If the number of potential dates, n, for Xand is 100,

0:09:270:09:31

his chance of success, P, is highest when he rejects

0:09:310:09:35

37% of potential partners.

0:09:350:09:38

His success rate drops off if he either reduces

0:09:380:09:42

or increases his rejection phase, r,

0:09:420:09:46

the time before he starts thinking seriously about a match.

0:09:460:09:49

OK, that is an amazing bit of maths. That is extraordinary.

0:09:490:09:53

So, imagine you decide to take 100 people,

0:09:530:10:00

reject the first 37 of them

0:10:000:10:03

and then pick the next person who comes along that's better than

0:10:030:10:06

everyone you've seen before and take that person on a date.

0:10:060:10:10

Wow!

0:10:100:10:12

-OK, so 100 people seems reasonable.

-Yeah, it's not crazy.

0:10:120:10:15

-Yeah, I can entertain 100 people.

-Of course.

-OK, so I just start...

0:10:150:10:18

Two, three, four...

0:10:180:10:20

Are there some of these that you would swipe right to?

0:10:200:10:23

This person, I would definitely swipe right for.

0:10:230:10:26

She looks lovely and according to you, I have to reject her,

0:10:260:10:29

because you told me to.

0:10:290:10:30

-So, like, this person seems nice. Nope, get rid of them.

-Yeah.

0:10:300:10:33

So, you carry on swiping left for the entire time

0:10:330:10:35

and then you end up, well, probably dying alone...

0:10:350:10:38

SHE LAUGHS

0:10:380:10:39

..nursing a deep hatred of mathematical formulas!

0:10:390:10:43

-And mathematicians!

-Yeah, probably.

0:10:430:10:45

You and the two guys at Harvard, primarily.

0:10:450:10:48

Back at home, I test Hannah's optimal stopping theory

0:10:540:10:57

on a commercial dating app.

0:10:570:10:59

I'm rejecting 37 potential dates.

0:10:590:11:03

And now the first person I see

0:11:030:11:05

who's better than everyone I've just rejected...

0:11:050:11:09

is Miss Right.

0:11:090:11:10

So, at 62, I decided to swipe right

0:11:130:11:17

and I think that this person was better than all the others.

0:11:170:11:20

I will send her a message and see if we can go on a date.

0:11:200:11:23

Well, I'm delighted she said yes and she's drop-dead gorgeous.

0:11:260:11:30

Do it again!

0:11:300:11:32

She's even volunteered to film the date

0:11:410:11:43

and tell me about her online dating experiences.

0:11:430:11:46

Surely, eHarmony's loss!

0:11:550:11:57

But as the date wore on, I wasn't convinced we were compatible.

0:11:570:12:00

It can be edited.

0:12:000:12:03

I started to wonder whether Hannah's optimal stopping theory had worked in this case.

0:12:100:12:15

I sincerely apologise for meeting Xand.

0:12:150:12:17

Well, I'm heading into work after my date with my supposed

0:12:220:12:25

Miss Right yesterday and it didn't work out.

0:12:250:12:29

So, back to the drawing board.

0:12:290:12:32

I guess choosing a date based purely on appearances is always

0:12:400:12:43

going to be a gamble.

0:12:430:12:45

However, all dating apps and websites do require a photo,

0:12:450:12:48

so it's clear that looks are important to everyone.

0:12:480:12:52

It's all about first impressions.

0:12:520:12:54

The picture engages you, then you read the profile.

0:12:540:12:57

I mean, let's be honest, it all boils down to the photos,

0:12:570:12:59

at the end of the day.

0:12:590:13:01

But it's not just attractiveness we're judged on.

0:13:010:13:04

Researchers at Princeton University have recently proved that

0:13:040:13:07

people use faces to make split-second judgements about our personalities, as well.

0:13:070:13:12

I think you can tell a lot about a person from the way they look.

0:13:130:13:16

It's really easy to make snap judgements about people based on their photo.

0:13:160:13:21

I think you can tell kindness, you can tell if someone's got a sense of fun.

0:13:210:13:25

I'd like to think that I can read intelligence in somebody's face,

0:13:250:13:29

maybe even the sense of humour that they have.

0:13:290:13:31

And while these judgements might not be right,

0:13:330:13:35

scientists have found that people tend to agree on what features

0:13:350:13:39

make someone appear likeable, trustworthy or competent.

0:13:390:13:42

And just put your chin down slightly.

0:13:420:13:44

I want to find out what people think of me.

0:13:440:13:48

So, I'm sending my picture to Dr Chris Olivola.

0:13:480:13:51

He's analysed hundreds of real online daters' reactions to profile

0:13:510:13:55

pictures and has discovered what facial attributes are most popular.

0:13:550:13:59

So, at least one of the websites I used is entirely based on

0:14:020:14:05

-swiping on pictures of people.

-Yes, I know what that site is.

0:14:050:14:08

-Yeah, so in that case, photos seem to be totally essential.

-Yes.

0:14:080:14:13

So, for women searching for men, they do care about attractiveness,

0:14:130:14:16

physical attractiveness,

0:14:160:14:18

but they also care about how fun and outgoing you are.

0:14:180:14:21

How warm and approachable you are.

0:14:210:14:23

So, if your goal is to try and get as many women interested in you as possible,

0:14:230:14:28

then looking more fun and outgoing is going to boost your chances

0:14:280:14:31

separately from looking more physically attractive.

0:14:310:14:34

OK.

0:14:340:14:36

To see if my face is generally perceived as fun and outgoing,

0:14:360:14:40

Chris has mapped my photo onto a 3-D model of a generic head.

0:14:400:14:44

-And, there we go.

-Oh!

-It doesn't do hair, so...

0:14:450:14:49

Looking at me there, I think I don't look competent,

0:14:490:14:52

I don't think I look particularly trustworthy

0:14:520:14:54

and I certainly don't think I look likeable.

0:14:540:14:57

With no hair and the computer's identification mark stamped on my forehead,

0:14:570:15:01

I think I look more like a cage fighter than a potential lover.

0:15:010:15:04

But what I think is irrelevant.

0:15:040:15:06

The computer program engineered by Chris's colleague Alex Todorov

0:15:070:15:11

combines the collective opinions of hundreds of people comparing

0:15:110:15:14

thousands of different faces.

0:15:140:15:17

-The model tells us your face looks likeable.

-OK.

0:15:170:15:20

-You do also look trustworthy.

-Oh!

0:15:200:15:23

In terms of competence, you have a fairly competent-looking face.

0:15:230:15:27

-Oh, wow! OK. Great.

-Yes.

0:15:270:15:29

What about the other things, then?

0:15:290:15:32

Fairly extroverted and quite dominant at the same time.

0:15:320:15:35

So, I think you have a face that's good for dating and job interviews,

0:15:350:15:38

which is great. I think most people, usually, it's one or the other.

0:15:380:15:41

It's very odd seeing this, because what I'm forgetting

0:15:410:15:44

is that these are not my character traits.

0:15:440:15:46

-You're not telling me about my personality.

-No.

0:15:460:15:49

You're simply telling me what a bunch of people would say

0:15:490:15:55

about my character traits glimpsing my face.

0:15:550:15:58

Yes, our faces say a lot.

0:15:580:16:00

And Chris has a way of showing me

0:16:000:16:02

how to look more likeable and trustworthy.

0:16:020:16:04

-Again, this can change my face.

-Yes.

-Do it, do it.

0:16:040:16:08

-OK, so which one do you want first?

