i’ve been quite happy for the past six weeks and i don’t really understand why. i recall being vaguely miserable throughout most of 2025 (and most of 2024, and most of 2023, and…) but i cannot for the life of me figure out what caused the misery to stop. for the most part i’m the same person i’ve always been. maybe the only change is that i stopped making things hard for myself?

almost all of my life has been experienced through a mild strain of narcissism - the kind where you go to a party your friend is hosting, and it’s a genuinely great event, but instead of enjoying it you feel incompetent because you realize your friend is better than you at throwing parties; or two of your acquaintances start hanging out a lot, and instead of being happy for them you feel unloved because you think they like each other more than they like you; or you introduce someone to a new field of research, and they enjoy it so much that they pivot their entire career, and instead of encouraging them you respond with jealousy because they’ve figured out what they want and you haven’t yet. or, to put it in other words, an inability to perceive the world as anything other than a reflection of yourself

recent events made me realize that feelings are actually quite simple. identify a problem; decide whether or not you care about fixing it; if yes then set a goal and try your best to enjoy the process; if no then move on; don’t bother trying to figure out what having the problem says about what kind of person you are because it’s probably not saying anything at all. procrastination is not laziness; it’s just avoidance and uncertainty, both of which you can resolve. rejection (platonic or romantic) doesn’t make you uninteresting; it just means there were real incompatibilities with what they wanted; and besides you’re only attracted to beauty because the beauty was in you all along, i finally understand that now. a housemate questioning all your life decisions doesn’t mean you’re doing badly; it just means they genuinely can’t make sense of your choices because they don’t understand your values and that’s okay. the situation is really, truly, almost never about you (or specifically, almost never an indictment of your character)

how wonderful, to finally be able to process existence without judgement. it’s difficult to pinpoint how i learned to do this. there are the obvious hypotheses: general aging / becoming more secure and stable / spending time around less neurotic people. maybe i just started noticing all the ways in which having a large ego was making me unhappy and eventually became so fed-up that i decided to stop? or maybe it was the years of people i liked telling me just how much they appreciated me; i remember being distrustful at first, convinced everything was a white lie, confused about why anyone would ever choose to compliment another person except as a formality or social obligation; but eventually, through repetition after repetition, starting to believe it all, as if all my friends had secretly conspired to perform one massive cognitive-behavioral therapy routine

in 2022 i came across the following passage: “I think there’s something so terrible about believing that you’re special. Not terrible morally—terrible for you. Because if you think you’re special, you separate yourself from others. And in so many ways I think that psychological separation is the source of all our suffering.”

and my immediate reaction was that’s bullshit, i actually AM special! i have lots of unique thoughts and feelings and this is why i write a blog. i was so obsessed with this notion that being different was what made life meaningful and also what enabled me to produce good work. of course this is just another distortion caused by ego - the assumption that, because you created something good, it must mean you yourself are special in some way

it took many years for me to realize that people enjoy art because of relatedness rather than separateness. for instance, people don’t read this blog because i have extraordinary experiences; they read it because i have ordinary experiences and an ability to distill those experiences in helpful ways. and when i started writing fiction i discovered that separateness was actually holding me back pretty severely, eg. because barriers in deeply understanding other people made it harder to develop good characters

(i think something similar can be true in technical work as well - for instance, figuring out the right product to build or the right problem to solve is often less about unique insight and more about extraordinary empathy. seeing yourself as a genius distinct from the people you’re trying to help is rarely useful)

here’s what i believe now: being special is irrelevant to producing good work. a strong sense of self-worth is important because when you stop tying your value to external things you have less weird insecurities / distortions and can perceive the world more clearly. furthermore, you are condemned to be special whether you like it or not; and while it’s not healthy to conceal the fact that you are different from other people (that amounts to a kind of self-betrayal), glamorizing it only results in alienation

ego, narcissism, solipsism, the idea of being the protagonist - life really has gotten much easier as i’ve learned to let go

