Resist categorization, become incompressible

the internet

imagine a stereotypical suburban American high school, the kind that's straight out of an early 2000s coming-of-age movie. during the good old days (think 60s to early 00s), interests and identities carried a notion of locality; who you became was most affected by your immediate surroundings: parents, neighbors, cable TV, local gossip. in our imaginary high school, (almost) everyone falls into predictable groups: jocks, nerds, skaters, theater kids, loners. because discovering new ideas or falling down rabbit holes required physical 'luck,' there existed a fundamental upper limit on what you can do and who you can be.

the internet flipped this dynamic on its head, because suddenly niche ideas became widely available and (more importantly) pursuing them no longer meant becoming socially isolated. compared to the old world (our model high school) where a nerd's purpose is to stay invisible and get good grades, a nerd can now explore anything, understand anything, become anyone. i'm very lucky to be exactly the age I am, because the internet reshaped what's possible right as i entered childhood and now LLMs are reshaping what's possible right as i enter adulthood. life is quite serendipitous in this way: if you look for signs of how impossibly lucky you are to be alive exactly when and where you are, you'll see them everywhere.

growing up, i watched a lot of youtube. i first started watching minecraft videos, but over time i've discovered and drifted towards many interests—mostly centered around how things work (unsolved mysteries, ARG analyses, political commentary, stories from many many many areas of modern history, breakdowns of ancient & modern technology, broader reflections/metacognitions from creators) and how to make working things (crazy ambitious garage engineering, multigenre music production, RL agents for videogame ragebaiting, generally fascinating artists and engineers displaying the act of arting and engineering). i also watch a lot of 'white boy friend group doing diabolical things' content (pranks, scavenger hunts, try not to laughs, public presentations, etc).

i believe collecting diverse knowledge about the world and the people in it is innately good, and the internet—the most expansive, infinitely accessible library to ever exist—makes this trivially easy. having grown up during the tail end of the world's digital migration, my mental model of high school is much more fluid than in the movies: the economist builds robots, the robotics captain is a linesman, the astrophysicist is hypersocial. in other words, i've discovered we have the power to go deep into things, but to also do this for many things, and move freely between them.

following threads

here are some things i've ~seriously tried, and what i learned (ie the things i'm glad to have learned about life, not necessarily the biggest takeaways of each field):

