The Arranged Marriage Algorithm

“So, Prithvi, we have heard that Indians mostly have arranged marriages. Is this true? How does that work?”

I was out at dinner with a bunch of business colleagues when I was asked this question. Conversation around the table paused. My colleagues were clearly interested in hearing about Indian arranged marriages. These colleagues included smart people from Brazil, Denmark, Poland, Bulgaria, Canada and Britain. I needed to come up with something truthful, credible, and that showed India in good light.

I had a hunch that this perception of Indian arranged marriages was shaped by Western media coverage of normal, happy girls from, say, suburban Birmingham, who are married off, against their will, to tribal chieftains in Kandahar province. These stories are true, and are terrible tragedies. However, my knowledge of this world in negligible, and comes from the same media reports as my Brazilian or Bulgarian colleagues.

My marriage, and marriages in my immediate family, have all been “love marriages”. However, several very good friends of ours have had arranged marriages. These couples generally are educated, professional, affluent, cosmopolitan, urban Indians. As far as I can tell, the texture of these arranged marriages is not all that different from love marriages. If anything, some of the most lovey couples I know - who cuddle together in company, and address each other as “honey” and “sweety” - came together through arranged marriages. I actively dislike the terms arranged and love marriages, partly because of the implication that arranged marriages are loveless.

I told my colleagues that while media reports about girls abducted to be married off against their will are true, they entirely outside my experience. Within the relatively privileged circles I inhabit, arranged marriages are quite common. However, here, they work quite differently from the media stereotype. Here, arranged marriages serve exactly the same purpose as the compatibility matching algorithms in dating websites like eHarmony.com or Match.com.

A couple brought together by an arranged marriage algorithm have a number of things in common. Their parents get along, or at least, are not contemptuous of each other. They are from similar social and economic backgrounds. They have similar levels of education, and are likely to have similar attitudes to a bunch of stuff. All these are good statistically significant predictors of marital success. What the arranged marriage algorithm, or the eHarmony.com algorithm, does not predict is chemistry - the magic electricity that crackles between, say, Elizabeth Taylor and Richard Burton. Which is OK, because the arranged marriage algorithm works just fine, as long as individuals can keep exploring options until they find someone with the right (or reasonably good) chemistry.

This gambit worked: Everyone at this table was a quant. It led to some companionable geek-talk about how one could improve the quality of these algorithms (should one hold out a control sample, of couples who are intentionally mis-matched, to train the matching algorithm?) until my Bulgarian colleague chipped in. She was one of two women at the table. To her, algorithms to predict compatibility are worse than useless, regardless of whether they’re authored by clans or by eHarmony.com. They totally miss the point. Chemistry is not just one more factor in a marriage. It is the central thing, the only thing that matters.

This served a nice segue to a set of stories about how women are more romantic than men, and my defense of the Indian Arranged Marriage was successfully concluded.