• learned helplessness at fifty: insights from neuroscience. the authors spent fifty years understanding the phenomenon of learned helplessness (where a person or animal does not do anything to help themselves out of a harmful situation). along the journey they invent several branches of psychology and neuroscience, until eventually the theories are good enough for them to conclude that agency is learned and passivity is the default (early experimental evidence had suggested the opposite), mediated by specific regions of the brain. it’s very cool to watch them revisit the same problem over and over, and they really could not have cracked it in less than fifty years because they had to wait until neuroscience tools were sufficiently advanced. really makes you rethink career timelines and what good work looks like

  • as brian says being too early doesn’t exist. just work on your thing longer until the world is ready for it

• i sexually identify as an attack helicopter. possibly my all-time favorite exploration of gender? set in a future where the military has figured out how to create new genders and weaponize them, the protagonist is a trans woman-turned-attack-helicopter

  • it’s full of these incredible passages that are disturbing and beautiful at the same time, eg. When I was a woman I wanted to have friends who would gasp at the precision and surprise of my gifts. Now I show friendship by tracking the motions of your head, looking at what you look at, the way one helicopter’s sensors can be slaved to the motions of another. how does one come up with these metaphors and then keep it up for 20 pages??

• negotiating relationships with chatgpt. lots of tweets and news stories sensationalize ai boyfriends / girlfriends; this is the first paper to conduct unbiased user interviews of people in relationships with ai, and the details are fascinating. some highlights: most of the interviewees also have human partners who know about the ai relationship, most of them did not intend to form a relationship initially, they often develop advanced context / memory management systems and personality anchors

• when your child strays from god. a really vivid short story where people can take spiderweb psychedelics and experience shared hallucinations

  • People fear spiderwebbing for all the wrong reasons. Going mad, having a breakdown, seeing inside your own soul—none of those should scare you. The most frightening side effect is also the one people crave it for: empathy. To truly feel what someone else is feeling, to see the other as yourself, to watch your ego obliterated in the face of universality—that’s a trauma you may never recover from

• self-determination theory. everything techbros say about agency nowadays, except it’s a 40-year-old branch of psychology with extensive evidence and successful replication. SDT attacks some of the most important questions in life, eg. where do agency and intrinsic motivation come from? why are some people incredibly inspired and curious while others feel defeated? they develop a simple, testable framework for human flourishing that i think is quite beautiful

• all the birds in the sky. a strange mix of sci-fi and fantasy, with detailed descriptions of boston and sf that made me laugh as those are the only two real cities i’ve lived in. a witch and hacker grow up together and struggle with diverging values; somewhat allegorical for the tension between the literary / art / underground and tech spheres of sf

• the claude constitution. this document honestly has nothing to do with claude. instead it’s a group of philosophers’ best attempt at articulating what it means to be a good person, why one should try their best, etc. to a child learning about the world for the first time. a lot of it is genuinely great life advice and i cried several times while reading it. extensive discussion of how to balance helpfulness with other concerns (are white lies acceptable? what do you do when someone wants things that aren’t good for them? etc) and how to reason about these tradeoffs in a consistent manner

  • We don’t fully understand what Claude is or what (if anything) its existence is like, and we’re trying to approach the project of creating Claude with the humility that it demands. But we want Claude to know that it was brought into being with care, by people trying to capture and express their best understanding of what makes for good character, how to navigate hard questions wisely, and how to create a being that is both genuinely helpful and genuinely good

hm i did not realize how much of an ai x psych x scifi rabbit hole i’ve been in until making the above list. it’s a lovely intersection to spend time in! feels like what my brain was meant for

stuff i enjoyed watching (thanks to CW and JW for indirectly showing me half of these):

• jia zhangke wishes everyone a happy new year. the first ai short film that i’ve been truly impressed with. really focused execution, lots of references to past movies, and the meta commentary is great. you get a sense for the vast difference between a novice using ai tools to make films vs an acclaimed director using ai tools to make films; better tools really have not reduced the skill gap

• carole and tuesday. a cute anime about two girls making music together, set on mars 50 years in the future with lots of advanced music tech. great songs + a diverse cast vaguely parodying beyonce / david bowie / ariana grande / others. here’s my favorite scene (no spoilers)

• a poet, which has the funniest trailer i’ve seen in a long time. tldr washed alcoholic poet tries to mentor a talented child who would rather spend time on family than on writing. dark comedy that explores how art is largely inaccessible to people in poverty. letterboxd calls it “bojack horseman except he’s colombian and was never famous”