  • the earliest creative pursuit i remember was making youtube videos: first about gaming, later skits, commentary, short films, cooking tutorials, and many more random things
    • especially early in life, i think it's really important to learn the 'full stack' of something: in this case the end-to-end optimization of filming, editing, publishing, building communities, analyzing to 'make number go up.' doing crazy amounts of varied work yourself builds intuition for how categories of choices affect complex systems.
    • i never had a lot of viewers, so most 'improvement' in my content came from self-feedback. comparing myself with videos i liked, i realized lots of subtle things greatly affect engagement: fonts, word choice, microphone quality (this is much more obvious, but surprisingly overlooked), scene-to-scene transitions. most viewers don't actually care about these technical details, but they feel the vibe and atmosphere created by adding them together. it's important to be aware of the atmosphere you want, and build it with intention.
    • in my eyes, art requires a certain lower bound of 'professionalism.' it's important to get infuriated by watermarks and ugly defaults, then know how to mask it away with the color match stamp brush thing on an online photoshop ripoff. once that bound is met (ie the unequivocally bad parts are gone) you're free to bend and break rules however you want. recognize that the goal isn't to make something technically perfect but to elicit feeling. it's always been to elicit feeling. feelings can be elicited in many different ways.
    • after ~3 years of inconsistent posting i abandoned the channel, but i'm often reminded just how valuable these specific skills of digital media manipulation are. during online school in 2021, photoshop edits i cooked up brought people together. across the years, the ability to make watchable homemade films has made many group projects memorable.
    • more from consuming videos than making them: i'm glad i learned how to tell stories. because everything online has already survived some test of attention, (actively) consuming heaps of interesting content gives you intuitions about the underlying qualities of 'interesting-ness.' i think it has something to do with the fact that a lot of our experiences and realizations have this inherent magical texture to them, like the felt absurdity of realizing something obvious in retrospect, or the relief of solving a deeply elusive problem. 'interesting-ness' is the ability to capture and convey with words the magic you've experienced. practically: create tension, and try to make emotions felt by those in the story stupid obvious. anything can be interesting if told properly.
    • i've also learned it's really hard to come up with good names for things 😆; i changed my channel name maybe 9 different times
  • eventually, i found myself programming (or to be more specific manipulating computers)
    • i got my start creating automations for boring, annoying processes: filling out free robux surveys, grinding in videogames, etc. primarily, i'd use autoclickers to record and replay sequences of mouse actions—pressing on app 1 to focus it, pressing a specific button, focusing app 2, pressing another specific button, etc. eventually, i realized i could create much more effective and controllable workflows by writing my own code.
    • though i've been thinking about how to make systems 'programmable' (how to break complex problems into steps computers can reliably execute) for a pretty long time, i wouldn't consider myself a 'cs kid.' i can't recall 'falling in love' with writing code, for me it's more of a means to an end; i want to achieve something in the digital world, and if it's more practical to, say, build a lego mechanism to move my mouse in specific increments, i'll do that instead.
    • when i was learning to make videogames, i discovered a really effective strategy for learning something completely new was to find tutorials making something similar to what i wanted, copy the tutorial exactly until i understood the tools enough, then stop watching and build out my new ideas using what i copied as a starting point.
  • somewhere around 2020, i really got into cooking.
    • making food and playing with ingredients in hack-y ways (like using flour to fix over-salting or espresso to enhance chocolate-y flavors) is really fun. it's very important to know what different ingredients and methods do, because then you can combine them to achieve new outcomes. it's important to understand the rules so deeply that breaking them makes complete sense.
    • 'cuisines' as a constraining construct are fake; you can just make interesting foods.
    • the ability to feel out flavors is much more important than the ability to measure precisely.
  • perhaps one of my most technically and personally formative life experiences is teaching myself to produce electronic music.
    • growing up i saw a lot of clips of people DJing and i wanted to be like that. when i was twelve, i discovered the world of music production (NOTE: making music is not the same as DJing. i realized making music was cooler...) and so i downloaded a music production software (aka a 'daw') and started making music, despite knowing nothing about music. with only youtube and sheer trial-and-error, i had to teach myself to use complex software, understand music theory, use synthesizers to create sounds, and hear deeper—all at the same time and all from essentially zero. even today i can't really play piano or read sheet music, but i've developed an intuitive feel for how sounds fit together.
    • the starving musician experience is/was also really exciting. at first i used a cheap, heavily limited $8 version of my daw (ableton live lite gang rise up), but eventually i discovered there was a 90 day free trial of the unrestricted version of the daw. after 90 days, i discovered if you found and wiped some local data you could get another 90 day trial with a new email. i used this for about two years, after which it stopped working. when your trial ended you could still make music, but you couldn't save it, so when i liked my creations i'd screenshot every configuration of every instrument and audio effect and saved it in pdfs, so that one day when i got the full version i could remake my creations. growing up, i was taught to be super rational about money and only buy things that are absolutely necessary, so i didn't buy a real license—even though being separated from music was tearing me apart—because i knew i could live without it. eventually i had enough and bought a real license of the unlimited ableton live suite, which i still use to this day. to clarify, money wasn't ever the problem (i'd built up a small fortune from Some Other Ventures) i held out because i was using sheer, misdirected willpower to stick to the principles i'd deeply internalized. in retrospect, my months of suffering were absolutely not worth it, especially because the money was already slated to be spent—whether today or in five years i knew i was going to buy the software anyway. it's not worth pushing off the inevitable. dreams deferred dry up.
