Dataclysm: Love, Sex, Race, and Identity

A partner search engine like OkCupid is a real treasure trove of data. Between May 2013 and April 2014 alone, around 10 million users went to Dating.

 What users reveal about themselves reveals much about the way they judge other people, what connects or separates them.

The average user answers about 300 questions on OkCupid. The answers provide interesting and thoroughly generalized insights into human behavior.                                                                                                                                               

For the following results not only the data of the OkCupid users, but also other data sources were used.

What age do women prefer to men, depending on their age? This question can be answered very precisely with the help of big data: Women are always attracted by men of the same age.

However, around the age of 30 something changes: Before this date, women prefer men who are slightly older, then men who are slightly younger. From 40, this tendency is exacerbated again: it seems that men of 40 years have reached the zenith of their attractiveness.

It is completely different for men when they want to assess the attractiveness of women: Whether 20 or 50 - men of all ages, the numbers suggest, find women between 20 and 23 years most attractive. So even though the men themselves are getting older, their preferences do not change.

The corresponding data was collected by presenting photos to men and women - so the results come from a rather indirect survey.

They show what both groups unconsciously desire. However, this is very different from what they actually say in partner search questionnaires.

For 44-year-old men are not looking for 20-year-old women, but rather 30- to 49-year-old women. In fact, they mainly contact 35-year olds.                                                                                                                                                                   

This creates a dilemma: women want men who grow old with them, men look for young women.

"Mathematically speaking, the age of a man and his sexual target group are independent variables: the first one progresses steadily, the second obviously not."

At OkCupid, users can rate each other up to five stars. Interestingly, many members with a few stars seem particularly attractive to some members. Some feel discouraged by the blemishes, but others are particularly attracted to them.

If women in the database reach a mean of three stars, then this value can be very different: Either they were rated by many men with three stars, or half of all men found them super sexy and the other was completely deterred.

In both cases, the mean is exactly the same. This distribution of data around the mean is called variance. Dating has shown that women with a high variance receive significantly more queries than others who achieve the same mean with less variance.

This is probably due to the fact that the number of 5-star admirers is much higher - and that is simply more interesting than the average rated women. Profile photos are allowed to polarize - with a funny facial expression, visible piercings or a toast in the hand.

This can even be decisive for the success of the contact ad. By the way, if you're looking for great love, it's a good idea not to lie, because otherwise you'll be assigned the wrong partners.

Anyone who has found potential partners in the database ideally gets in contact with them. How does that happen? The databases also provide information about this.

There are many users who send out long-term contact requests for several pages, for which they need hours, others are short, concise and spontaneous. And then there's the copy-and-paste faction, which sends in-hand or copied contact emails to multiple potential partners.

They are not original, they get about 25 percent fewer responses, but they are incredibly efficient: the ratio of effort and number of answers is very good, despite the unimaginative copying.

A contradictory behavior showed the OkCupid users, as the operators of the platform took an experiment. Under the motto "Love is blind" all profile pictures were deactivated for one day. An app instead generated blind dates.

The resulting couples had no picture of each other until they met live in the café. The app also allowed the rating of the meeting. It turned out that both women and men were very satisfied with their meetings, regardless of the attractiveness of the other person

With activated profile picture, the whole thing looks different again: If the man is less attractive than the woman, she rarely answers the request. If he is more attractive than she is, her frequency of response exceeds a normal value many times over.

The men showed similar values. The criterion of appearance on the Internet obviously plays a big role in the selection - when getting to know each other in real life, it is not so important anymore.

Things like looks, a certain political orientation or denomination are often meaningless for a relationship - yet they are claimed by the users. The "Love-is-blind-day" brought interesting data to light, but was never repeated.

If one were to allow the algorithm of OkCupid and trust it only the inner values ​​of the men and women who are looking for a partner, the skin color would play just as large a role as the zodiac: none at all. Everyone would get along pretty well.

In the US, there are four groups of skin colors: Asians, whites, blacks and latinos. If the respective preferences for the different groups are included, big differences are revealed.

No matter if you use the data of OkCupid or other dating sites: black women are always rejected by men of other skin colors. Women show a high acceptance for men of their own skin color and wish white men as a second choice.

Another line separates the beautiful from the rest: just as little as about your skin color, you can do something about your appearance, but in the evaluation it plays a major role. Especially for women: Women in the top spots of the attractiveness scale receive disproportionately many offers.

Incidentally, this is also evident in professional life: While the appearance of men has hardly any influence on their chances of being hired, the factor of attractiveness in women makes the chances of success shoot up exponentially.

"OkCupid is one of the deepest distinctions, perhaps the deepest, between the beautiful and all the others." Who is beautiful, is also good. This myth seems to have prevailed everywhere, even in areas where it is often completely useless, as in the working world.

The problem: The ideals of beauty - even completely outlandish, how to be super thin - rampant in social networks and lead to psychic or even physical disorders in unsettled girls.

How do the users of OkCupid describe themselves? This question can be answered with an analysis of the word frequencies. The white men and also women brag about their blue eyes, the blond - in women also red - hair and an amazing love for outdoor activities and campfires.

Asians most often mention that they are "quite big for an Asian," Latinos are noted for their mention of Latin American musical styles, and blacks often write about black artists like Jill Scott, Jamie Fox, or Luther Vandross.

As far as sexual orientation is concerned, Big Data offers more insights than official requests. Looking for the preference for gay pornography, it turns out that in the US about 5 percent of searches go in this direction, therefore 1 in 20 men seems to be gay.

"51 percent of women and 18 percent of men have ever had a gay experience or would like to try it." Officially, in states like North Dakota, only 1.7 percent of men are gay. Conclusion: Another 3.3 percent are, but hide it.

The question of whether heterosexual men and women have ever had a sexual experience with a same-sex partner reveals the following: 33 percent of women and only 13 percent of men say yes, and 17 percent of women and 5 percent of men would like to try it out.

Does the place of residence decide how one feels and what one thinks? Big data can also help answer that question because the data can be geographically assigned.

When asked "Should it be forbidden to burn the American flag?", A clear urban-rural divide emerges: It does not bother the townspeople, it looks quite different in rural areas.

Even the frequency of daily body care can be explored. Surprising result: Regardless of the climate, New Jersey is more fussy about hygiene than surrounding countries - and in Vermont, locals do not really care if they stink or not.

There is a surprising answer to the question "Sex or Love?": In the north of the Midwest, people seem to be more driven by lust. That does not correspond to the conservative stereotype.

"Google has almost incidentally become the register of the collective unconscious of humanity."

The data we leave behind on the net is a trail of breadcrumbs that allow data collectors to get us straight to the bottom.                                                                                                                                                                  

Algorithms can use Facebook likes to reliably calculate gender, age, sexual orientation, drug use and political attitudes.

Facebook has only been around for a few years. How will our world change when a generation grows up entirely on the Internet and then, for example, when entering the workforce, can already be fully screened?

And without the option not to participate, for example in the recruitment test or assessment center. No doubt this is one of the dark sides of Big Data.