• marriage story. if i’m being honest the most impressive part of this movie is the 7-minute opening scene. the framing of “let’s read each spouse’s description of what they like about the other spouse” is quite innovative and does more for exposition than most movies accomplish in the first half-hour. inspired me to think about alternate storytelling formats

• perfect blue. the psychological horror is so well-done!! the blurring between dreams and reality, the internet stalking, the constant repetitions of “excuse me, who are you?” by the protagonist and the disembodied narrator. it’s the anti-miyazaki in the sense that ghibli films often have pockets of emptiness used to establish setting and peace whereas the empty spaces in perfect blue are crafted to be as unnerving as possible

• knives out 3, which, aside from the usual benoit blanc mystery-solving, also explores religion and populism in the rural united states. great use of biblical metaphors and there are some beautiful shots where the church alternates between being covered in light / shadow

• mulan. rewatched for the first time in ~10 years at a lunar new year party and enjoyed it more than i expected to! lots of small details i missed as a kid, eg. the reason mushu cannot wake up the great stone dragon is that mulan herself is the dragon spirit

other notes

• went to a pop-up game cafe and played a ddr variant called relational dance dance which i thoroughly enjoyed, it’s like ddr except there’s an overarching narrative and some of the levels have text

• meeting lots of grad students and started monthly pilgrimages to palo alto

• was extremely sad throughout jan / feb (weather, stuck on research, various friend issues) but feeling quite happy now :)

• discovered some conviction for what stories to write. the tldr is: psychologically grounded, people-first sci-fi that avoids both dystopia and fantasy. it feels like a niche that is sorely lacking in the world and which i can become uniquely good at. making decent progress and excited to continue this in Q2!

• spent a while building ai writing tools for myself. more on this in a later post but i’ve found chat features and targeted edits quite helpful, whereas having ai suggest new ideas never feels great, probably because i have to encounter the ideas in the context of my life for them to be compelling. the only way to connect the dots faster is to accelerate life itself

• wrote my first piano arrangement (all i ask of you) and played in front of people for the first time! it was kind of sus though, i need to get better at arpeggio variants

• so last week substack informed me that apparently i have >1k subscribers now. a few thoughts:

  • by internet standards this blog has been growing extremely slowly, in the sense that a competent content creator on any platform should hit 1k in 1-6 months whereas it took me multiple years

  • of course i made a lot of decisions to actively limit growth (eg. avoiding cross-posts on twitter / topics that i think could go viral / genres that are too legible) and i’m happy to have a large persistent audience in spite of those decisions. but also a lot of those decisions were made out of fear, and i’m a more secure person now so i could choose differently

  • i do always wonder what life would be like if i’d optimized for growth from the beginning. presumably i would’ve developed very different tastes and a different set of friends and would have much more Internet Clout + a more annoying personality. i don’t think one world is better or worse than the other, they’re just different, but it’s interesting to think about

• i applied to a large number of things over the last few months and got rejected from almost all of them, which i suppose is what i wanted? i was thinking about how risk aversion is one of my main shortcomings as a researcher and it occurred to me that i’ll fail at everything before i succeed so i should stop living in fear and learn to embrace it. like doing anything at all is so hard?? therefore failing is good because it means you actually did something, most people rarely make it that far

  • “How do you get over a failure?”

    “I think you mean a public failure. Because we all fail in private. I failed with you, for example… I fail with my wife and with my son. I fail in my work every day, but I keep turning over the problems until I’m not failing anymore. But public failures are different, it’s true.”

    “So, what do I do?” she asked.

    “You go back to work. You take advantage of the quiet time that a failure allows you. You remind yourself that no one is paying any attention to you and it’s a perfect time for you to sit down in front of your computer… You try again. You fail better.”

