Inside OKCupid: The math of online dating – Christian Rudder

July 13, 2016 - Comment

  View full lesson: When two people join a dating website, they are matched according to shared interests and how they answer a number of personal questions. But how do sites calculate the likelihood of a successful relationship? Christian Rudder, one of the founders of popular dating site OKCupid, details the algorithm behind 'hitting it


 


View full lesson:

When two people join a dating website, they are matched according to shared interests and how they answer a number of personal questions. But how do sites calculate the likelihood of a successful relationship? Christian Rudder, one of the founders of popular dating site OKCupid, details the algorithm behind 'hitting it off.'

Lesson by Christian Rudder, animation by TED-Ed.

Comments

Dion7 says:

It’s very interesting, though it requires that a person knows what they
want in their partner. In my experience, it’s sometimes the partner who
surprises you and has a personality that you didn’t know you were attracted
to who is really good for you. Though it is a lot of fun to fill out a
questionnaire like that and see how your dates filled out theirs.

Artbug says:

+Dion7
In my experience, people don’t give too much credence to the match
percentage unless they’re really low… in which case there’s a good reason
for it. People can always update their answers if they find characteristics
in a person they find they like once they experience them (since you very
rarely come across someone who is a 100% match – unless you’ve only
answered about 5 questions out of the thousands of questions they have).

LordoftheJamesClan says:

1. I have about 4 girls to talk to in my area.

2. None of them talk to me.

Lara Orane says:

+MuffHam Yeah, that’s literally my life.

ihaveagun22 says:

+LordoftheJamesClan *sobs in defeat*

Rawr_Kitten says:

was on ok cupid for 3 years… lots of guys in my area… all wanted sex
not dateing… right when i leave okcupid and go to another site… i find
a guy..

Richard Duryea says:

+LordoftheJamesClan I have hundreds of women in my area with a +90% match
and none of them respond to me.

Lily Rose says:

I met my husband on OkCupid, we’ve been together for 3 years. We really
love each other and have a lot in common, but we also have our differences.
Just recently he told me that his friend set up his profile for him, and
answered all the questions for him randomly.

frosecold says:

+Lily Rose update, she dumped me… She was crazy and though i was cheating
on her

Artbug says:

+Lily Rose
But did you give him a chance based solely on the match percentage? …Or
something else?

dracocrusher says:

+frosecold
Jeez… why can’t people just trust and have faith in the people they’re
going to move in with?

Vdaddy says:

+frosecold Might as well cheat on her now and prove she was right.

scott- o says:

And exactly how does your algorithm handle the man from Nigeria that’s
pretending to be a woman?

Ben Wall says:

+jonas samuel Or that Nigerians pretending to be people they aren’t is a
bit of meme because internet culture is a thing.

Artbug says:

+scott- o
There’s a “Report this account” button for that.

Colin van Dam says:

+Momma llama In a Can
yes.

The Big Medley says:

+scott- o Are you sure she is not a trans woman?

jonas samuel says:

+Ben Wall
or that^^.

but then the same question could be asked for the origin of a meme, but i
would say using a (if we assume it would have) a slightly racist origin, is
not automatically racist. ^^.

Ian Korish says:

I want to create my own dating website and pair up shitty people with
equally shitty people.

Bret Bennett says:

Hi, My bf and I have recently got married, she is from Belarus and I am
from Philippines. We met in this dating site http://il0veu.cf/s1ngles4u

Michał Kowalski says:

+Ian Korish Before i read your comment i thought there should be just one
extra question: How atractive do you think you are: give it a number from 1
to 10. And ppl would be matched by this and there would be much more happy
couples as a result. What do you think about that?

Quick Silver says:

+Ian Korish lmfao! Omg dude I’m fucking rolling

alexclaton says:

+Ian Korish Isn’t that plenty of whales?

Okane says:

The math of online dating:
99% males 1% females :P.

James Reilly says:

+Dalish Elf Still a 1% 6.8 billion people makes 68 million asexual.

jonas samuel says:

+Dalish Elf maybe it helps not to throw statements like "100% of the
time" around, just allways say 99% or "mostly" it shows that
you actually think critically and don't claim absolute truth as many
people do.

(i don't say you are i just say statements like that make you seem like
on of those people.)

and lastly, the thing with the 5% is not universal, in some cases 0,5%
(often 2,5%) is a significant diversion it depends on which field of
science you are in, what kind of test you are doing (and in what magnitude)

oh and by the way, there are many happy couples who wait several months to
have sex for the first time so your statement is at least uncomplete.