-Make me more likeable.

0:16:080:16:12

-So, if I ramp up your likeability...

-I get thinner!

0:16:120:16:15

-And you're smiling more.

-Yeah, OK.

0:16:150:16:18

So smiling is a simple and easy tip.

0:16:180:16:21

-So, my eyebrows are a little further apart...

-Yes.

0:16:210:16:24

-..and I'm slightly more smiley.

-Yes.

0:16:240:16:27

And what about trustworthiness? If we increased that?

0:16:270:16:31

Dial it up.

0:16:310:16:33

See, that would be hard to do without plastic surgery.

0:16:330:16:36

The BBC are refusing to pay for plastic surgery,

0:16:380:16:41

so the only alternative is for me to try and take a more fun,

0:16:410:16:45

outgoing profile picture to boost my online appeal.

0:16:450:16:48

I have to say, I'm cringing right now, but if this is what it takes,

0:16:500:16:54

I guess I'll give it a go.

0:16:540:16:56

Someone once told me you have to cough out a laugh, so...

0:16:570:17:00

HE COUGHS

0:17:000:17:01

Yes, Xand. Simply smiling can help you look more fun and outgoing,

0:17:050:17:10

but choosing your own profile picture can be counterintuitive.

0:17:100:17:15

When you look through online dating websites' data,

0:17:150:17:19

it says that actually being different is the thing that counts.

0:17:190:17:23

Dividing opinion is much better than just having everybody think

0:17:230:17:27

that you're generically attractive.

0:17:270:17:30

But the trouble is, when it comes to being objective about yourself,

0:17:300:17:33

it's easier said than done.

0:17:330:17:35

I've asked photographer Scott Chasserot to demonstrate this.

0:17:370:17:42

Scott's going to take me through the process, which includes

0:17:420:17:45

having my portrait taken with no make-up or accessories,

0:17:450:17:50

which I'm not going to lie, I feel quite nervous about,

0:17:500:17:53

but never let your vanity get in the way of doing a good job!

0:17:530:17:58

So, here goes.

0:17:580:18:01

My face has to be completely clear of both make-up...

0:18:010:18:04

# Girls on film... #

0:18:040:18:06

..and hair.

0:18:060:18:08

# ..Girls on film... #

0:18:080:18:09

-These bits?

-Yeah.

-God, you're like the blooming hair police.

0:18:090:18:11

# ..Girls on film... #

0:18:110:18:13

It's quite important to keep a straight neck.

0:18:130:18:15

# ..Girls on film. #

0:18:150:18:17

Scott's going to manipulate my picture and show me

0:18:190:18:22

lots of different versions to compare what I think I think

0:18:220:18:26

is the most attractive, with my brain's reaction before I've

0:18:260:18:30

had time to consciously think about it.

0:18:300:18:33

An EEG monitor will measure electrical activity

0:18:330:18:37

in my cerebral cortex as I see each image.

0:18:370:18:40

But this consumer headset isn't foolproof

0:18:400:18:43

because facial expressions will give a false reading.

0:18:430:18:47

I mean, you can try it, if you smile... Look at that!

0:18:470:18:50

SHE LAUGHS

0:18:500:18:51

Yeah. Any movement, I can see it.

0:18:510:18:53

So, it would be very hard to separate that,

0:18:530:18:55

-tease that apart from what's actually going on in the brain.

-Mm-hm.

0:18:550:18:58

So, this fun test requires my best poker face.

0:18:580:19:02

It's so weird!

0:19:050:19:07

But Scott's playing dirty, because as well as the original image,

0:19:090:19:13

he's shuffled in five versions of me that have been modified

0:19:130:19:16

according to theories about femininity, facial symmetry

0:19:160:19:20

and skin tone,

0:19:200:19:22

and five less attractive versions of me.

0:19:220:19:25

There are some that are really horrible.

0:19:250:19:28

By the end of the test,

0:19:300:19:31

I'm pretty confident about which version of myself I like the best.

0:19:310:19:35

I wouldn't mind being her!

0:19:380:19:41

But it's not until we get the results of the EEG back

0:19:430:19:46

from Scott's colleague in New York University

0:19:460:19:49

that we can see what my pre-conscious brain found

0:19:490:19:53

the most striking.

0:19:530:19:54

The results have shown that you've had a strongest reaction

0:19:540:19:58

to the sixth of those 11 images. So, that's one towards...

0:19:580:20:02

-Beauty.

-To beauty, yeah.

0:20:020:20:04

-And can it see what image this one is, though?

-Yeah. So...

0:20:040:20:08

That is not the one I was expecting you to show.

0:20:080:20:11

-I don't even really like that picture.

-Right.

0:20:110:20:15

See, I think that looks like I'm on a two-week holiday in Magaluf

0:20:150:20:18

and have spent a bit too much money on fake tan.

0:20:180:20:21

Which one were you expecting to see?

0:20:210:20:23

I thought it would be the one where you'd done...

0:20:230:20:27

you'd changed everything.

0:20:270:20:29

Where I had still quite dark eyes, but I had a smaller jaw,

0:20:290:20:33

thinner face, thinner nose,

0:20:330:20:35

you'd pinned my ears back, you'd made my neck longer.

0:20:350:20:38

Yeah. Yeah, that was the one.

0:20:380:20:39

That was certainly the one that I picked out as my favourite.

0:20:390:20:42

Right, your verbal choice was that one, yeah.

0:20:420:20:46

I consciously preferred the picture

0:20:460:20:48

that has been made much more feminine,

0:20:480:20:50

but my pre-conscious brain paid more attention to the image where only my

0:20:500:20:54

skin tone had changed.

0:20:540:20:57

A scientific study has shown that this carotenoid skin tone

0:20:570:21:01

is considered more attractive than pale skin,

0:21:010:21:04

yet my rational brain discarded this image. This shows why it can be

0:21:040:21:09

difficult to pick the best picture to represent ourselves online.

0:21:090:21:14

And I think the real lesson there is that when it comes to choosing

0:21:140:21:17

an online dating profile picture,

0:21:170:21:18

you should really get your friends to help you.

0:21:180:21:21

If my new picture is working, I should be more attractive to women.

0:21:240:21:28

But I don't want to spend even more time than I already do

0:21:280:21:31

swiping through potential dates.

0:21:310:21:34

Vancouver-based software engineer Justin Long came up with

0:21:340:21:38

a hi-tech solution to the problem of time-consuming searching.

0:21:380:21:41

I created an application that helps you automate everything on Tinder.

0:21:410:21:45

Using a computer program, or bot, to meet girls just sounds

0:21:460:21:50

a little dubious.

0:21:500:21:51

So, I've arranged an online chat with Justin.

0:21:510:21:54

I realised that after using Tinder for a while,

0:21:550:21:57

it became a situation where, you know,

0:21:570:21:59

I was using a lot of my time to swipe left and right

0:21:590:22:03

and it would be up to an hour or more a day...

0:22:030:22:05

And that's what I found I was doing. I think an hour a day

0:22:050:22:08

-is almost a conservative estimate.

-Absolutely.

0:22:080:22:11

And the same with my friends.

0:22:110:22:12

If we were all out at a bar, or we were out at dinner,

0:22:120:22:15

everyone would be literally swiping left or right on their phones

0:22:150:22:19

and I figured, well, why not build a bot that actually automates this?

0:22:190:22:22

So, how does the bot work?