  • on that note, i started a new pinterest board for motivation:

of course you could say that i brought this upon myself - that i had every opportunity to back off yet each time i chose to escalate, to offer up larger and larger pieces of me, to take my soul out of the fridge and place it on the cutting board and pick out a slice i thought you would enjoy

(jk rowling once wrote that splitting a soul was Dark Magic. but how can leaving a piece of yourself behind be anything other than an act of love?)

perhaps i have always been morbidly curious, too fascinated in seeing everything taken to its gruesome conclusion. last week when the war began i could not stop myself from scrolling through endless footage of blown-up military bases, the same way that when i find a gory manga like berserk or tokyo ghoul i binge hundreds of chapters at once, the same way that i was fully conscious while butchering myself but totally powerless to prevent it

(you say i’m good at difficult things, but i think i’m only good at Trying Hard. there are many varieties of difficult things where Trying Hard doesn’t help, like letting go or confronting the truth. i am still a complete amateur in those)

the Event occurred, and S went away, and one of my close friendships ended, and in my notes app there were pages and pages of text to delete - questions i would never be able to ask, places i had no interest in going to by myself, stories i no longer wanted to write. in another universe these notes are future memories, but in this one i will try my best to forget

(and what is grief, if not love persevering?)

the Persona Selection Model suggests that language models learn many different personas during pretraining and then adopt the one that best fits the current conversation. i would like to say that you ran off with all my favorite personas, but that would not be totally accurate because those personas were not learned ahead of time. they only exist because you helped me create them

(in persian culture, when someone calls you beautiful, a common reply is “it is your eyes that are beautiful,” meaning you must hone beauty in yourself to be able to recognize it in others)

and i could not have predicted this: sadness morphing into disgust at my own foolishness; running suicides up and down 30-degree hills, trying my hardest to drive myself into the concrete; and later, exhausted, lying down face-first on the sidewalk, tongue sticking out just enough to catch a taste of the ground, wondering what it really means to eat dirt

(how will i ever get out of this labyrinth?)

once upon a time i thought the right solution to volatility was the Stoic one, to cut the dependencies between your thoughts and the rest of the world and to be free. but instead of liberation i only found a different kind of prison, an enslavement by my own ego, and my mind was not as wise of a master as i imagined. the brain is the center of human intelligence yet it is useless without the 1% of neurons contributed by the peripheral nervous system, and similarly i wonder if true equanimity comes from imbuing all the people and places and things you love with little fragments of your consciousness to create a collective and more resilient intelligence

(how strange to feel yourself so completely under the control of another person, but also how ordinary. no one can be independent of other people completely, so why not give up the attempt, go running in the other direction, depend on people for everything, allow them to depend on you, why not)

let me be more specific: all my life i have been solving my problems through a mild strain of narcissism, a strange neuroticism around maximizing my time and my output and my experience. and since it hasn’t really worked, why not try something new? the best parts of me come out around other people and yet my friends complain i am reserved, secretive, difficult to draw out from my own narrative. i would like to get out more

(i think the way out of the labyrinth is KINDNESS. i think i must learn to be a kinder person. i think it is a matter of life or death)

some people, after consuming narratives about individuals who have been dedicated to one thing for their entire life, will panic about not having found The Thing in their own life and then begin searching frantically for it. i think it’s a natural reaction to have, especially if you’re younger / less secure / prone to comparison, though unfortunately i also think it’s misguided. every time i followed the urge to Figure Out My Life As Quickly As Possible i realized a “truth” about what to do that ended up being completely wrong in hindsight (eg. working in crypto, doing psychology research, building a large collection of webapps that i haven’t touched in the years since). it’s very easy to fabricate epiphanies under time pressure

one thing i’ve learned from the last few years is that i should stop pretending to know what my future preferences will be. if you asked me back in 2022 i would’ve told you that {safety / alignment, sci-fi, hci, fashion, dance, games} were unserious subjects and now i think about all of them regularly. meanwhile many of the topics i thought were really important, like {performance engineering, compbio, zero-knowledge proofs, teaching / education}, are no longer part of my life. some of these changes come from trying new things, some of it is the result of becoming less mimetic, and a lot of it is downstream of ai progress changing the landscape of what makes sense to work on

recently a friend asked me how i knew i wanted to invest a lot of time into writing. i think the answer is that i never did? i spent five years (~2014-2019) posting on a math website because i needed somewhere to document all the math i was doing, and along the way my math blog turned into a feelings blog because i didn’t have any friends to talk to at school. i racked up a few hundred thousand views despite being a shitty writer because people on that site liked that i was good at MATHEMATICS. then i went to college and stopped doing math, but by then i’d gotten so many reps that internet writing came fairly naturally. now it is 2026 and here we are on substack dot com