    • music production is a really good entry point to many different technical specializations: signal processing, latency-sensitive programming, analog circuitry, machine learning, and lots more. i've explored all these areas and permutations of these areas in at least some capacity, though i now focus on machine learning.
    • making music for such a significant portion of my life, i've developed a very personal connection with sound. i believe sound and music are deeply connected to how my mind is wired and how i perceive the world. i love the feeling of warmth, being surrounded and cradled by sound, escaping my body and realizing nothing matters but what i hear right now. i love the feeling of awe, my mind exploding when i hear something i can't begin to comprehend the making of. i love discovering new things in the songs i love. but this blog post is already very long, so i will save these theses (and other random music production stories) for another day :)
    • interacting with lots of europeans (particularly french, german, and dutch people) online has irrevocably changed how i pronounce lots of words :)
  • throughout my life, i've built and fixed things with crude materials.
    • binder clips and staples are very versatile tools. one time, i fixed my broken backpack zipper by threading staples through the zipper loops and adjacent fabric, then folding the staple shut. because the staples were just thin enough, the slider could still pass over them and fasten shut. if you understand at a very low level what needs to be done, it's very easy to make a solution with any object or material. think tony stark in the cave.
    • something i noticed recently is a 'soft' tradeoff between 'love' and 'hygiene.' at my senior prom, my fingers felt embarrassingly nude, so i spent the entire event clutching and fidgeting with a cheap eiffel tower keychain i got at the event's entrance. eventually, i got frustrated by the keychain being slightly too long to clasp the tower perfectly in my palm, so i started taking apart links. a common way to do this is with your teeth—since they can generate more leverage on tiny, tightly clasped pieces—which immediately feels very 'dirty' to me. but i then thought of it as me putting my personhood (and i guess on a more abstract level my 'love') into the objects i keep close, and the feeling of that intimacy outweighed my other concerns.
    • i've learned a bunch of specific quirks of plastics, metals, and paper/cardboard, most of which i don't readily remember but trust will come back when i need them to.
  • in high school i joined a competitive robotics team (btw frc mogs everything else).
    • this is another thing that have its own ten blog posts. overall, being in this environment helped me formalize some of my intuitions. my robotics mentors taught me how to actually write code, not just chicken-scratch python scripts that happen to execute (the fix was mostly to lint & use git—i lowkey been vibe coding since even before LLMs). they showed me how to think through and sketch out mechanical systems more complex than anything i'd previously thought about. rather than feeling through things, they showed me how important it is to be principled, think about best practices, and work backwards from clearly defined goals.
    • this is a bit silly, but i remember entering my team with extremely misguided notions of precision. because i viewed myself as finally in the big leagues, i'd get hung up over mechanism ideas that couldn't achieve tail-end performance on everything ('it needs to carry 155 balls, 150 isn't enough'), or when preparing aluminum tubing for assembly i'd mark and re-mark and re-mark where to drill to avoid being even 1/64th of an inch off. i've discovered that in the real world, precision has upper limits: carrying more than 140 balls causes the drivetrain to draw too much power, and anyway we only have time to pick up 110 in matches; our rickety old drill press never travels perfectly straight down. engineering is about making good judgment calls, and inherently comes with doing lots of 'handwave-y' things. it's important to calibrate precision to the task at hand; some things (like mounting beam-break sensors) inherently require more precision than others (like riveting metal plates together)
  • after getting my first phone at 14, i realized i really enjoy taking photos.
    • i always hated taking photos, but after getting my own phone i discovered that i naturally see things compositionally, and often notice when something i see could make a cool photograph. i realize i never hated photography itself, just how my family members' obsession with 'taking photos' interrupts actual lived experience. for me, photos are more about capturing moments than taking a perfect, manufactured smiling portrait. i still occasionally take selfies and posed photos of my own volition, but i'll never interrupt a good moment to do so, and it'll never be the reason i go somewhere or do something. i'll start a photography archive soon 😄
    • https://x.com/01shuklabhay/status/2049791857941745683
  • occasionally i will run, bike, hike, or otherwise engage in physical activities.
    • i really enjoy that feeling where nothing else matters but the road in front of you, and the actions your body is taking to produce movement are completely out of sight, out of mind. conceptually, i like that the only barrier to running faster, pedaling harder, is the voice in my mind telling me i'm at my maximum. i like to tell it to shut up.
  • most recently, i've been training neural networks.
    • next fall, i will begin my first year of university studying 'computer science.' rather than traditional 'computer science,' i'm fascinated on many levels by the process of training neural networks. the math is interesting and the systems optimization-y problems around efficiently using compute are fun, but on a humanistic level there's something so incredible about the idea that we can take a slice of how humans perceive the world and condense it into a single algorithm. on a philosophical level there's something so uncanny about the idea that we, for the first time in human history, seem to have access to plato's world of forms.
    • being fascinated by a tool rather than a goal slightly contradicts some of my earlier ramblings, but there really is so much more that's possible with neural networks—in the digital and physical worlds.

