Authors: Emilio Cecconi
Michelle really has a way of undermining me. She’s right about a lot of things though. She always has been. I still can’t stop thinking about some of the things she said. Actually, I can’t believe she told me that she thought online dating was a pretty good thing.
It really surprises me how many people use online dating websites. Each website has its own audience and niche. One in four relationships in the United States now begins online. That sickens me. Online dating has been around since the beginning of the internet, except it wasn’t called “online dating” back in the 90s. It started with people meeting in IRC channels or ICQ chat rooms under aliases such as AZNL33T85 and PRINSEZ86. It was such a 90’s thing for you to spell things in improperly. It was the cool thing to do for technophiles. For those of you who are too young to remember what I’m talking about think about AIM chat rooms. I can’t even think of a present day equivalent of a chat room. Do chat rooms even exist anymore? I’m sure they do on some parts of the internet, probably in video camera format.
I forgot to mention, I know a lot about online dating. It’s quite the story why I know so much about it. I’ve never actually used it myself though. I feel like it would be cheating if I do. Here goes.
At the turn of the century some guy had the idea that you could make a lot of money if you made a website that had people profiles you could search in order to date them. This initially only appealed to the people who were extremely tech savvy and very nerdy. Why in the hell would an average person spend days searching these anonymous profiles, most of which were fake, trying to get some person’s attention when they could just go to a local bar?
Boom.
Online dating was born. Websites were started and now these places started displacing chat rooms as the de facto way to meet a person of the opposite sex on the internet. You know, things worked but people dissatisfied with the signal to noise ratio on these websites.
A few years later a group of computer scientists came along and said, “We’ll make computers match people up together automatically. Numbers don’t lie. We can map people together based on their interests and personality.” These computer scientists turned self-proclaimed love doctors believed that the world would be a better place if a computer could find your mate for you.
These people were convinced that a computer could do a better job at pairing people up than you could. They believed that trying to meet people in real life was too messy and unpredictable. How could you meet more than a few people per month? There was just one catch, how do you get the first people to an empty party?
So they devised a questionnaire filled with tons of questions. They ran ads saying we will successfully find someone for you on our dating website based on the results of our ‘proprietary algorithm.’
What they really did was match people up on a likeness scale. These scientists thought that the idea of ‘opposites attract’ was complete bullshit. So they paired people up on how alike they said they were. That was their master algorithm: date a person who is a lot like you.
The most surprising thing about this whole ordeal is that it actually worked. Online dating companies started running ads about success stories and people ended up getting married by the bundles. You know, I still don’t buy it. I believe the reason why this rudimentary pairing process worked was because the kind of people who went on these websites already pre-committed to a relationship. It’s the damn confirmation bias.
For years online dating started gaining more traction. Advertisements said that they were able to pair you up with someone you can fall in love with or your money back.
Things were all great in the online dating world until
YinYang came along in 2006. It was a dating website started by a Harvard Economist and a couple of M.I.T. computer scientists. They thought that they whole questionnaire part of the process was troubling. Why? Because they believed that people did not know themselves. They didn’t think people knew who they were or what they wanted.
For example, here’s something I would say about myself.
“Hi my name is Jake and I am a 26 year old male. I’m organized and neat. I like romantic comedies and long walks on the beach. I’m looking for a girl who could share my passion for linguistics.”
This is what
YinYang would say about my self analysis,
“Um, I don’t believe a single word you just said. I’m going to observe you for the next week to gather my own opinion on what kind of person I think you are and what kind of woman you would be into.” That’s what the
YingYang people would say about my little segment on myself.
The
YinYang website still had a questionnaire that they say maps people to their potential mates. Seems all the same right?
No. Not in the least.
YinYang only used the questionnaire as 5% of their matching algorithm. You know all those questions you answered? They really didn’t care about what you said about yourself. They only wanted you to believe that they cared about what you had to say about yourself.
What does that mean? It means that
YinYang believed people were only 5% knowledgeable about who they were and what they want. So you might ask, what’s the other 95% of the matching algorithm? How do they recommend people for you to date?
Machine learning.
Yep. If this answer sounds to you like artificial intelligence, it’s because it is. The algorithm is changing every day based on past results. Don’t worry it’s none of this
The Matrix
stuff. Machine learning is used on all the major search engines online, that’s how they “learn” what relevant ads to show you after you type a query. Applied to online dating, the results can be explosive.
You’ll notice that
YinYang doesn’t recommend ‘matches’ until after you spend 60 minutes on the website “looking around and examining its functionality.” They also tell you they do this to discourage fake accounts. If you were a fake person signing up for an account, you could make hundreds of accounts in an hour. Well that gets discouraged if you have to surf around and examine the website for yourself. But there is more than meets the eye.
What they don’t tell you? The entire time you are surfing the site they are recording so much information on you it’s scary.
What kind of internet browser do you use? Are you a Mac or a PC person? What queries are you entering into the search box? How much time do you spend on each person’s profile? How much time do you spend on your own profile? Where on the page do you click your mouse, how long do you keep the window active before you switch tabs, what time of day do you log in to, what mobile device do you check the website on. Whose profiles do you spend the most time on? What are the occupations you are searching out for potential mates? When evaluating other people’s profiles do you look at the about me section or pictures section more? The list goes on and on and on. Everything gets logged and categorized.