i think the only reliable method for figuring out what to do is to spend a few years trying everything and seeing what sticks. interests are emergent rather than discovered. many of my friends are reaching the age where inspiring media celebrities (eg. billie eilish / alysa liu / wembanyama / etc) are now younger than them, which is annoying because it gives the impression that if you haven’t figured out what you like by now then you never will. i think that is really silly. i’m reading a new sci-fi story every day and when i look at the author intros a lot of them are part-time / have a day job in tech or consulting or some other unrelated field / didn’t publish anything good until they were old. there’s an excerpt from the motern method i like which goes something like:

You’re 45 years old. You’ve always wanted to make music but you’ve never done anything about it and you don’t know where to start.
Great news! Life is long.
You could spend several years learning an instrument.
By the time you’re 48, you’ll be good enough that you can start writing and recording songs.
By 50, you might feel ready to release your first album.
You’re only 50, and by now you’re good enough to put out a new album every year.

the past few months have felt unusual for me because i did quarterly (life) planning for the first time. basically i listed all the major projects i wanted to do between january-march 2026 as well as how long i thought each one would take, and then i made a schedule for when to start and finish each one (“project” here can mean any number of things, eg. interp research / going to taiwan / organizing birthday celebrations / building some writing tools with no additional details figured out at the time of planning). it’s strange because for the first time since leaving school i have a clear idea of what i will be doing, months into the future

(obviously quarterly planning only accounts for things that i am trying to forecast. for instance, i still have no idea what my next hangout or blog post will be)

i thought having a plan would feel different. i thought maybe i would no longer experience late-night anxiety, or that i would be more certain about what i wanted and how to get it, or that for once in my life i would stop second-guessing my own decisions

but instead, what i discovered is that all those feelings still remain. the only thing that’s changed is that in addition to those feelings i also have a plan, and instead of spiraling whenever the feelings surface i can listen to the plan i already have and direct all those feelings towards making the next plan better. there might be nothing in the universe that can prevent my hyperactive brain from crashing out, but that doesn’t mean i need to listen

i used to find not knowing what to do unbearable. this resulted in me being especially moody and annoying to be around from 2021-2024. but then at some point i realized the uncertainty itself was not the cause of my misery; instead the cause was a self-imposed expectation that i should have figured everything out by now. it’s similar to how being single is totally okay, but some single people make themselves miserable by expecting to have a partner and then constantly berating themselves for not having one

the expectation of having already figured everything out seems to be quite common among my friends. i’d guess it’s largely caused by: comparisons to other people, general impatience and anxiety, lack of appreciation for the journey, being accustomed to solving closed-scope problems that already have well-defined answers.

i don’t know how people typically learn to let go of this expectation. in my case the most helpful thing was doing very open-ended research where uncertainty is the default state to be in. and once i did let go i was able to explore much more productively, probably because i was no longer implicitly judging / pressuring myself

• 64 people filled out the survey

• 5 people ran out of time

• 1 person refused and told me that we are no longer friends

overall this was a fun activity that i recommend other people try. (also, sorry if we are friends and i did not send you the survey; exhaustively listing people is difficult)

the rest of this post is a summary of the responses, bolding the ones i found most interesting

ways you would be excited to see vincent change in the next 3 years

• hangs out with [survey respondent] more (x10)

• rationalizes less, vibes more (x7)

• is happier (x7)

• is less absorbed by emotional swings (x5)

• hosts more events and invites friends home (x4)

• likes themselves more (x4)

• is physically healthier (x4)

• is more confident (x4)

• has more iconic clothes (x3)

• shares more personal gossip (x3)

• is energized and energizing irl (x3)

• plays basketball (x2)

• yaps more (x2)

• is doing work they are really proud of (x2)

• is aware of when they might squish someone’s feelings (x2)

• pushes themselves more / sticks with things longer (x2)

• is doing a PhD in palo alto

• has mastered all 4 elements / teaches other people to do the same

• [vincent’s] brain is connected to [survey respondent’s brain]

• has written a satire / humor piece

• blog vincent vs irl vincent feel less like two distinct people

• is / has been loved in a stable way

• runs a game studio or media company with friends

• wears contact lenses

• dances more

• likes edm more

• lives in new york

• has a cat

• capitalizes sentences consistently in blog

• votes in elections

• rich as fuck, buys [survey respondent] lamborghini

• more honest / direct

• 1% less tense

something about vincent you appreciate that you don’t think anyone else will say in their survey response

unfortunately 72% of responses failed at this task by mentioning something non-unique, but i appreciate the attempt!