0:22:220:22:24

It looks at the facial structure of the person

0:22:240:22:28

and it's building a computer model behind the scenes of what

0:22:280:22:31

those people look like and it compares that facial structure

0:22:310:22:35

to the differences between other facial structures.

0:22:350:22:37

Because when you're swiping left and right, you're actually telling it

0:22:370:22:41

this is who I find attractive

0:22:410:22:43

and this is who I don't find attractive.

0:22:430:22:45

So, I've been using this bot that you built and now it knows what I like,

0:22:450:22:48

-it can make those decisions for me and I don't need to be involved any more.

-Yes, that's right.

0:22:480:22:52

So, what's the next step?

0:22:520:22:53

This is actually where the bot gets more interesting.

0:22:530:22:56

And it actually saves you more time.

0:22:560:22:58

What you can do is you can customise the bot where you type in

0:22:580:23:02

your own messages to introduce yourself

0:23:020:23:05

and if they keep responding positively to your introductory messages,

0:23:050:23:10

you will then get a notification on your computer saying,

0:23:100:23:13

hey, this person's interested, you need to talk to them.

0:23:130:23:16

It's funny, I have this vague discomfort with the machine doing it all.

0:23:160:23:20

Did the people that you were talking to know that it was a bot?

0:23:200:23:23

No-one ever figured it out because even though you're having a bot

0:23:230:23:27

setting up the introduction,

0:23:270:23:29

you've still written the introduction yourself.

0:23:290:23:31

So, it's still you.

0:23:310:23:32

OK. All right. Look, thanks very much indeed, Justin.

0:23:320:23:35

-Not a problem.

-Take care, bye.

0:23:350:23:37

I'm not sure getting a machine to choose me a date on the basis

0:23:420:23:45

of looks alone is going to be any more successful than my last date.

0:23:450:23:49

I'm about to go on my first Tinder bot date

0:23:490:23:52

and I don't know if it was more efficient than organising it myself,

0:23:520:23:55

but maybe it'll be a better date.

0:23:550:23:57

Maybe she'll just think I'm weirdo for using a robot.

0:23:570:23:59

Did you know you were talking to a computer?

0:23:590:24:02

Did that even cross your mind?

0:24:020:24:04

HE LAUGHS

0:24:080:24:10

-That is exactly what I was doing.

-Yeah.

-That's not cool, is it?

0:24:150:24:19

No-one wants to feel like you're one of hundreds.

0:24:190:24:21

'So, Tinder bot makes me look like a player and while I enjoyed the date,

0:24:240:24:28

'she thought I was a bit odd and we didn't click. Again.'

0:24:280:24:32

So far, my dates chosen randomly by looks alone haven't worked out.

0:24:320:24:37

So, I'm dumping the swipe apps

0:24:370:24:39

and I'm going to pay more attention to written profiles.

0:24:390:24:43

I think it's definitely more important what a person has written

0:24:470:24:50

about themselves than what they look like.

0:24:500:24:53

The profile takes quite a while, it's what sells you,

0:24:530:24:56

so you don't want to scrimp and save on it.

0:24:560:24:58

You can tell somebody's personality, from the way that they write.

0:24:580:25:01

And that can be quite nerve-racking in itself.

0:25:010:25:04

It's so important that you separate yourself

0:25:040:25:06

and make yourself definable.

0:25:060:25:09

Hannah's given me early access to the Horizon dating website.

0:25:100:25:14

It requires a detailed written profile,

0:25:140:25:17

but how should I describe myself to attract the most people online?

0:25:170:25:20

I'm seeking help from Professor Khalid Khan.

0:25:230:25:26

He analysed nearly 4,000 scientific papers to find the best method

0:25:260:25:30

of optimising an online dating profile

0:25:300:25:33

and published the results in a prestigious medical journal.

0:25:330:25:37

His friend and co-author Sameer Chaudry was the first to benefit.

0:25:370:25:41

Previously, seven years single, he had registered with four

0:25:410:25:44

different sites and went on hundreds of dates.

0:25:440:25:47

And then he applied the Khan technique.

0:25:480:25:51

And so what outcome did you get?

0:25:510:25:53

Within three dates, he was able to then find a partner with whom

0:25:530:25:58

-he is still in partnership for the last four years.

-Wow! OK.

0:25:580:26:03

So, this all seems very good for me.

0:26:030:26:05

Horizon have built this online experimental dating site.

0:26:050:26:09

-Can you help me fill it out?

-Sure. Let's have a go.

-OK.

0:26:090:26:12

The first thing we've got to do is choose a username.

0:26:120:26:16

Everyone calls me Xand so, Xand seems like, I don't know,

0:26:160:26:18

XandVT would be kind of what I'd go towards.

0:26:180:26:21

-Do we have evidence about how to choose a username?

-Yes.

0:26:210:26:24

I would propose that you consider something that is closer to the top

0:26:240:26:28

-of the alphabet.

-Really?

0:26:280:26:30

Does that actually make a difference,

0:26:300:26:32

whether or not it's an A or an X?

0:26:320:26:34

It's more or less like the Yellow Pages effect,

0:26:340:26:37

in that the traders listed at the top of the alphabet

0:26:370:26:41

tend to receive more calls for their business

0:26:410:26:44

-than those at the bottom of the alphabet.

-OK.

0:26:440:26:46

So, the next bit of the site says tell us a bit about yourself,

0:26:460:26:49

sell yourself in less than 250 words.

0:26:490:26:52

And this is my nightmare.

0:26:520:26:54

I think this is one of the big barriers to online dating,

0:26:540:26:56

because no-one knows what to put.

0:26:560:26:58

You don't want to be boastful, you want to be humble, but you need to sell yourself -

0:26:580:27:01

it's like this impossible balance of stuff.

0:27:010:27:04

So, what do women look for in this kind of thing?

0:27:040:27:06

So, the important thing to understand is that, in general,

0:27:060:27:09

women prefer that men demonstrate courageousness,

0:27:090:27:13

they prefer the ability to take risks

0:27:130:27:16

and they don't particularly like submissiveness or kindness.

0:27:160:27:23

-So, just being a nice guy is not necessarily the best pitch.

-Really?!

0:27:230:27:28

-So, nice guys finish last, basically.

-That is correct.

0:27:280:27:31

I mean, that's a bit depressing, isn't it? This is my problem.

0:27:330:27:37

I'm too nice!

0:27:370:27:39

It sounds almost like a cliche,

0:27:390:27:41

but Khalid's meta-analysis of other scientific studies proves that

0:27:410:27:44

in the absence of familiarity, women do prefer bravery over altruism.

0:27:440:27:49

So, I'm not trying to sell myself as a humanitarian,

0:27:500:27:53

it's better to say - I don't know, what have I done?

0:27:530:27:57

Like, I've worked in war zones, you know.

0:27:570:27:59

I should emphasise that rather than saying I fed starving children.

0:27:590:28:02

-That's correct.

-And what about the tone of this? Is humour important?

0:28:020:28:06

Humour is important and it is also important to demonstrate humour

0:28:060:28:09

-without saying the words.

-Be funny, don't say you're funny.

-Yes.

-OK.

0:28:090:28:14

So, I'm going to do this now and I will let you know how I get on.

0:28:140:28:20

I wish you very good luck in making progress.

0:28:200:28:23

-I hope your outcome will be...

-As good as Sameer's?

0:28:230:28:26

As good as Sameer's, yes.

0:28:260:28:28

Xand needs to hurry up and write his profile,

0:28:300:28:33

because we're about to open the website to the public.

