## Potpourri

==> My interview with Titus Gebel on his proposal for free private cities.

==> Carlos and I discuss the 2017 tax reform’s impact on charitable giving.

==> This YouTube video using “optimal stopping theory” to guide you in dating wasn’t what I was expecting. In the search models we studied in grad school, you would decide on your reservation wage and keep going on interviews until you got an offer at least that high. In contrast, the video describes a strategy where you *don’t *pick a reservation level of desirability in your partner, but instead decide on how much time you will sample, and then you switch to saying, “Now, from this point forward, I will marry the person who is the best I’ve yet dated.”

Disclaimer: I haven’t watched the video.

But isn’t there a practical problem with saying “From this point forward, I will marry the person who is the best I’ve dated?” Mainly, that most people you date won’t let you date other people, and are quite unlikely to take you back if you break up with them and try to reconcile later?

If you aren’t a plate-juggling Chad, you can’t date 10 women at once and pick the best (unless you’re on The Bachelor). Instead, you have to date one woman at a time, sequentially. So say you date 10, in a row, and after you break up with the 10th, you look back and decide the 3rd one you dated was the best, so you should marry her. The odds of her going along with this (assuming she hasn’t already moved on to someone else, which, if she’s the best you dated, seems unlikely in and of itself) are quite low. And if you dare to be honest with her about how/why you came to this decision, practically zero…

You’re misunderstanding the rule, Matt. You never go back to someone.

What you do is date (say) 10 people over the course of your first 5 years dating. Then, starting in year 6, you keep going on dates and if the person isn’t the best you’ve ever dated, you move on. Once you find someone who is better than anyone you ever dated during the first 5 years, you propose.

While I love this one simple rule. It reminded me of this explanation of it which also includes an overall chart.

http://datagenetics.com/blog/december32012/index.html

How much would you regret missing out or having to compromise later in life? Maybe you would be ‘optimal’ when weighted for different scenarios to only reject for the first 20%.

Or maybe you don’t believe you can really KNOW who is optimal. Swallow humility and be content for anything in the top 10%. Maybe that means you should only reject the first 15%?

Or maybe you don’t need to date people to get a feeling for people’s character, maybe your calibration by which you’re judging what is optimal in the first place means you already know what kinda people are out there and you just settle for the first who is good enough. I think that makes perfect sense.

I have a lot of questions but only two of them are not variations of “Who would do such a thing?”:

(1) If you pre-commit to dating some number of people before switching to the “marry the best person” strategy, and the last person you date before switching strategies is the best person you’ve dated so far, should you just marry that person or follow through with your pre-commitment and dump that person to look for someone even better?

(2) Can you and should you ever tell any of the people you are dating, either before or after you switch strategies and/or get married, about your plan?

Andrew,

On (1), no you shouldn’t, because suppose you *did* do that. Then that effectively would be the same as doing the original strategy, but only with the first 9 people. (I’m assuming the original strategy said to go the first 10, then start monitoring.) And then, if you were doing *that*, it would raise the question: What if the 9th person happens to be the best so far? Shouldn’t you stick with that mate, rather than following the modified strategy? etc.

On (2) well, this lies outside the scope of the assumptions. And probably applying this technique to dating just guarantees we’re going to get into awkward spots. (The point of the exercise is more about optimal search theory, rather than dating.) But even on its own terms, it’s actually not that crazy for someone to say at the start of a relationship, “I’m not looking to settle down.” I think people in the real world do that all the time.

No strategy is perfect. There is a chance that the first person you date is the best one for you, but that is very unlikely. Following this strategy you run the risk of missing out on the perfect partner in the “trial period”. but statistically you maximise your chances of finding a better partner overall.

The idea is not “how to find your perfect partner” but “how to maximise your chances of finding a good partner.”

By the rules of this strategy, if you find the perfect partner during the trial period, then you’ll never marry anyone.

That is a problem, but is is not quite right. If your perfect match was in the trial period you end up with the last one you try. The problem with this is that you must pre-define either the number of dates or the time period.

The guy in the video, using OST, ended up with 6, 77 and 2 away from his “perfect” 10. He miss-applied the rule, because in the second try he said he remained single. In fact, the rule states that you must accept the last candidate if you have not chosen by then, and in this case the last number generated was 87.

Just using his intuition he ended up with 6,4 56. He was closer overall just using intuition. Mathematics tells us this would not continue over an infinite number of plays.

In life there is the obvious problem that it is not all down to you – the other person has to say “I do.” This means leaving it until the very last person is a big risk that you will remain single or fail to employ a secretary.

Numberphile did an analysis on this using the best toilet at a music festival. It starts here:

https://www.youtube.com/watch?v=ZWib5olGbQ0

There is a follow up that goes into the maths a bit more. The probability of selecting the very best is 37% and also the point where you stop checking and start selecting is also 37% of the total number.

As I understand it, which is not fully, this is the best way to maximise your chances of selecting the very best option, and 37% chance does seem pretty good. This is not the same as maximising your chances of getting an acceptable result. For example, if the best is in the first 37% you will end up with the last one – an essentially random selection. So you have a 37% chance of selecting the best, but also a 37% chance of a completely random pick. It seems there should be some algorithm that gives you, say a very good chance of getting in the top 10% of choices and very little chance of picking a complete dud.

This is essentially what the guy in the video was doing, for which the optimal stopping point is not an optimised solution (I think).