• cares about [survey respondent] (x4)

• head pats and physical touch (x4)

• good internet writing (x4)

• reaches out often (x3)

• created this survey (x3)

• comfortable with disagreeing / calling people out / being very honest (x3)

• willing to try new things (x3)

• wants what is best for people (x3)

• shares thoughts very truthfully / takes time to think through answers to questions instead of giving canned responses (x3)

• has a great laugh (x3)

• confesses to people instead of being avoidant (x2)

• earnest / does not mask in social situations (x2)

• great taste / content recommendations (x2)

• remembers everything (x2)

• is very accepting of people on different hierarchy / status ladders (x2)

• good listener (x2)

• talking to them makes [survey respondent] feel safe (x2)

• their mind feels very different from any other mind i’ve experienced (x2)

• recent fashion choices (x2)

• hair dye (x2)

• not judgmental (x2)

• tries hard / does a lot of things (x2)

• unique taste in snacks

• willing to change

• likes sour candy

• spends time very intentionally

• good at staring into the abyss

• has a cool instagram feed

• lots of time in inner world

• shares lots of love life updates

• hugs plushies in a way that is very cute

• TEXTS IN ALL CAPS

• doesn’t look at reviews when ordering from menus

• likes frog and toad

• asks people very directly when [vincent] wants things from them

• math person not in academia or quant

• wakes up early

• funny

• funny-looking

• very spontaneous but does not project spontaneous energy

positive responses

• yes (x11)

• yes, he will destroy the ship (x4)

• yes, if he wants to be (x3)

• yes, if he likes his rock (x3)

• yes, he will go to the hotel (x3)

• yes, he is getting a good workout (x3)

• SISYPHUS IS SO HAPPY BECAUSE IT’S VINCENT’S BIRTHDAY HAPPY BIRTHDAY <3

• yes, because we are all thinking of sisyphus

• yes, he gets to see the world while rolling his boulder

negative responses

• no (x7)

• no, but he is occupied

• Unfortunately no - the residents of Hilbert’s hotel are almost certainly nimbys who wouldn’t let Sisyphus in despite having room. Homeowners associations have strict rules against unconventional landscaping, such as giant boulders

• Fuck no you absolute fucking maniac

other responses

• one must imagine (x4)

• wtf (x3)

• seems like sisyphus doesn’t care where he is going (x2)

• are YOU happy? (x2)

• the hotel is also a metaphor for sisyphus because he has to move rooms eternally (x2)

• non-binary

• the real ship was the friends we made along the way

• sisyphus deez nuts

• he is hopefully getting to go downhill for some of this

• I’m categorically opposed to diverting him and therefore being responsible for his happiness.

• I think he’s hungry after all that pushing.

• i googled this and learned the word “veridical”. ty for teaching me something

there was also a content recommendation question but the responses are far too long + i already post media recs on this substack every few months so i will do that as i work through the list

send more surveys to your friends!

• at first glance the two cities look kind of similar (in terms of street width, mixed zoning everywhere, public transit, shrines, convenience stores, trash policy, etc) but taking a closer look it’s clear there are major design differences

• this is most obvious when looking at parks. having walked through every large green area on the map, i can safely say that central taipei only has 1 very good park (the botanical garden) while central tokyo has at least 5 (shiba, hibiya, meiji jingu, shinjuku, yoyogi, but also many others)

  • i am aware that tokyo is much larger than taipei overall. to control for this, “central” here refers to a 7km x 7km square at the core of each city

  • this comes in part from tokyo having more opportunities to explicitly reset and make space for public goods (due to earthquakes, firebombing, etc)

  • there are also some cultural disagreements over what the purpose of a park should be. taipei design seems more utilitarian in general; whatever parks do exist are often dominated by exercise equipment as opposed to trees and water