0:28:330:28:36

The challenge is Xand choosing a girl himself

0:28:360:28:39

versus my algorithm matching for him.

0:28:390:28:42

I think that online dating sites can match people.

0:28:420:28:45

I think it's totally possible.

0:28:450:28:47

I guess from our perspective,

0:28:470:28:49

we feel like we were matched really well.

0:28:490:28:51

98 to 99% match.

0:28:510:28:54

It's just maths. Maths got us together.

0:28:540:28:56

-We're definitely pro-algorithm!

-Yeah.

0:28:560:28:59

Just like commercial websites,

0:29:000:29:02

the Horizon algorithm takes inputs from an extensive questionnaire.

0:29:020:29:06

Now, our website is based on three inputs in total.

0:29:060:29:09

First, it asks you a questionnaire to find out a little bit about you.

0:29:090:29:13

It also asks you what type of things you're looking for in your partner.

0:29:130:29:17

And thirdly and most importantly, it allows you to rate how important

0:29:170:29:22

those characteristics are in a potential date.

0:29:220:29:25

Now, this last bit is particularly important,

0:29:250:29:28

because you have to have room to set those criteria yourself,

0:29:280:29:32

rather than have a computer set them for you.

0:29:320:29:35

Now, our questionnaire has almost 300 questions,

0:29:350:29:39

so it should give us a really rich understanding of both the people

0:29:390:29:42

that are signing up and how good they'll be for Xand to date.

0:29:420:29:46

But the problem is, is that ultimately a lot of people

0:29:460:29:49

just don't really know what they want until they find it

0:29:490:29:52

and I'm slightly concerned that that might be Xand's problem.

0:29:520:29:56

Maybe not knowing what I should be looking for

0:30:000:30:02

has always been my problem. I've been on loads of dates

0:30:020:30:05

but I'm still no closer to finding a soulmate.

0:30:050:30:08

I'm on top of the Empire State Building.

0:30:130:30:15

Now, this is meant to be

0:30:150:30:16

one of the most romantic locations in the world.

0:30:160:30:19

King Kong met his end here looking for love.

0:30:190:30:22

Tom Hanks did a little better.

0:30:220:30:24

We know that a New York minute

0:30:240:30:25

can be crammed with sex in the city, but what about love?

0:30:250:30:29

Well, they say that if you can make it here, you can make it anywhere,

0:30:290:30:32

but in five years, it hasn't worked out for me.

0:30:320:30:35

Did Frank Sinatra set the bar just a little bit too high?

0:30:350:30:38

Why haven't I found love?

0:30:380:30:40

And why do we fall in love with some people and not others?

0:30:400:30:43

One online dating company was keen to find the answer,

0:30:450:30:48

so approached a scientist who has spent 30 years studying love

0:30:480:30:51

and attraction, Dr Helen Fisher.

0:30:510:30:54

Helen is an expert on what's happening in your brain when you're in love.

0:30:540:30:58

She's now taking this one step further,

0:30:580:31:00

claiming she can match people using a personality questionnaire.

0:31:000:31:04

But as she hasn't yet published her scientific paper,

0:31:040:31:07

I'm curious to find out how it works.

0:31:070:31:10

So, how did you come up with the questionnaire?

0:31:100:31:12

I had studied dopamine in the brain. So, I pulled out a sheet of paper

0:31:120:31:15

and I wrote down "dopamine" at the top of the paper and I listed all

0:31:150:31:18

of the traits that are linked with the dopamine system in the brain.

0:31:180:31:22

Being curious, creative, spontaneous, energetic,

0:31:220:31:25

risk-taking, novelty-seeking, mentally flexible.

0:31:250:31:30

I saw that list and I said,

0:31:300:31:32

"Well, you also know something about serotonin system in the brain,"

0:31:320:31:35

and so I wrote "serotonin" on another sheet of paper

0:31:350:31:37

and I listed the traits linked with the serotonin system.

0:31:370:31:40

Being traditional, conventional, following the rules,

0:31:400:31:43

respecting authority.

0:31:430:31:45

And I had written a book on gender differences in the brain,

0:31:450:31:48

so I knew the traits linked with the testosterone system

0:31:480:31:50

and the oestrogen system, so I said, "I'm going to make a questionnaire

0:31:500:31:54

"to see the degree to which you express the traits

0:31:540:31:57

"linked with all four of these brain systems,

0:31:570:32:01

"and then watch on this dating site who's naturally drawn to whom."

0:32:010:32:05

Oh, wow! So, what happens?

0:32:050:32:07

Well, as it turns out,

0:32:070:32:09

people who are very expressive of the dopamine system -

0:32:090:32:12

I call them explorers -

0:32:120:32:13

they're naturally drawn to people like themselves.

0:32:130:32:16

People who are very expressive in the serotonin system -

0:32:160:32:18

I call them builders - they tend to be...

0:32:180:32:20

Traditional goes for traditional.

0:32:200:32:22

Traditional people want traditional people.

0:32:220:32:24

In those two cases, similarity attracts.

0:32:240:32:27

In the other two cases, opposites attract.

0:32:270:32:29

The high testosterone - analytical, logical, direct, decisive -

0:32:290:32:32

goes for the high oestrogen - empathetic, emotionally expressive,

0:32:320:32:36

good with people. So those two types go for their opposite.

0:32:360:32:40

Helen collected data from 40,000 people on the dating website

0:32:410:32:45

and although she hasn't published it yet, she says it proves her theory.

0:32:450:32:49

Testosterone-driven directors were drawn to

0:32:490:32:52

oestrogen-driven negotiators.

0:32:520:32:54

But serotonin-driven builders and dopamine-driven explorers

0:32:540:32:58

got on best with personality types like themselves.

0:32:580:33:03

So, if I take your personality questionnaire, you should be able to

0:33:030:33:06

figure out a lot more about who I am likely to be attracted to.

0:33:060:33:11

Yes. Absolutely.

0:33:110:33:13

Without seeing the data myself, it's difficult to know if it stands up,

0:33:140:33:18

but I'd like to test Helen's theory

0:33:180:33:20

with help from my married twin brother.

0:33:200:33:23

Our personalities are quite similar, but they're not identical, are they?

0:33:230:33:27

Why are you talking about this? Do the test and then you'll find out.

0:33:270:33:30

That's the difference!

0:33:300:33:32

Will Chris and his wife fit Helen's theory?

0:33:340:33:36

If Chris and I are the same, then maybe his wife's personality

0:33:360:33:39

might be a good indicator of what I should be looking for.

0:33:390:33:42

I can change my mind easily. Yes.

0:33:420:33:44

No!

0:33:440:33:45

Yes.

0:33:450:33:47

Will you do the test?!

0:33:470:33:48

Both Helen and her research associate, neurologist Lucy Brown,

0:33:510:33:55

have been looking at our questionnaires

0:33:550:33:58

and I've come to see what they think.

0:33:580:34:00

You were just a little bit more of an explorer than he.

0:34:000:34:03

He's a little bit more expressive, actually, of the oestrogen system

0:34:030:34:06

and you're just a tiny little bit more expressive of the dopamine system.

0:34:060:34:10

But you're very, very similar.

0:34:100:34:12

So, biologically speaking,

0:34:120:34:14

the woman that you end up falling in love with, she's going to have some

0:34:140:34:17

of these basic personality traits that your brother's wife also has.

0:34:170:34:21

Wow! OK, OK.

0:34:210:34:23

So, according to the personality questionnaire,

0:34:230:34:26

what kind of people should I be looking for?

