Progress and Pacifism
Life is an infinitely complicated model that we aren't supposed to optimize
It is straightforward but not easy to reach any goal. If you want to become a software engineer, you can find people (on YouTube, Reddit, etc.) who list the steps to follow, depending on wherever you are in life. In college? Great; take coding classes, go to office hours, learn data structures, Leetcode, read books to crack coding interviews, do personal projects, tailor your resume to every company, reach out to recruiters or employees, and apply to as many jobs as possible. Doing each step increases the likelihood of success, with many other possible steps to further increase the probability of success.
With only one goal, the path is by no means easy, but it is straightforward. But life is never uni-directional. On top of that, every decision opens and closes doors that inadvertently impact probabilities for whatever goals you are pursuing.
There is a chance that if you went to that party last week instead of studying, you would have found someone who would have become your well-connected best friend and would have given you that job that you were studying for in the first place, improbable as it may be. Let’s build a model to capture this possibility. Or why not make a model that captures every possibility?
And so the race to model every moment in life begins. If you create a model that includes every event that has ever and will ever exist, no matter how improbable it may seem, you can find the path of least resistance to reach any set of dynamically changing goals. If I want to be the prime minister of Canada in exactly 20 years, I should be able to use this model to output an infinitely long set of finitely-long exact decisions I need to make every day for the next 20 years in order to reach this goal.
But problems quickly sprout. If you do this, you’ll quickly realize that for any goal, the outliers drastically outbalance your model. If your goal is to live a long and healthy life and the outcome is measured in “years of healthy life”, then the improbable outcome of choosing a path that results in “infinite years of healthy living” outweighs all other paths. If there is an infinitely improbable possibility that if you drop out of school, sell all worldly possessions, and spend every waking hour researching aging, you will find the cure for aging and unlock eternal juvenescence, then it's worth it to pursue this since infinite life is worth infinitely more than any other outcome.
So let’s remove these outliers by limiting possibilities to the extent of current academic research. But it takes time to figure out what the bounds of possibilities are. Okay, let’s schedule a flexible 2-hour block every day to keep track of what the frontiers are. Let’s follow top researchers, explore new industries, and follow Twitter accounts. And let’s write these 2 hours each day off as a sunk cost to this model.
Okay, your 120 minutes today are up, it's time to start working towards your goals. But tonight you have to get dinner and catch up with a friend you haven’t seen in many months. It’s okay, Today’s break only costs 4 hours and in the grand scheme of the model, it marginally changes the probabilities. But after dinner, you are tired and sleep is more important in the long-term compared to doing work. So let’s sleep and re-calculate probabilities when you wake up. And so after even just a few days, regret starts to seep in. “If only I spent an extra 2 hours each day studying, I would have been much more likely to get an A on the final exam.” But maybe, if your goal is to be worth a million dollars as soon as possible, there is a possibility that when you fail, you’ll meet someone on the way out of the exam hall who hypnotizes the professor and offers you a million-dollar check because she likes your confidence. Anything is possible, right?
Okay, let’s refactor this model. Instead of the only cost being time, let’s also include physical and mental strain, money, and future strain. Let’s also more heavily weigh short-term outcomes so that the model is focused more towards my current goodwill instead of solely focusing on exact momentary future outcomes. And of course, the cost of every action is based on leading research. For example, if I sleep between 8 and 10 hours, it’s a net positive on future mental and physical strain. If it’s over 12 or under 8, it’s a net negative. Going back to the possibility that I fail an exam and get a million dollars, it suddenly becomes much less desirable to fail and more desirable to study because the short-term impact of a fail in nearly 100% of the decision trees is a much larger net negative to the goal of reaching a million dollars and sleeping more, than the impossibly improbable outcome of being handed a million dollars.
There are infinite ways to optimize this model, and the infinite time and mental strain required to calculate and recalculate probabilities is an infinite cost to justify the deterministic, infinitely useful traits of the model.
However, it quickly becomes obvious that implementation of the model is impossible. Every decision is somehow solely the responsibility of the individual, yet it is impossible to optimize quality for any given task. Human error runs rampant because of the human operator, whose entire existence is plagued by residual error.
~~~
This model is stupid. And yet I sometimes find myself believing parts of it because it makes me feel like I’m in control. According to this model, every mistake is a recalculation in a determined “best path” and every moment of pleasure is a wasted moment that reduces the probability of future pleasure. In effect, it lets me justify the bad decisions I make and the misery in my life because I don’t have control over my life, the all-seeing model does.
The original purpose of the model is to bring urgency to life. Self-help gurus and thought leaders espouse the benefits of waking up at 5 a.m., eating exactly 3.5 meals a day, working infinitely hard for 12 hours, and sleeping exactly 9.69 hours a day (aligned with the gravitational shift of the stars, of course). But precision leads to absurdity, where the goal isn’t to reach your goals but to build the best machine to reach your goals - a devious halting problem that none of us will be alive for when it outputs 42.
The model also leads to a sunk cost fallacy - if I’m this far, the only way to determine if the model actually works is to continue believing in it. No wonder there are cult-like followings for so many impossibly improbable beliefs and belief systems.
So how should you and I live our lives? Maybe, we should live every day knowing that we have the freedom to do whatever we want in our life. Goals are just events in time and space, and if you think too hard, they become impossibly distant to reach. Take that leap of faith and do whatever the hell you want.