Dealing with yourself

A few months ago I received a call from one of my friends. “They have unlimited pani puri,” she said, “you in?”

Without thinking twice, I ordered an Uber and was on my way. $17 for an Uber for some $10 unlimited pani puri. A few weeks prior I spent $33 Ubering to a tennis match and back. On the surface this sounds egregious, especially making a habit out of it. But it’s intentional.

I don’t own a car

Owning a car would easily cost me at least $6k/year amortized (parking alone is $3k/yr at my apartment). I live in a walkable neighborhood and walk to work, so the only real value I’d get from a car is convenience for spontaneous trips.

So instead I decided to spend freely on Uber. This is still way cheaper than owning a car, since I could afford an hour of Ubers per week (with ~$100/week to spend)¹, and I get all the upside of a car (spontaneous trips, convenience) without the downside (parking, maintenance, traffic).

But importantly, this system only works if I actually take the Ubers when the time comes. If I chicken out on $17 Ubers because “that’s too expensive” and start avoiding these spontaneous trips, I might start fantasizing about purchasing a car. And just like that, a seemingly responsible decision cripples my past self’s entire strategy.

Principal-agent problems

Okay, so your present self can undermine your past self’s plans, and just stick to the plan and be disciplined and be smart and think and good things will happen, blah blah. Where is the insight?

This issue is very similar to the principal-agent problem in economics, where one person (the principal) delegates work to another (the agent) whose incentives are not aligned. For example, a company hiring a contractor, who cuts corners since they don’t care about the long-term stability of what they build.

I think something like this happens inside a single person. Past-you is the principal. Present-you is the agent.

The principal designed the policy (e.g. never hesitate to Uber), and present-you, the agent, while having more information, is also much more short-sighted.

Constrain the agent

In 1519, Hernán Cortés landed in Mexico. He knew his men might want to retreat, but rather than wasting his time cooking up a motivational speech, he just destroyed his own boats. In doing so, he shifted agency from the agents (his men) back to himself (the principal). You can apply the same tactic to your own problems.

Just shift agency from your present self to your past self where possible. In some cases you can remove almost all agency from your present self. For example, by buying and pre-loading Uber gift cards into your account, or prepaying for a gym class.

And in cases where you can’t shift all of the agency, you can oftentimes still increase the cost of defecting. For example, telling all your friends about your plans, so that you feel guiltier about defecting (see also: weddings?).

But what if the principal was wrong?

Fair question. If the agent constantly disagrees with the principal’s plans, then one of you is wrong — but it doesn’t matter who, because either way the fix falls to the principal, not the agent. The moment of execution is the wrong time to evaluate it. The agent should ideally just follow the policy and make a note for the principal to look at later. Maybe the Uber budget should be lower. Maybe the policy needs to change altogether. But this reconsideration should happen later, back when you are in “principal-mode”.

But what exactly is “principal-mode,” anyway?

If you zoom out, you might notice that each principal is just an agent for an even longer time-horizon principal. The you who plans this week answers to the you who plans this year, who answers to the you who plans your life.

And then… what does that version of you answer to?

I’m not sure. But this is getting to be too much.

I’m going back to my pani puri.

◆ ◆ ◆

Thanks to draft reader Michael