So okay, you might say… they have all this information. So what? I’ll tell you what. They combined forces with a couple of experts from Stanford and created what they called
LoveRank. They use very complicated algorithms to match people together.
For example, say I spend my time looking at blondes on the website who are mainly veterinarians.
YinYang will look for a blonde veterinarian who is looking for a management consultant who has a peculiar interest in the humanities like me. Then they will match us together. This of course is hypothetical as I do not use these sites. My example was also overly simplistic. The ways they actually pair people together are downright scary.
YinYang
pretty much exploded and started taking market share from all of their competitors. People go on there every day not knowing that their actions are what are being used to map them to someone else.
You may ask how I know all of these things. The year after I graduated college was when
YinYang was getting really big. Like really big. People were leaving the other major dating sites pretty rapidly to join YinYang. The other companies didn’t know what to do, except for to emulate YinYang’s matching algorithm.
My first consulting engagement was to bring the Titan of online dating sites to the
YinYang era. Yes the CEO actually called it the YinYang era.
My first order of business was to convince the CEO that this was the “Machine Learning Era.” My Firm’s partner leading this engagement kept repeating this phrase every time I’d see him on a call with clients. He made sure to include it in every PowerPoint presentation I had ever seen him do. Sorry, I didn’t mean to be esoteric – here’s a definition I pulled from Wikipedia. It’s the second time I’ve used the term.
Machine Learning, a branch of artificial intelligence, is a scientific discipline concerned with the design and development of algorithms that allow computers to evolve behaviors based on empirical data, such as from sensor data or databases. - Wikipedia
Once we got that point settled with the CEO, I began hacking away. I was paired up with their data scientists and started looking at aggregated data that they had collected over the last few years from their users. I spent months using fuzzy logic and neural networks to make inferences on how I think people should be paired up.
That’s when I met Punjab. He was this rock star from Bombay who could make computers track anything his imagination wanted. Even better, he knew how to turn my imaginative ideas into computer code. I remember staying up all night talking to him over the internet during his workday so I could get ready for presentations in the morning.
So now you think on top of everything else I told you also think I exploit outsourced labor? Not quite. Punjab lived in Bombay, but he was employed by my firm as a US worker. He went to Stanford, naturally, and has a Ph.D. in computer science. He spent a few years working R&D at my firm until he realized he really didn’t like the USA that much. He packed his bags ready to go back to India looking for professorships there. A few of the Partners at my firm flipped. After a little back and forth, they agreed to pay Punjab US salary plus true up taxes and relocation to India. He works remote for us from India now. I think he made out pretty well on his decision.
In four months’ time we implemented what I called
Genesis
at this online dating website. Our goal was to pretty much emulate the algorithms that YinYang was using to pair people together. I think we accomplished that and more. You know, I had some crazy ideas on how people should be connected.
Punjab and I made out like heroes. Month after month, we saw that people were flocking back to this website ready to get some action on. Oh and to Michelle’s point, not a single Monte Carlo method was used in
Genesis.
Everything was modeled via machine learning and data mining. Nothing was hypothetical. I still can’t get past that statement from last month at the Museum of Fine Arts.
I kept working with Punjab wherever I went and we did some work for some Wall Street banks automating many of their financial modeling procedures. Some of the bankers weren’t too happy as I was making their job obsolete. When I was in the elevator one day some junior banker looked at my shoes and said, “Doesn’t it suck that if your salary plus bonus doesn’t put you at the top tax bracket?”
I thought to myself, “since when are Bruno Magli shoes considered thrifty? Maybe it was my Hugo Boss suit?”
I really hope that banker’s best skill was the one I had just automated with
Death Star.
It was a little asset developed for mergers and acquisitions.
A few months passed by and word eventually got out about Punjab and I with the dating sites. Soon all the major competitors were asking us to re-haul their dating algorithms. So we did. I spend the next year and a half bringing
Genesis
to other companies. Based on the demographics of the users the website had and the underlying technology of each website, I tweaked the algorithms here and there.
So I guess now I’ve explained why I have such a distaste for online dating, I feel somewhat responsible for all of these people making bad decisions and finding their mate online. Whatever happened to looking someone in the eyes and just being stunned by their presence? Instead, now what we have is a system where your online actions determine your future. Doesn’t that take away from the magic and uncertainty of life? Online dating is too efficient. It’s emotionless.
I also can’t use it because any attempt for me to use it will not work the way it’s intended. I could game the system every time I log on, as I’ve developed so many of the techniques that match people together. It’s like taking a personality test in which you wrote the tests and answers. You’d score however you wanted to score. Oh well, it’s not something I ever wanted to dabble in anyway.
Too bad I signed all these non-disclosure agreements on my work. I would have loved to tell Michelle that I helped shape online dating as we know it right now. Well, even if I modeled a lot of the work off
YinYang. That would work her up wouldn’t it?
As much as I dislike the idea of online dating I did have fun mapping those relationships between people, their actions, and their potential mates. It reminded me of the days in which I mapped the relationships between words, languages, geography, and time with Eden. So what, I’ve always looked at numerical patterns to explain the world around me. There’s nothing wrong about that.