core realizations, closing notes

there are many more things i'm far too detached from to properly remember (animating, 3D modeling, being in a play, cartography, being in a yt automation farm, speaking/debating), and many more things i'm just now getting into (fabrics), but writing all this out i realize there are three underlying 'themes' of my life.

first, serendipity. encountering new ideas requires some inherent randomness. in biology, evolution is driven by random mutations. in my case, recommendation algorithms pushed me toward new ideas. in real life, this could mean going outside and meeting people and pushing yourself to be 100% even if you feel 95%. maybe you WILL meet your wife during your ft. lauderdale spring break trip 😁😁. i mean to say that i believe life is serendipitous, and a wonderful set of coincidences. but you must first give serendipity opportunities to happen; the greatest gift is to be present, because presence allows you to be conscious of just how much the world has to offer to you, and just how much you can offer it in return.

the second overarching idea is to pay attention. noticing is an art. to give attention is to love. despite having little prior or formal knowledge, i find myself constantly reading super technical writing on machine learning, game theory, rl, social theory, politics. i approach learning from the top down: i start with a complex idea and piece together all the foundations i need to understand it properly. when you understand what things are at their most basic levels, you can take control—build them from scratch, make them work in new contexts, whatever. but understanding first is important. to effectively break rules, they must first make complete sense (i speak in absolutes because they're poetic; this is usually but not always true). i implore you to look deeper than what's visible at first glance, because it'll train your first glances to capture more and more of life's elegant beauty. sometimes i listen to the specific ways people say their sentences; lots of people have very distinct syntactical signatures, shaped by their lived experiences.

third, and perhaps most important, is to recognize just how transferrable good principles are. love operating roughly inversely to hygiene is as true for metal-bending as it is for food. my learnings of precision's upper bounds apply just as much to bryan johnson's duels with turf. clean code and clean cad follow a lot of the same principles (define constraints, abstract purposefully, compartmentalize). good principles are ubiquitous because they target underlying reality of... reality, rather than any particular bitstream (technically everything is a bit, so this metaphor works :p). it's important to build intuitions and recognize when they transfer, or more generally to know things you can't name. even as i write this, i'm learning so many names for things i can describe but can't identify. life is allll about pattern matching.

expanding outwards into social commentary, i think it's really important for us to recognize the extent to which we condense and classify and segregate based on imaginary notions of what should and shouldn't go together. most dichotomies are deeply untrue. mathematics, for example, emerged as an evidence-based perspective in philosophical debate.

i like to define people, things, and myself with adjectives rather than nouns: 'curious,' 'relentless,' 'humanistic,' 'intrinsically motivated.' even if it doesn't fit cleanly into a single story, i believe it's important to maximally embody yourself: be the person who explores everything they're interested in, be the person who shoots twitter DM airball after airball to meet people who seem interesting, be the person who moshes while screaming to anyone who'll listen about how the pushing forces you're experiencing can be modeled with differential equations. be whoever you want to be and do whatever you want to do and don't be afraid to leave your footprint in the world.

resist categorization, become incompressible.

afterward

exploring my interests and building broad intuitions has brought me tremendous technical and personal growth, and i highly recommend doing so to anyone my age, but i think it's also important to be aware of the tradeoff with impact. because i've explored so much i've learned a lot and built a little in everything i'm interested in, but i've also never been able to scale anything up to the degree i want to/feel i can reach. historically, this hasn't bothered me—my goal has always been to understand the world around me—but as i enter adulthood i realize i want to live a life where i've burnt off as much of my potential as possible.

right now though, i feel blocked by an immense psychological burn, which i think stems from the fact that i've had to explore almost all of my interests independently. i hypothesize that this burn is mostly an issue of environment: around the right people everything feels so different. in 11th grade i became close with a ML hyperscaler at my school, and realized it's not weird at all to be so particular about the things you care about; the summer after 10th grade at a ucla summer program i realized how quickly i could pick up new things, and how easy it is to feel 'accepted' with the right people.

something i've realized talking to some friends is that impact is bottlenecked not by what you can or can't do, but how fast you can do it. sure, one might have the skillset and patience to run every department of a company independently, but at the end of the day all that matters is how fast one can go from having an idea to changing lives. so, even though we've built up this capacity to do everything with a high degree of competence, the road to impact is to relentlessly curate: relinquish control, occasionally advise, and actively choose to not do things—because there are other things even more worth doing.

i suppose it's important to know how to do everything because it teaches you that you can be. once you wholeheartedly believe you can be anyone and do anything, you're left with the harder, more important question: what is worth becoming?

credit, inspiration:

sometimes i think of myself as nothing more than a filter, building my worldview by sorting and synthesizing perspectives from everyone i encounter (in person, through writing, or otherwise).

this work was heavily inspired by writing from: jason liu, rami seid, ashray gupta, andrew wu, hannah gao, tiger zhang, brangus, alvin djajadikerta, will savage

and conversations with: avni iyer, max nguyen, nancy zuo, brady gho, neel redkar, hudhafaya nazoordeen, maria kostylew, ruben stephen, ayaan bhambri, pinak paliwal, david yu, yan zhang, ulisse mini, aiden jeong, andrew wu, aaron lin, gloria ma, michelle la, bhargav nemani, ewan bailey

fwiw, this song was also playing in my head during much of the writing process: https://open.spotify.com/track/6FororaZLRknUirUap7cyh?si=d9de8f1561db4ee6