Modeling Satisfying Life Decisions

Making difficult decisions even harder, but more satisfying and interesting

Every so often, I’m asked a question like “Should I go to college A or college B?”

I’ve always found myself somewhat confounded by this. Not only because I dropped out of a Canadian university, so I hardly know about US colleges. Instead, something about the shape of the question itself bugs me.

Surely between Choice A and Choice B, there is a choice that is “better” than the other right? Surely, one choice will be a better fit, yield more of what I value, and lead to less regret than the other?

If we were in a deterministic world where we can have complete information, then this would have to be true. Though, that’s certainly not the world we are in.

Not only do we not have complete information, but for many decisions, we can’t even be certain about how much information we don’t have. Any attempt to gain more information to increase the chance of making the “right” decision is useful but seldom conclusive.

But we can at least build a probabilistic model to account for incomplete information. This way we can still improve our decisions based on evidence … right?

Here, it’s useful to distinguish two classes of decisions that we make. One where the value of a decision is pre-defined and one where the value of a decision is under-defined.

In the pre-defined case, we can think of a context like poker. Many decisions are made in poker, under incomplete information. However, one thing about the context is clear: the value of the decisions. This means that while I may not know how exactly to do so, I know that maximizing the amount of chips in my possession when I end the game is unambiguously the objective. The game’s value function is defined, which means that uncertainty in terms of incomplete information can be compressed into probabilistic reasoning like "pot odds", distilling theoretically optimal plays. There are many such cases from Chess to Monopoly, corresponding well to James Carse’s notion of Finite Games. We know what the objective is, thus what is valuable, and the only uncertainty is how best to achieve it.

In the under-defined case, decisions are usually made in complex environments, which inconveniently include many of our most important life choices. Which school should I go to? Which career should I pursue? Who should be my partner? Where should I live? In these questions, the objective of the game is either unclear (e.g. how could I know where I should live given: climate change, changing geopolitics, social scene, etc.) or far too abstract (e.g. I’d like to be happy). The objective(s) resists formal modeling and leaves us with conflicting intuitions with not only the uncertainty of success but also the uncertainty of what success should even look like.

The pre-defined case is a relatively well-studied subject. I’m interested in how to make better decisions in the latter, including wrestling with its nebulosity so that we can derive some nuggets of formalization that could inform our intuition.

In decision-making within these under-defined contexts, I think it’s helpful to at least start with the acknowledgment that this can feel scary. It’s hard to think critically about a subject when we are scared or anxious. These decisions often represent significant commitments, and in the modern environment of abundant choice, we are all too clear about the opportunity costs of such choices.¹ Despite this post’s purpose not being one of working out psychological blocks around choice-anxiety, it is often the essential first step before going further. Otherwise, we risk our own anxiousness grasping for certainty in such a way that we lose sight of what we care about.

Using “Which school is better” as an example, I want to try to outline a different approach from the usual paradigm of “increasing confidence in making the correct choice between A and B”. Instead, I’d like to propose:

“Increasing confidence in landing in a satisfying world, for either the decision A or B.”

Here’s what I mean.

Suppose I am to decide between School A and School B, and for the sake of illustration, we close off the possibility of any third choice C (e.g. not going to school at all). Instead of thinking that I’m faced with a binary choice of A or B, each with a corresponding and distinct future, we recognize that there are many futures associated with A or B.²

This is to say that if I choose A, I could have worlds where I might meet my best friend, find a great mentor or become depressed, and so on, including any combinations of these possible events. The same could be said for B. This is because by choosing a school, I am choosing a context that shapes my future choices, a context that contains many possible worlds.

Despite this possibly being obvious, big choices like this can often feel like a single thing. Going to Harvard can feel just like “going to Harvard” – especially for an incoming student with high uncertainty, all of those futures can feel compressed into one “Harvard path/future”. However, as any alum could probably share, there are many ways to “do” Harvard and there’s a great degree of diversity from how a school is experienced and how the future pans out.

Zooming in on School A as a choice, we might have something like:

Each of these distinct lives will become the basis on which one would evaluate if a correct decision is made. Some future versions of ourselves will look back and feel deeply glad about our choice. Other future versions of ourselves would wonder what the alternative could have been like in regret. Perhaps like this:

Now, in addition to wondering whether A or B is the right choice, we can also evaluate how likely we are to end up in the “Satisfying Worlds” of A. After all, we may discern with some confidence that School A might have more satisfying worlds than B and thus feel inclined to choose A.