Mind Chow says:

+MosukaDreamer I have a decent job and won’t ask for Sex(too busy for
that). but I’m kinda ugly. am I worthy?

jonas samuel says:

+Joe Manco
what? i would be fine if you are using actual statistics, but saying
“Woman, if the sex is bad, the relationship is bad 100% of the time.” is
either a joke, wrong generalisation or stupid, pick your poison :).

DéJi Vu says:

At this bit I almost cried…
0 = Irrelevant
1 = A little important
10 = Somewhat important
50 = Very important
250 = Mandatory

Alright, that’s just stupid. What arbitrary numbers to pick. At the end of
the day, whichever the highest number you pick is, is simply 100%, and any
other number is itself divided by the highest number (in percent). So “Very
important” here is 50 ÷ 250, which is 0.2, or 20%… whiiiich… doesn’t
seem so important to me. Clearly it should be done the intuitive way, so
that ‘A little important’ is 25%, ‘Somewhat important’ is 50% and ‘Very
Important’ is 75%. More user friendly then, especially those able to use
their brains.

Of course, to be more technical and accurate, computers don’t have
percentages built into their logic, so they instead use real numbers (which
are often approximations due to memory limits) where 20% is 0.2 and 100% is
1.

Also something I thought from the start is how do you actually compare
these at the end of the day? If someone is very tidy and they say that it’s
“very important” that their match be “moderately messy”, is that literally
going to set a precedence for the people who describe themselves as messy?
I mean what it *should* do is search for people “at least this *tidy*”, not
restrict the search *to* people that untidy, et cetera. Then you just get
to the more complicated questions where judging on being closest to your
preference isn’t fitting – there could be extra conditions where that’s
fine, as long as it’s closer enough to another preference, etc.

Of course, as someone who already figured out the advice given at the end
of this video and having become a pretty good programmer, I would never use
such a feature for the same reason I habitually exclude non-important words
from Google searches. Knowing computers gives you this burden where you
generally just don’t trust anything you feel can’t be done perfectly by a
computer, and using things like that gently starts to become second nature.

The irony is that the dream goal of programming is to overcome these
limitations with stuff like AI, so us programmers (or those with similar
logical thinking skills) will usually assume something wouldn’t work and
not even try to see if it does. So basically the more a developer
programmer does, the more user programmers will be utterly stumped by the
end product.

Similarly, I have a tendency to get overly frustrated with software, most
probably because I know how they should work is relatively simple so it’s
frustrating when I’m not noticing quick and simple front-end effects. I’m
also obviously rather anti-javascript. Those ‘parallax’ websites that
scroll in two directions at once make me sick, quite literally, and end up
skipping the actual content – sure they look awesome, but user experience
is everything, so remember that programmers (or those with similar logical
thinking skills) make up quite a good portion of your end-users (especially
good, because they’re the smart ones, unlike the majority who are maybe
better at sports or relationship building or something).

jonas samuel says:

+DéJi Vu well i think the little important to somewhat important gap with a
factor of 10 to 1 might be a bit big in comparison,

but the problem is that some things just are waaaay more important then
others to us, but i would prefere to know these numbers when i would use
such a site to evaluate myself, maybe even write a value of 0 to 255 (maybe
represented as a colour 😉 )

DéJi Vu says:

+jonas samuel Better than that, add a control to set how the value is
compared. Do you want someone more than 50% messy, or less? Do you want
someone exactly that messy? Do you want someone within 15% of either
totally messy or totally clean? Or do you want somewhere in the middle,
maybe leaning a bit more towards the clean side? Perhaps you don’t care how
messy they are as long as it’s under a certain value, or within a certain
range, but do if it’s another etc.

Just saying “somewhat messy”, “very messy”, etc. is far too vague. This is
a software which purports to being able to solve a very big life decision,
and they expect people to be okay with using such vague parameters? And
don’t get me started on the questions they do have. I was just using messy
as an example, but it appears dating sites put all the example questions
into the real site, instead of asking some stuff which is more likely to
pair people up with socially compatible others. Basically, a psychology
test… yeah, well, this shouldn’t be fun, it’s dating.

jonas samuel says:

+DéJi Vu
yeh people are to lazy for that^^

Murdy Baskan says:

+DéJi Vu This concept is called as fuzzy logic, they create a scale to
something cannot be measured. But every problem in world cannot be solved
easily.

Good partner may become hostile at their period. How can you determine
that? But it is ok to make some guess.

WIZARD Z says:

Fucking useless nobody uses it. All they care about is how you look

player276 says:

+Dalish Elf Someone has bad experience with online dating.

player276 says:

+Dalish Elf Have any data to back that up?