0:34:260:34:28

Well, certainly other explorers like yourself.

0:34:280:34:30

Otherwise, I think that you'll get bored.

0:34:300:34:33

Helen's theory about the personality matching runs true for Chris

0:34:350:34:38

and his wife Dinah. She, like him, is an explorer.

0:34:380:34:41

But how deep is their love?

0:34:410:34:43

Thanks to neuroscience, we should be able to see if the love is there.

0:34:430:34:49

We can now scan people and find out whether they're really in love.

0:34:490:34:52

In fact, we have had the experience of someone about to get married,

0:34:520:34:57

and I didn't see anything

0:34:570:34:59

and it was someone I knew and I didn't know really what to say.

0:34:590:35:02

-She was divorced within a year.

-HE GASPS

0:35:020:35:05

-Really?!

-Yes.

-Yeah.

-Oh!

-Yeah.

0:35:050:35:08

There's a bit of "I told you so" there. Yeah, OK.

0:35:080:35:13

In an act of brotherly love, Chris has agreed to put his marriage

0:35:140:35:17

under the microscope - or rather, the MRI scanner -

0:35:170:35:21

to find out how much he loves his explorer wife.

0:35:210:35:24

That's a nice picture of Dinah you've chosen. Or did she choose it?

0:35:290:35:32

It's a picture that Dinah chose!

0:35:320:35:33

HE LAUGHS

0:35:330:35:35

-OK, it's starting now.

-OK.

0:35:350:35:37

OK.

0:35:410:35:43

And as I'm not in love, I'm the control.

0:35:430:35:47

I'm looking at an ex-girlfriend from nine years ago.

0:35:470:35:50

I shan't identify her as she's moved on in her life.

0:35:500:35:54

But as I'm still single, I'm starting to wonder

0:35:540:35:56

if my brain is equipped for romance.

0:35:560:35:59

What do you find when you scan Chris's brain versus my brain?

0:35:590:36:02

So, first let's look at Chris's brain here.

0:36:020:36:05

So, here's the brainstem and there's the ventral tegmental area.

0:36:050:36:09

It's red and yellow.

0:36:090:36:10

And he showed up a robust activation

0:36:100:36:15

in that area in response to his wife.

0:36:150:36:18

I love the idea that Chris says he's madly in love and you say,

0:36:180:36:21

"Yes, he has a robust activation in his ventral tegmental area."

0:36:210:36:26

Like that's great, that's proper science, isn't it?

0:36:260:36:29

-He's madly in love with his wife.

-So, he is madly in love.

0:36:290:36:31

Good, they're not just putting on a good show.

0:36:310:36:33

There's another sweet thing here.

0:36:330:36:35

This is called the dorsolateral prefrontal cortex,

0:36:350:36:38

but it's a very cognitive area.

0:36:380:36:40

I mean, this is an area of the brain that you use to do higher-level

0:36:400:36:44

cognitive things, like spatial calculations, that kind of thing.

0:36:440:36:48

And even reasoning.

0:36:480:36:49

And reasoning. Reasoning is a great way to summarise it.

0:36:490:36:53

And he's shutting it down.

0:36:530:36:56

-He's literally not thinking.

-Yeah.

0:36:560:36:59

Chris is also deactivating an area involved in social judgment -

0:36:590:37:03

in other words, he's not being critical of his wife.

0:37:030:37:06

So, if you just lived with a friend, them leaving the toilet seat up

0:37:060:37:10

would drive you crazy,

0:37:100:37:11

but you forgive it in the person you're in love with.

0:37:110:37:13

You overlook the negative.

0:37:130:37:15

And I would guess that it evolved for a very important purpose.

0:37:150:37:19

I mean, this is going to be your breeding partner, you're going

0:37:190:37:22

to spend years trying to raise some DNA together

0:37:220:37:25

and it would be very adaptive to be able to overlook

0:37:250:37:29

the toilet seat issue.

0:37:290:37:31

So, if you are with someone that you know really likes you,

0:37:310:37:34

you can pretty much behave as you want?! Is that...

0:37:340:37:37

That's what I'm taking away from this! Great.

0:37:370:37:41

-I hadn't thought of that.

-No, me neither.

0:37:410:37:44

So, Chris has really got the full package here, hasn't he?

0:37:440:37:47

So, he's got intense romantic love

0:37:470:37:49

and he's suppressing negative thoughts and he's also foolish.

0:37:490:37:53

He's suppressing his thoughts in general.

0:37:530:37:55

Like, he's just a love fool with her.

0:37:550:37:57

For you, you were looking at a former girlfriend

0:37:570:38:00

and you also show that suspension of negative judgment,

0:38:000:38:05

ability to overlook some of her faults.

0:38:050:38:09

She is the latest, flakiest person I know and I forgive her every time.

0:38:090:38:14

Oh! That's great!

0:38:140:38:17

No, no. It's true, it's true, it's true.

0:38:170:38:20

In anyone else, it would drive me crazy,

0:38:200:38:22

but I'm enormously fond of her. Yeah, yeah.

0:38:220:38:24

So, it's not that my brain is simply poorly equipped for romantic love?

0:38:240:38:28

No, it's well equipped.

0:38:280:38:31

So, I'm just as capable of overlooking the negative

0:38:310:38:34

as my brother, which is great news.

0:38:340:38:36

But Lucy has spotted some other brain activity.

0:38:360:38:39

This is all your somatosensory system,

0:38:390:38:42

and there's a huge red blob there.

0:38:420:38:45

As you know, there is a body map, right?

0:38:450:38:49

-Yes. The homunculus.

-The homunculus.

0:38:490:38:52

So, leg, arm, face...

0:38:520:38:54

and the genitals, I know...

0:38:540:38:56

HE LAUGHS

0:38:560:38:58

-And that became active?

-Yes.

-Really?

0:38:580:39:00

-Really?!

-Yes.

0:39:000:39:02

HE LAUGHS

0:39:020:39:05

That's very funny. So, this is very strange,

0:39:050:39:08

to sit with you looking at my brain and you're going...right there.

0:39:080:39:12

-Right.

-That's where your genitals lit up.

0:39:120:39:15

There really is no hiding how you feel from these scientists.

0:39:160:39:19

But I'm happy that my brain has the capacity for love.

0:39:190:39:24

It's renewed my enthusiasm for online dating

0:39:240:39:27

and given me an insight into

0:39:270:39:29

the personality type I should be trying to choose.

0:39:290:39:31

-Cuddling?

-Yeah, cuddling is very important.

-Everyone loves cuddling.

0:39:340:39:37

The Horizon website has been live for five weeks now

0:39:370:39:40

and there are plenty of women for Xand to choose from.

0:39:400:39:43

But will the algorithm do better than him?

0:39:430:39:46

It's matching on shared values and ideals,

0:39:460:39:49

rather than personality traits.

0:39:490:39:51

They've got a very good match at 82.

0:39:510:39:53

But we're not just looking for a girl for Xand.

0:39:530:39:57

200 men and women of all ages and orientations

0:39:570:40:00

have signed up to the site and some will be invited

0:40:000:40:03

to the event where we test the matching system.

0:40:030:40:06

-And they look pretty amenable, right?

-Mm-hm.

0:40:060:40:09

They're quite happy to get on with a lot of people.

0:40:090:40:11

Because it's a relatively small sample size, our algorithm

0:40:110:40:15

only matches one couple at 85%.

0:40:150:40:19

But when we reduce the match percentage,

0:40:190:40:21

more and more people become compatible with each other.