Expanding to include B this time could look like this:

With B in the picture, we can now ask some nice questions like:³

• Which of School A or School B offers the greater possibility space? How confident am I in this?

• How likely am I to be in a Satisfying World given that I choose School A or B?

• Which choice offers the largest amount of satisfying worlds or the smallest amount of unsatisfying worlds?

• How well am I capturing all possible satisfying/unsatisfying worlds?

It’s a very different question to ask someone at School A: “Do you think I should go to School A?” versus “Tell me about how many ‘satisfying worlds’ you see at this school”.⁴ These questions allow for more rigorous investigation, with distinct categories of information to seek, in order to form a more complete picture.

In contrast, the more common way of comparing two options, where we determine several parameters and compare them (prestige, tuition, average income after graduation, etc) is mostly suggestive of the option space but does not offer concrete perspectives.⁵ This is especially exacerbated when every school markets slogans and qualities that approximate “Granting Our Alumni The Brightest Future”. Instead, evaluating by seeking out distinct futures invites us to qualify those lofty claims. Just what kind of Bright Futures are they offering? How easy is it for me to find out about these futures or are they all behind a veneer of marketing?

What emerges is a picture that de-emphasizes the binary choice of A or B. Instead, we foreground the evaluation of the various worlds in A and B, and how to land in a satisfying one, almost regardless of the choice itself. This is because “winning” exists in both A and B, and we are looking to maximize the likelihood of being in the green area or minimize the area of red, regardless of A or B⁶:

Now, one might say that while this may grant more ways to investigate the choice, it doesn’t necessarily bring clarity, so how am I supposed to choose between A and B? To that, I would say “the game” isn’t even to choose between A and B. Sure, at some deadline associated with a decision we will find ourselves in one of the universes A or B and not both.⁷ That fork will come and pass one way or another. We are looking for “Satisfying Worlds” not the “Correct Choice”:

The choice was never School A or School B. It’s always between what’s wanted and not wanted, of which the choice of schools is a proxy.

We care about being satisfied with our choices and by extension our life. This is not about grasping onto the hope that by making the correct choice, we would be finally, decisively “happy” or “free” or “belonged”. If this is what’s driving the decision, it’s recommended that we return to the earlier mentioned first step, recognizing that we are probably scared and addressing that. It can be quite scary!

By focusing on satisfying worlds, we are instead loading/potentiating our minds to be sensitive to subsequent paths for either choice. Like a standup comic who has run 100 variations of the same routine, they will be able to improvise based on the audience with an immense degree of creativity, we are improving our ability to improvise in either the worlds of A or B.⁸

Notably, this model is cognitively demanding. Rather than settling for some easy-to-determine criteria that ChatGPT could do about as well, we are constantly searching for details-within-worlds and worlds-yet-considered. This makes each choice made under this model somewhat undecidable, that is to say, we can keep on considering possible worlds forever – amusingly even beyond the point the decision has even been made.⁹

For me though, cognitively demanding as it might be (especially when starting off), this is a skill that can be honed and is worth honing. Specifically, it’s a “learning engine” that I believe is natural to human learning/cognition. Intuitively, this shouldn’t be difficult to imagine. A general who has studied many battles (by leading or studying) learns to disseminate the patterns in those battles as bits to recompose into creative new strategies. A good general wouldn’t be satisfied with simply having learned stratagems, but rather to keep considering all the ways things can go wrong (or right). Over time, this forms something vaguely labeled as Wisdom, where they gain the ability to approach new and old contexts creatively.

The honed ability here is what allows us to confidently interface with uncertainty – including decisions like choosing between a school. By confidently I mean the solidity in the general’s ability to face a new battle and feel in their body a sense of “I don’t know what’s going on yet, but I expect I can figure it out” – self-efficacy or “trust in the self”. This is why the fear of these decisions, if present, is important to recognize first. In the case of “going to Harvard or not”, a rising freshman might find it intimidating whereas a graduating senior might find it rich and thought-provoking. The difference is in this quality of self-efficacy, and as we gain this sense of trust in ourselves, decisions transform in nature.

When faced with a choice, someone who fears all options except the “correct” one and someone who trusts they can navigate any world is not making the same decision.

Fortunately, we can hone this skill, and in fact, we already do this naturally. With each decision, we naturally evaluate how well we did and update on this information. The key is to consciously purpose it towards how well we evaluate worlds. Perhaps we are glad of the choice, indicating that we are in a satisfying world. Did we foresee this world ahead of time? Why or why not?

