[Unfinished] Regret Minimization as a Life Policy

How do you live an aesthetic/good/satisfying life? You can prioritize optionality, control, or a plethora of other metrics. How does regret compare to these other life policies? What does it mean to minimize regret?

The "regret minimization framework" is a framework popularized by Jeff Bezos as one of his driving motivations for founding Amazon. He argues that he is likely to feel more regret at the end of his life by never trying to start a company than he would feel trying and failing to start a company. This relatively simple calculus led him to start Amazon and create the company we know today.

To understand regret minimization, we need to differentiate it from alternative life policies. Through this differentiation, we can learn what might be better or worse about regret minimization. In short, regret minimization is a patient policy with continual reevaluation.

Greedy Policy vs Patient Policy

All life policies can be broken down into two categories: greedy policies and patient policies.

A greedy policy evaluates candidate choices against their immediate effect on one's goals. "If I eat this, I won't be hungry anymore." Greedy policies are fast to calculate and easy to understand.

A patient policy evaluated candidate choices against their effect on one's goals some arbitrary distance in the future. "If I eat this, I will reduce my quality of sleep tonight, that means I might have less fun at the brunch I have scheduled tomorrow." Patient policies are a bit more difficult to calculate. "Well if I don't eat this, I might go to bed hangry and have an even worse night of sleep." There are so many candidate futures to consider that your decisions will, by necessity, take longer to resolve.

There are appropriate times and places for each policy style. If you touch a hot stove, you most certainly don't want to wait for a patient policy to percolate down your arm asking inane questions like, "well maybe this burn will take me to a hospital in which I will meet the love of my life." You want to take your hand off the stove. Alternatively, "I should probably leave my money in the index fund instead of YOLO'ing it into GME because I know, long term, that will be enough."

These two policy styles can be applied at multiple levels of abstraction in your own life. How do we know what the single highest level policy should have as its style? Intuition says that the delayed gratification associated with patient policies will be the winner - and fortunately for us, the regret minimization framework is a patient policy.

Single Evaluation vs Continual Evaluation

Ethics tell you what is right to do or wrong to do. Morals tell you if what you did was right or wrong. An ethical code can foster trust in a community, even if it produces locally un-optimal decisions (e.g., no man left behind). Morals save your from committing atrocities in the name of some misguided and man-made ethical binding. This distinction is similar to the distinction between single evaluation and continual evaluation policies.

In a single evaluation policy, you might wake up, make a five year plan to make the most money you can, execute that plan faithfully, then wake up at the end and say "wait, no one needs COVID test strips anymore?" A continual evaluation policy has you wake up every day, make that five year plan, execute it for the next day, then write it all over again. In many ways, this is good! You won't wake up one day and realize that the world has passed you by. However, you might trick yourself into following fads and fashions instead of working on what is important to you.

What Benefits can Regret Minimization Provide?

Suppose one wants to live a good life. By what mechanism is a good life measured? Is it measured by oneself, at the end of ones life? Is it measured by the summary of one's peers? The excess goodness in the world that can be directly attributed. to ones actions? For each of these mechanisms, we need to consider some set of counterfactuals: if you had not done what you did do and did something else entirely, what would be different?

If you hadn't stopped that person from stepping out into a busy street, they might have been harmed. Your actions, when measured against their hypothetical absence, were good. But how do you measure the "goodness" of doing something like starting a company? We most certainly shouldn't have the ego to assume that what we want is good - though, is it reasonable to assume that a future you would be able to look back and reflect honestly on a decision you made and measure it's goodness? This assumption is certainly more reasonable than believing you have the tools to fully know what is good now?

So how do we act in the "now"? We can not stand in the certain shoes of the future, look back, and judge out past decisions - we can try. At its core, the regret minimization framework is standing in the "now" asking yourself what your future self would think, looking back on a currently prospective decision, and casting this judgement as honestly as you can.

So, what might this practice actually do for me? TODO

Alternatives to the Regret Minimization Framework

In 1745 Days and 2000 Days I establish and equivalence between the two life policies of maximizing present optionality and maximizing present control. Additionally, there are some intuitive life frameworks like: maximize present happiness, maximize number of children, make the most money possible. Some of these evaluate the "now" and some are future-facing.