0:40:210:40:24

So, we know that this is a group of people who are quite intelligent,

0:40:240:40:28

who like science - goes without saying -

0:40:280:40:30

but find geekiness quite sexy.

0:40:300:40:33

This is a group of people who are quite romantic,

0:40:330:40:36

who are looking for a long-term relationships,

0:40:360:40:38

rather than just hook-ups,

0:40:380:40:39

but they also like beer and are very open to new ideas.

0:40:390:40:44

So, generally a good bunch, I think.

0:40:440:40:47

It's boding well for finding a date for Xand,

0:40:470:40:49

but I want to dig into some of the things I know are important to him.

0:40:490:40:53

What about the cats question, though?

0:40:530:40:55

Because I know cats is very important for Xand.

0:40:550:40:57

So, the algorithm has given us a shortlist.

0:40:570:41:01

We think that there are two, maybe three matches

0:41:010:41:05

that we can set up for Xand of potential dates.

0:41:050:41:08

But do all daters believe these algorithms work?

0:41:100:41:14

The matching's easy, it's the chemistry bit which is the hard bit.

0:41:140:41:18

I don't think it can just boil down to a computer program that decides

0:41:180:41:21

whether you get on with somebody or not.

0:41:210:41:23

I like to think that love is much more impulsive and spontaneous

0:41:230:41:26

than science can ever create.

0:41:260:41:27

Lots of people tell you that their eyes met across a crowded bar

0:41:270:41:30

and from that moment onwards they were smitten, and that's not

0:41:300:41:33

an algorithm or a spreadsheet that's telling them that.

0:41:330:41:36

That's something that if you could bottle it, you'd be extremely rich.

0:41:360:41:40

Online dating is lucrative business

0:41:410:41:44

and some websites make strong claims about their algorithm's ability

0:41:440:41:47

to find people's soulmates.

0:41:470:41:50

But social psychologist Eli Finkel doesn't believe the hype.

0:41:500:41:53

He can't prove it, because the sites haven't disclosed their algorithms,

0:41:530:41:57

but in his paper critiquing the industry, he argues that dating

0:41:570:42:00

companies haven't published any evidence to support their claims.

0:42:000:42:04

They have spent hundreds of millions of dollars telling the world that

0:42:050:42:08

there are soulmates, but it turns out that even believing that there

0:42:080:42:13

is such a thing as a soulmate tends to be destructive for relationships.

0:42:130:42:17

I wouldn't say I'm particularly

0:42:170:42:19

romantic or naive about these things,

0:42:190:42:22

but that's a bit of a body blow, I have to say,

0:42:220:42:24

-that I don't have a soulmate.

-You don't!

-OK.

0:42:240:42:27

But my assumption was that because they're getting such massive

0:42:270:42:32

quantities of data, the algorithms would do a better job than I would.

0:42:320:42:35

That they would be more sophisticated than a swiping app.

0:42:350:42:38

The truth is, there is no way that

0:42:380:42:40

these algorithms can do what they're claiming to do.

0:42:400:42:44

-Really?!

-Yes.

0:42:440:42:46

They're claiming that they can set you up with somebody who is

0:42:460:42:49

more romantically compatible with you than some other person

0:42:490:42:53

chosen more and less at random. And, yes, the scientific community

0:42:530:42:58

who don't have a horse in the race, there's a pretty wide consensus

0:42:580:43:01

here is that none of these algorithms can succeed in that task.

0:43:010:43:05

If these online dating sites conducted one rigorous,

0:43:050:43:09

compelling study showing nothing but increased satisfaction

0:43:090:43:13

following a first date, I would already take back my words.

0:43:130:43:17

But having sifted through the psychological literature, Eli thinks

0:43:220:43:26

there's one area where algorithms can work.

0:43:260:43:29

People who are highly neurotic are in fact, on average,

0:43:290:43:33

not very good relationship partners.

0:43:330:43:35

They have way more conflict and they're much more difficult.

0:43:350:43:38

Do I think eHarmony any can figure out who's neurotic? Yes.

0:43:380:43:41

It's not a hard thing to do.

0:43:410:43:43

And the reason why I believe that eHarmony actually does assess

0:43:430:43:47

that stuff and use it in their algorithm

0:43:470:43:49

is they frequently tell people we don't have anybody for you.

0:43:490:43:53

And my guess is what they're saying is,

0:43:530:43:56

you're a lousy relationship partner

0:43:560:43:59

-and we're not going to take your money.

-Wow!

0:43:590:44:03

Because we think you will pollute our pool.

0:44:030:44:05

So, even though dating sites can't match you with your soulmate,

0:44:050:44:09

at least they can sift out undesirables. But is that enough?

0:44:090:44:13

I often get presented as like a scold,

0:44:130:44:16

somebody who's taking the industry to task,

0:44:160:44:18

and I suppose that's fair enough, but if the question is,

0:44:180:44:22

is the overall wellbeing in the world better

0:44:220:44:25

because online dating exists?

0:44:250:44:27

-The answer is, without any qualification, yes.

-Wow! OK.

0:44:270:44:31

-Thank you very much.

-My pleasure.

0:44:310:44:33

You've got to keep you posted on how it goes.

0:44:330:44:37

So having spoken to Eli about this, I guess I feel quite conflicted.

0:44:390:44:42

On the one hand, he is sceptical about algorithms,

0:44:420:44:44

but on the other hand, he thinks that online dating -

0:44:440:44:47

and I guess dating in general - is a good thing,

0:44:470:44:49

that's how I'll get what I want.

0:44:490:44:51

So, now it's time for me to pick my date from the Horizon dating site.

0:44:510:44:56

Helen Fisher and Lucy Brown have told me I should be looking

0:44:560:44:59

for someone who's adventurous like me.

0:44:590:45:02

So, I made a shortlist, I'm hoping my written profile -

0:45:020:45:06

and writing that profile was a nightmare -

0:45:060:45:08

but I'm hoping it makes me sound funny AND courageous - not easy.

0:45:080:45:12

My profile picture I'm hoping looks fun and outgoing enough

0:45:120:45:16

for people to want to meet me. So, here goes.

0:45:160:45:19

But before Xand can find out who likes him,

0:45:220:45:24

I want to do an experiment with the rest

0:45:240:45:26

of Horizon's online dating guinea pigs.

0:45:260:45:29

Of the 200 people who signed up to my dating site, we've invited 50

0:45:290:45:34

to help me test whether algorithms

0:45:340:45:37

are as ineffective as Eli Finkel believes.

0:45:370:45:40

MUSIC: Ooh La La by Goldfrapp

0:45:400:45:43

I guess when I arrived I thought I'd feel sorry for them,

0:45:460:45:49

like, "Oh, you poor old single losers!"

0:45:490:45:51

But because I'm one of them, I now look at them and go...

0:45:510:45:53

-You are a single loser!

-I know! I know,

0:45:530:45:55

-and I really admire them for doing this.

-Fortune favours the brave.

0:45:550:45:59

But first I want to determine how much of the success of online dating

0:45:590:46:04

comes down to actual matching

0:46:040:46:06

and how much is due to the power of suggestion - the placebo effect.

0:46:060:46:12

So, we'll run an experiment.

0:46:120:46:14

Four different groups, each of them slightly different.

0:46:140:46:17

Group A, over in that corner there, they are well matched

0:46:170:46:22

and we've told them that they're all well matched together.