Every “mistaken” decision is then incorporated into an evaluation of why we did not effectively foresee unsatisfying worlds or navigate towards satisfying worlds. We then gain valuable information on how to model more nuanced worlds for subsequent decisions. In fact, we may discover that most of the time, the “mistake” lies in reality having details that were totally out of what we can conceive of before.

In this model, we are no longer even minimizing regret, but rather transcending regret – or Post-Regret. We can still have thoughts shaped like “I wish I did that differently”, but rather than regret, we feed these thoughts into the learning engine that informs our subsequent evaluations of satisfying worlds, building our ability to navigate more and more worlds, thus developing self-efficacy.

It’s worth noting that this model is meant to be compatible with two assumptions about the nature of Choice.

• The first is that choices are not made in perfect agency (i.e. a perfect individual rational agent). We are embedded (i.e. in relationship), and sometimes we don’t get to choose what we want.¹⁰ This helps us make sure that however the choice pans out, we are equipped to navigate it well.

• The second is that each decision is not a conclusive choice. The choice continues in a continuum of choosing, each shaping the next. To think that one choice (like getting into the right college or job or romantic partner) will definitively conclude some quest for satisfaction is both epistemically inaccurate and ineffective.

Returning to the concrete example: “so, like, School A or School B?”

The answer, if this model works well, should increasingly sound like: “I don’t know! But I do know that both schools seem really great!”

I swear, this really is a real solution lol. The amusing part of this “solution” is that it solves not the content of the problem but the context of the problem – it solves the feeling of “there is a problem here” from the problem. The content remains, but is transformed: School A vs School B is no longer in a context of scarcity (i.e. one of the many choices must be best) but rather one of abundance (i.e. many of the many choices are great).

Solving problems in abundance is very different in nature than solving problems in scarcity. We are ever increasing the attractive options we face, which can feel overwhelming – necessitating a set of skills quite different from more common reductive modes of decision-making. This warrants an expansion in a separate piece of writing, but illustratively improv is a good example of where such skills are built. “Problem-solving” in improv should taste very different than normal problem-solving because we are firmly in the Infinite Game. Since life is also an infinite game, practicing solving abundance-based problems is certainly a worthwhile endeavor.

Finally, not only do I think it’s possible to learn this, but teaching this should be well within the possible. The pedagogy still seems poorly understood (i.e. I barely scratched the formal possibilities here) and poorly distributed (i.e. we don’t see this in regular education). This is why various jobs that require interfacing with complexity and uncertainty like business management or military special forces often select for a certain “je ne sais quoi” that is difficult to train for – call it “Leadership” or “Ability to Problem Solve” or “Hacker Mindset”.

Researching and implementing learning engines like this in “contexts for learning” (e.g. businesses, schools, relationships, etc) constitutes a good portion of my work. If you are interested, consider reaching out.

1. Even choosing which restaurants would be good for a date can sometimes feel overwhelming, never mind which career path one should be taking.

2. For the math inclined, I mean that A is a set that contains {a1, a2 … an}, and choosing A means choosing the set, rather than a specific outcome.

3. Correspondingly:

1. Is Set A > Set B?

2. The conditional probability of Satisfying World given I chose School A/B.

3. Is p(Satisfying World | A) > p(Satisfying World | B)?

4. My confidence in the accuracy of any sets and subsets in the diagram.

4. Not only would it likely yield more useful information for the asker, but the person responding will be invited to reflect on their own experience in a way that likely yields more insight for them too. Conveniently this usually makes people more willing to talk, therefore yielding more information – a sound tactic for the investigation.

5. To be clear, I’m not saying don’t do this usual way of evaluating but more like use this as a tool that informs our exploring of worlds.

6. For the crafty and ambitious, one can also seek to maximize and expand the area of the entire option space. This is often a bit more difficult, though the skill is certainly worth honing, in my experience.

7. Except for when one might later decide to transfer or study at both schools or neither schools or any other infinite possibilities. The binary choice can be deceptive.

8. In a predictive processing sense, we are creating volumes of predictions representing many potential worlds, such that when reality meets us, we are already sensitized to emerging, subtle details of reality matching up to satisfying worlds that we have been predicting.

9. Though since most of the time, we are endlessly cycling anxiously on this kind of decision anyways, we might as well cycle more usefully on this type of computation rather than “what if I make the wrong choice”. 🤷

10. This can be good! I almost joined a pyramid scheme when I was a teenager which my mother thwarted despite my protests.