[Question] How would one go about defining the ideal personality compatibility test?

[digitalcaffeine]

Let’s talk “soulmates”.

Pop culture has more or less drilled into our minds that the idea of machine-found love is dystopian and scary in nature, as if intoxicatedly hitting on members of the other sex at bars, or vainly swiping on profile pictures could possibly lead a person to their ideal match.

On the other hand, we already do have personality compatibility tests such as the Myers Briggs, not that its 16 personality results serve as anything other than a vague, pseudoscience guideline in regards to whom you would probably get along with. I’d like to think we can do better than this.

So, here’s my question for you, LessWrong: Using only practical elements (re: technologies we have available to us) and preferably a reasonable number of questions (<100?), what would you define as a quiz set that would work best in matching two hypercompatible personalities together? Assume you’re tasked with this in the hopes of creating romantic couples, and that you have a reasonably large user population to work with.

[Bucky]

I think I would have to pull Efficient Market Hypothesis on this and direct you to e.g. Match.com. Huge multinational dating sites are dependent on making good matches and I can’t think of any strong enough reasons to expect the market to be inefficient enough for me to do better.

You can read a bit about how Match.com do it here.

[digitalcaffeine]

Doubtful. For-profit dating services make their profits off of keeping people on their platforms for as long as possible. Thus, it’s very much in Match’s interests to prevent people from meeting any hyperidealized “soulmate” for as long as possible.

[Bucky]

For-profits operate in a marketplace. Provided the marketplace is working, it doesn’t matter if they would rather keep people on the site by giving poor matches—if they don’t match people well then another company will give the people what they want and take their market share.

First you need to show that a company can operate against the interests of its customers for an extended time period without market consequences, then you can talk about what the company would like to do with this ability.

I don’t claim this can’t happen, just that the market working properly should be the null hypothesis.

[Dagon]

My model of people and matching is _not_ what you describe. There is no “ideal match”. There are hundreds of millions of great matches, and it’s more about timing and environment than about finding the perfect one. Arranged marriages work. Dating apps work. Meeting in bars or activity centers works. Friend of a friend works. Finding the person isn’t hard. Finding a person _at the time when you’re both actually open to the sacrifices and compromises to make it permanent_ is hard. And maintaining that willingness over time is extremely hard. A bit of self-deception helps in this, which is why pop culture shows it so much: if you can convince yourself that this is “the one”, it’s way easier to make the choice to continue.

I love my wife of 16 years, and can’t seriously imagine making a life with anyone else. But I know that’s my brain lying to me, to help me reinforce my commitment and joy. In fact, it’s likely that there are many millions who would have matched just as well.