0:46:220:46:26

These yellow wristband wearers are part of a group of people who,

0:46:260:46:30

according to the algorithm,

0:46:300:46:32

should match well to a number of different partners.

0:46:320:46:35

Group B, just behind me here, they are also very well matched,

0:46:350:46:39

but we've told them that they're a terrible match for each other.

0:46:390:46:43

It'll be interesting to see whether these pink wristband wearers

0:46:430:46:46

are more influenced by the algorithm or the power of suggestion.

0:46:460:46:52

We've also got the people who are actually badly matched.

0:46:520:46:55

So, groups C and D over here.

0:46:550:46:58

Of these badly matched daters, Group C, green,

0:46:580:47:02

were told that they were well matched.

0:47:020:47:04

But Group D, purple, were told the truth.

0:47:040:47:08

Maybe it's just that the perception of being well matched is enough

0:47:080:47:11

to trick people into thinking their dates were better than they were.

0:47:110:47:15

At this exact moment, having split them up into the different groups,

0:47:150:47:18

it doesn't look as though one group is having a better time than others.

0:47:180:47:22

After half an hour, everyone puts their wristbands

0:47:230:47:26

into a container according to how well they got on with their dates.

0:47:260:47:30

Statistically, if neither algorithm nor the power of suggestion

0:47:300:47:34

has an effect, we should find the wristbands equally distributed

0:47:340:47:39

between all three baskets.

0:47:390:47:40

So, first up in the sad basket -

0:47:400:47:43

the people who did not have a very good time -

0:47:430:47:45

what is interesting is that there are only pink and purple in here.

0:47:450:47:50

Pink and purple were the two groups that were told that they were

0:47:500:47:54

-not well matched with each other.

-Power of suggestion.

0:47:540:47:57

Yeah, implying that people who were expecting to have a bad time

0:47:570:48:00

really did have a bad time.

0:48:000:48:02

Nobody who was told that they were well matched ended up

0:48:020:48:04

not having a good time. It's extraordinary. Suggestion works.

0:48:040:48:08

There's a hint there. OK.

0:48:080:48:10

But if you look at the basket where people had a really good time,

0:48:100:48:14

the biggest number of wristbands in here are the yellow wristbands,

0:48:140:48:19

which is where people were told they were going to have a good time,

0:48:190:48:24

but also, the algorithm said that they would be well matched.

0:48:240:48:27

-That's the highest number in here.

-That's really good.

0:48:270:48:30

So, your algorithm did make people have a good time,

0:48:300:48:33

-even beyond what we told them.

-Yeah, absolutely.

0:48:330:48:36

I think, generally, what this is saying is that both things

0:48:360:48:39

make a difference -

0:48:390:48:40

both what you tell people - this power of suggestion -

0:48:400:48:42

but also that the algorithm seems to have some effect.

0:48:420:48:45

I think it's really funny that you're surprised by this.

0:48:450:48:48

Well, yeah. We have to be scientists here, in that this is...

0:48:480:48:53

You do, I'm just looking for love!

0:48:530:48:55

SHE LAUGHS

0:48:550:48:57

As there were equal numbers of the two groups who were duped

0:48:570:48:59

- green and pink -

0:48:590:49:01

we'll have to call it a draw between algorithm and placebo.

0:49:010:49:05

Our little test mirrors the results of a much larger online experiment

0:49:050:49:10

by OkCupid.

0:49:100:49:12

'Now it's time to introduce our daters to their best matches...'

0:49:130:49:17

-Recovering-Cyclist...

-With Miss...

0:49:170:49:20

'..to see if Hannah's algorithm can find love among our volunteers.'

0:49:200:49:26

-Adman91.

-You're with Babe2.

0:49:260:49:30

'If successful, it should bode well for Xand's dates tomorrow.'

0:49:300:49:35

-Alien-Turned-Human.

-You are with Knotted-Sheep.

0:49:350:49:39

Oh, gosh. You're already standing together!

0:49:390:49:42

What a surprise!

0:49:420:49:45

That's a good sign.

0:49:450:49:47

It's a fun way of approaching topology, isn't it?

0:49:510:49:54

Yeah, absolutely.

0:49:540:49:55

Do you feel like you wrote this algorithm and you are fiddling with their lives in quite a weird way?

0:49:580:50:04

In a way, it would be very cool if some people actually got together

0:50:040:50:07

this evening and we got to go to a wedding in a couple of years' time.

0:50:070:50:10

You would get to officiate that wedding. You'd be...

0:50:100:50:14

You'd be like the guest of honour. You'd be on top table, for sure.

0:50:140:50:17

I want a little statue of me on top of their cake. That's all I'm saying.

0:50:170:50:20

THEY LAUGH

0:50:200:50:22

But to make sure everyone got paired up,

0:50:220:50:25

not every single match was perfect.

0:50:250:50:27

Well, I had an interesting conversation with my match,

0:50:270:50:30

but there was no initial attraction there, no.

0:50:300:50:33

-I think he likes someone else.

-Really?

-So, yeah.

0:50:330:50:36

She was very lovely and we had quite a good conversation,

0:50:360:50:39

but I don't think we had a whole lot in common.

0:50:390:50:42

But most seemed quite happy with their match up.

0:50:420:50:44

She lives in Bristol, which is where I'm originally from.

0:50:440:50:47

I do go back to Bristol quite a bit. So, who knows?

0:50:470:50:51

He's quite funny and do you know what, if a guy makes me laugh, I'm putty in their hand. So, yeah.

0:50:510:50:56

We found out we actually had a lot in common.

0:50:560:50:59

And actually we're getting on really well right now.

0:50:590:51:01

So, I guess there might be something to this algorithm.

0:51:010:51:04

But what about the algorithm's very best-matched couple, at 85%?

0:51:040:51:10

-Have you found that you've got quite a lot in common, then?

-Yes.

-Yes.

0:51:100:51:14

It was not a surprise that we were matched up. We've just been talking the whole day.

0:51:140:51:18

-We started talking while we were still outside.

-Do you think you'll stay in touch?

0:51:180:51:21

-Do you think you'll want to see each other again?

-I think we'll stay in touch.

-I would like to, yes.

0:51:210:51:27

I think from a small sample of people

0:51:270:51:29

over quite a short period of time, you did really well.

0:51:290:51:32

Some of them left arm in arm.

0:51:320:51:34

Yeah, yeah.

0:51:350:51:36

So, I think the algorithm will have picked out some ladies

0:51:360:51:41

for you tomorrow that you'll, at the very least, get on well with.

0:51:410:51:45

All right, we'll see. We'll see.

0:51:450:51:47

I'm looking forward to it much more now than I was before this started.

0:51:470:51:53

-I think that's a good place to start, at the very least.

-OK.

0:51:530:51:58

Finally, I'm going to meet my dates.

0:52:010:52:04

-Very nice to meet you.

-Nice to meet you.

0:52:040:52:07

The first girl I'm meeting is one I picked using just her photo

0:52:070:52:10

and profile information.

0:52:100:52:13

And since the whole challenge is riding on these dates,

0:52:140:52:17

I'm keeping an eye on proceedings.

0:52:170:52:20

-Are you a chemist by background, then?

-I'm a chemist, yeah.

0:52:200:52:23

I did nanochemistry for my PhD.

0:52:230:52:25

I thought Cat was beautiful,

0:52:250:52:26

she's obviously very intelligent, which I really like.

0:52:260:52:29

Do you have strong opinions about national service?

0:52:290:52:32

I don't really have strong opinions on a lot of things.

