uncertainty by Vincent H. '23

recently i looked through a list of all my friends who graduated from college recently, and i realized that all of them seem less interesting to me now than they did while they were in school

i have a bad habit where i’m attracted to uncertainty. blog ideas and unfinished drafts usually feel better than complete posts because the process of completion removes uncertainty, and in doing so shrinks the realm of possibility for how good a post can be. i am optimistic by nature, so whenever i’m working on a half-finished project i tend to assume a positive outcome for the part i have yet to finish, and as a result i am usually discontent afterwards

i think this is actually a pretty common bias. it’s closely related to why people are often disappointed when they achieve their dreams, and why many people have commitment issues (i don’t think i do, but i empathize with the state of mind of not wanting to commit because you’re worried about missing out on better options that may not even exist)

last summer i was working at a startup whose focus was to build infrastructure to make ai systems as efficient as possible. afterwards my boss told me they were pivoting away from developing the infrastructure, and their new focus would be using their existing infrastructure to make user-facing ai products, the first of which was a code generation tool. my initial reaction was “you have some of the best ai infrastructure in the world and could build any ai-powered product, and the best thing you could come up with was code generation?” it wasn’t until much later that i realized maybe i was being a bit unfair; i probably would’ve had a similar reaction regardless of what direction they settled on

one related idea i often think about is the concept of “beginner gains” in the context of learning. that is to say, if you’re a curious person with an open mind, you can convince yourself that practically any subject is interesting while you’re in the initial stages of learning that subject. this happens because, when you’re a beginner, you’re mostly learning generalities and new abstractions, which are usually pretty enjoyable to reason about; it’s only much later that you confront the reality of what it’s like to wrestle with a field, to have to memorize page after page of granular definitions or to spend hours working through tedious calculations or to conduct experiments that proceed glacially. learning is always fun when it’s done at a high level with few details attached; it’s only after you get past the beginner phase and are studying something fairly specific that you can determine how much you truly enjoy thinking about a subject

it’s not always the case that specific outcomes are less appealing than uncertainty though. once in a while you encounter something which is so far out-of-distribution that it surpasses your imagination, something so incredible that if you described it to someone who didn’t already know about it and then showed it to them they would be more impressed by the thing itself than by its description. i’ve written about this idea before in the context of meeting new people – that some people are so much more competent in an area than you that their work is indistinguishable from magic – but occasionally you’ll also find outlier products or ideas or art (for me some examples in these categories include the internet, cantor’s diagonalization argument, and bojack horseman respectively)

when i was first writing this post i thought the takeaway was going to be that certainty is always less attractive than uncertainty, and therefore you shouldn’t feel bad if your life turns out to be significantly more dull than you had envisioned. then i thought about the magicians i know and realized this probably isn’t the right conclusion to draw. i think it’s important to recognize that magicians exist and you can do extremely interesting things if you want to, but also that interestingness is overvalued and a lot of important work is quite dull. a lot of education work is just ordinary teaching and curriculum development, a lot of ai research is just scaling and fine-tuning existing research, a lot of climate work is relatively routine materials science and policy work, and so on. maybe it was always inevitable that i’d become a boring adult