Aaron Richards says:

+Dalish Elf this is why we need waifus

Quick Silver says:

+WIZARD Z and that’s fucking stupid (unless of course they only want a
one-night-stand). I’ll never settle for a hot dumbass just sayin

soonsims says:

You forget one more variable. People lie. And more often than not, people
lie to themselves.

Xantiago P says:

so deep

soonsims says:

+Xantiago P tnx

Jai Verma says:

Every man, no matter how strong he is, lies to himself about something,
I’ll find your lie and I’ll break you.

soonsims says:

+Jai Verma Yes absolutely true. And that’s the biggest flaw I with these
kind of studies which rely on self-evaluation.

I'm that Gringo says:

+Jai Verma I never lie to myself…tf

Teencat says:

Annnd this is why I don’t use dating websites: the comments. They reflect
how shallow human nature is. But, congrats to those for whom the website
does work ^_^

Alex Stein says:

The problem is that people lie (social desirability bias), overestimate the
extent to which they understand their own behavioural traits
(Dunning-Kruger effect), and either ignore the numbers because they find
someone attractive (infatuation) or rely on them too much (anchoring). So
long as human error plays a role, matchmaking will be a messy affair.
It’s better to spend time with people and judge face-to-face whether or not
they make you happy, have an idea of basic traits that are desirable
(compassion, honesty, reliability), and put your apish emotions to one side
as much as possible when choosing a mate.

J.R. Scott says:

OK Cupid sucks. If you’re a girl then you will get thousands of messages
and if you’re a guy you have to look like a male model or your message will
just get lost in the mix and you will never get a date. Bullshit site.

Artbug says:

+Jason Scott
That’s not OK Cupid… That’s life!

Tayet4Buri says:

+Jason Scott Everyone just writes “hi” or “how are you” and in the worst
case scenario “how U doin?”
The one I ended up dating was the one who wrote “Let it snow…” In the
header and asked me if I liked snow, because he would love to have a coffee
date in the park tomorrow, and walk through the glistening frostyness.
It did snow that evening and because it was kind of funny, I said yes.

Alex Shor says:

+Tayet4Buri Cannot confirm… I always send unique interesting messages
like that and I hardly ever even get responses, let alone date requests.

Tayet4Buri says:

+Alex Shor Then those you write to aren’t worth your time. If they don’t
even bother they obviously aren’t deserving you. The only exception would
be if you write stuff like “wanna f#@k”, then you are the baddy.

ROBIE says:

I don’t even want a relationship but I just want to fill out this quiz and
see who comes up.

evergreen 13 says:

can you just play a little bit with this?…i’m curious…

Pika Zilla says:

I noticed a few common statements about OKCupid;

answer 1000 questions at least so that the algorithms work the best they can
be patient: don’t expect anything serious in the first weeks
don’t make a majority of your questions mandatory: that could make
realistically pointless questions skew your results
be picky: don’t date everyone above 50%. aim for 96% to 99%

Dick “Seymour” Maus says:

OkCupid is just full of people who are either a total wreck, or looking for
meaningless sex. The real people are too scattered to show up as matches.
So it’s just a pimping site.

Tanner Herzman says:

im a hopeless bastard 🙂

Xantiago P says:

all i could think during the video was that the right boob was hanging
lower than the left boob

Andrew Smith says:

How can I use OK Cupid to get casual sex? I’ve heard that it’s a great site
for casual sex. I’ve sent messages to hundreds of girls on there – the vast
majority never reply. The tiny minority who reply do so by insulting me.

Andrew Smith says:

+coolguythinking Do you get casual sex with any of them?

coolguythinking says:

+Andrew Smith I just started doing this so not yet.

Andrew Smith says:

+coolguythinking If you’ve not had any success with online dating, what do
you mean when you say that ‘online is easier’? Having a 0% success rate
isn’t good – even if you’ve not been trying for long.

coolguythinking says:

+Andrew Smith I said online is easier for me, not that it’s easier for
anyone else. If you’re ugly you should stay away from online dating and
stick to reality. Since I look decent it’s easier for me to stick with
online since I don’t have to approach a girl in the same way I would have
to in the real world. All I can do is send a message and ask for a meet up.
I’ve already been liked by several girls but right now I’m trying to score
with a really hot chick. Getting hotter girls takes more effort than
getting attention from average looking girls.

Andrew Smith says:

+coolguythinking I’m rejected by girls in person, just like I am online.

Comments are disabled for this post.