0:52:320:52:34

She was really good fun.

0:52:340:52:35

Oh, she's Iranian. Great, I'll definitely bring that up.

0:52:350:52:39

That can be fed in somehow.

0:52:390:52:40

I think she quite fancies Xand.

0:52:420:52:44

Yeah, he's cute.

0:52:440:52:46

Yeah, I really enjoyed it.

0:52:460:52:48

I don't know if I'd go on a second date and I kind of think,

0:52:500:52:53

if I don't want to go on a second date with her,

0:52:530:52:55

then what am I doing? Who would I go on a second date with?

0:52:550:52:57

I discovered I didn't like New York at New Year.

0:52:570:53:00

-Nowhere's fun at New Year.

-No.

0:53:000:53:03

So, Xand isn't as good at picking a partner as he thinks he is.

0:53:070:53:11

The next date was matched by the algorithm, according to

0:53:110:53:14

shared interests and opinions.

0:53:140:53:16

-Hello.

-Very nice to meet you. How are you doing?

-I'm good. Nice to meet you.

-Thank you for coming.

0:53:160:53:21

But will she have that special something Xand is looking for?

0:53:210:53:24

Do you have, like, do you think you've got good at dating?

0:53:240:53:28

Erm...I don't know, I'm not sure.

0:53:280:53:31

-That's a really unfair question, I think you're good at dating!

-Yeah, thanks. Yeah.

0:53:310:53:35

I liked her, I thought she was nice, I thought she was attractive.

0:53:350:53:38

You know, it's a bit like when you interview people for a job...

0:53:380:53:41

I didn't think the algorithm did a bad job...

0:53:410:53:43

I think dates are a little bit like that. You know straight away if you're going to like someone or not.

0:53:430:53:48

Ooh, there was locked eye contact and a smile right then!

0:53:480:53:51

He could probably chat the hind legs off a donkey, I'm pretty sure.

0:53:560:54:00

But, yeah, he seems like a nice, genuine guy.

0:54:000:54:03

I think subtle personality traits meant that there wasn't a spark,

0:54:030:54:06

but I don't know, is that the algorithm's fault? I suspect not.

0:54:060:54:09

I think, I guess, if I met 50 people like her,

0:54:090:54:13

one of them might be perfect.

0:54:130:54:15

-The format of a date...

-Yeah.

-..it is a terrible idea.

0:54:150:54:18

Yeah, yeah. It is.

0:54:180:54:19

Having watched both Xand's dates,

0:54:190:54:22

I don't think either of us have won the bet.

0:54:220:54:25

But there's one more date left.

0:54:250:54:27

How are you doing? Oh, really nice to meet you.

0:54:300:54:33

-Come in, this way.

-Great.

0:54:330:54:35

This date appeared on both the list of good matches,

0:54:350:54:39

according to the algorithm and the list that we asked Xand to put

0:54:390:54:43

together himself of people he'd like to meet and date.

0:54:430:54:46

Because we both selected her,

0:54:460:54:48

neither of us can claim this as a win if this date goes well.

0:54:480:54:53

-Your username was Little Burp, wasn't it?

-Yes.

-Yeah.

-That's right.

0:54:530:54:56

Why was it Little Burp?

0:54:560:54:58

Well, it's a bit silly really, I've got lots of bird tattoos, so

0:54:580:55:01

for a long time my friends have called me Little Bird,

0:55:010:55:04

which is a bit naff, so eventually that became Little Burp,

0:55:040:55:06

because it's funnier and more appropriate.

0:55:060:55:09

From behind, that basically could be my head!

0:55:090:55:13

She looks really similar to me.

0:55:130:55:16

Although, she's much cuter!

0:55:160:55:18

I feel fairly convinced that they both fancy each other.

0:55:220:55:25

I mean, Xand's very dishy.

0:55:250:55:27

I was really bored with my hometown

0:55:320:55:35

by the time I was about ten, I think.

0:55:350:55:37

I'm from Beccles, which is right on the...

0:55:370:55:39

Er, Beccles, just so you know,

0:55:390:55:41

is where my family are from, where my English family are from.

0:55:410:55:44

It's a tiny little town, it's where I'm spending Christmas!

0:55:440:55:47

I've lived in London for 13 years this week.

0:55:470:55:50

-Really? You have, like, a London anniversary.

-I do, 17th September.

0:55:500:55:53

I've lived in London for 13 years this week!

0:55:530:55:56

-I deliberately didn't spend lots of time thinking about what you might be like.

-OK.

0:55:560:56:00

I had to make the choice, so maybe I'm a bit more...

0:56:000:56:03

Maybe I'm a bit more invested.

0:56:030:56:04

-Oh, bless you!

-HE LAUGHS

0:56:040:56:06

I think the date went pretty well.

0:56:060:56:09

As soon as she arrived, I just thought she looked really great.

0:56:090:56:12

Just everything about her was really nice.

0:56:120:56:15

She has a great smile, she was really...

0:56:150:56:17

She was just someone I quite fancied.

0:56:170:56:20

He's a very, sort of, attractive character. He's very articulate.

0:56:200:56:26

I think the more you like someone,

0:56:260:56:28

the harder it is to tell what they think of you.

0:56:280:56:32

So, yeah, maybe there is something in the algorithm.

0:56:320:56:35

Are you texting someone there, Xand?

0:56:400:56:42

SHE LAUGHS

0:56:430:56:45

So, it went well, then?

0:56:450:56:47

Yes. So, my date with Cindy was great,

0:56:490:56:52

but I still have the nervousness of going, "Did she like me?"

0:56:520:56:55

So, she hasn't responded to my text message saying,

0:56:550:56:57

would she like to go for a drink?

0:56:570:56:59

You sent it about three minutes ago!

0:56:590:57:01

Well, no, but... I might check it again!

0:57:010:57:04

No, still nothing.

0:57:050:57:07

-Mate, that's still three-and-a-half minutes later!

-Yeah.

0:57:070:57:10

Very seriously, from watching it,

0:57:100:57:13

it was like a completely different date.

0:57:130:57:16

So, I think the algorithm does well,

0:57:160:57:20

but I also think what it can't do is tell you about that spark.

0:57:200:57:25

There's something extra, which you just can't define or capture,

0:57:260:57:30

and you only know it when it's put in front of you.

0:57:300:57:33

But I think that's still massively impressive.

0:57:330:57:37

It's so hard to meet people - my life is so busy,

0:57:370:57:41

my friends are so married.

0:57:410:57:43

Prior to doing this, I'd never written a profile

0:57:430:57:46

and I had never used an algorithm-driven site,

0:57:460:57:49

I'd just used a swiping app,

0:57:490:57:51

and I wouldn't bother with that any more.

0:57:510:57:54

I think online dating is just an introductory service.

0:57:540:57:57

So, it is just a numbers game.

0:57:570:57:58

I wouldn't say it's JUST a numbers game, I think it's, like,

0:57:580:58:01

the most important numbers game you can possibly play!

0:58:010:58:04

I'm still rolling the dice, right? I'm like the last guy in the casino

0:58:040:58:07

when they're trying to turn off the lights.

0:58:070:58:10

I guess what I mean is, I think the internet dating websites

0:58:100:58:15

and their algorithms do do something,

0:58:150:58:19

but I think, ultimately,

0:58:190:58:20

it doesn't guarantee that every date will be good,

0:58:200:58:23

but it gives you a good solid basis on which to build.

0:58:230:58:27

Download Subtitles

SRT

ASS