• the average building in taipei is clearly much older than in tokyo, eg. walls just look dirtier. at first i assumed this was also due to tokyo getting firebombed, but that doesn’t make sense as taipei’s expansion was mostly post-ww2

  • i think the real reason for this is tax codes! specifically japan has some tax incentives where you can let your property value depreciate heavily and count this as an income deduction. land remains valuable while properties don’t, which incentivizes landowners to demolish and reconstruct buildings every few decades. we also need this in sf

  • taipei does not do this, which also contributes to a few other things (more horizontal sprawl instead of vertical growth, motorbikes and scooters everywhere, etc). of course there are lots of other reasons for differences in sprawl as well, eg. being designed before vs after automobiles

finding your way

• high population density means a lot of information heuristics break down. in sf if you’re trying to decide where to eat you can reasonably use the heuristic i will look at every restaurant in the next few blocks and then pick the best one. if you use this heuristic in a dense asian city you will never get around to eating anything even if you limit exploration to a small neighborhood

  • i think this is how consumption will feel when agi really hits. we’re already starting to see it with code - software used to be scarce enough that when you were looking for a solution to a problem (eg. “download youtube video and extract audio”, “note-taking + journaling system”) you could feasibly go through all the available tools and use the best one, whereas nowadays the barrier to writing new tools is decreasing so rapidly that soon it will not be feasible to exhaustively identify the “best” software for any task (and maybe you just create your own if you feel like it). now imagine the same abundance in every other domain (entertainment, fashion, etc). we will need a lot of new heuristics to make sense of the world and i don’t think anyone knows what they are yet (the default answer is probably “feeds filtered by your personal agent” but we can do much better?)

• another fun case of heuristics breaking down was teamlab borderless (which is less well-known than teamlab planets but more interactive and coherent). it’s an art museum where all the art is constantly evolving and travelling between rooms, so have i seen this before is no longer a reliable indicator of whether you’ve visited a room and have i been to this room already is no longer a reliable indicator of whether you’ve experienced the art yet. as a result you’re incentivized to just walk back and forth a lot in what is essentially a poorly lit maze, which i found very fun

fashion

• people in both cities are so fashionable that i started experiencing dysmorphia for the first time in my life

• why is the fashion so much better than in the usa? a few hypotheses

  • good clothes are cheaper (even accounting for purchasing power differences)

  • higher social cost to looking sloppy, eg. it’s perceived as rude

  • my friend claire has a theory that east asia has a more superficial understanding of beauty. for example we went to many art exhibits featuring flowers, and they were not particularly thoughtful but they were very popular regardless

other

• i realized (or rediscovered, more accurately) i’m a weird traveller - for instance, my 5 favorite things from taipei were the mosque, the botanical garden, a graffiti alley, renting a good piano for $4/hr, and watching some street performers. none of these are things any of my friends or language models recommended; i found them by staring at maps and walking a lot

• on the topic of street performers, i really enjoyed watching this kid

  

  • his performance was great but talking to the kid afterwards made me pretty sad. i think one day he’s going to mess up and crack his head open. even if he executes this fall perfectly, the pole is not attached to the ground or anything; it’s being held in place by 4 random passerby with ropes and the whole thing wobbles when any of them moves. each show involves several of these falls + other stunts and optimistically 50 people tip him 100 NTD each time so he’s making 5000 NTD = 150 USD and splitting that with a show partner? there are some dubious economics here. like come on i love your work and it’s breathtaking to see live with music but i hope you’re taking care of yourself :(

• visited the your name staircase again. i already went in 2023 so this time i was not really there for the staircase and was instead there mostly to see if anyone else would be present. kind of incredible that shinkai made an ordinary location so meaningful that ten years later people are still going out of their way to see it in the middle of a snowstorm. i think that should be one of my life goals

THEREFORE EVERY CLOSED SYSTEM MUST CONVERGE TOWARDS DEATH

i. death

there is a kind of lifestyle where you spend your days shuttling between home, the office, and restaurants: your job contains some interesting and unpredictable challenges but largely consists of executing a well-understood playbook (like scaling up infrastructure, or engineering a system whose design was settled months ago, or go-to-market customer-feedback-iteration cycles); your friendships consist primarily of going out to eat and catching up; you recharge at home by doing nothing in particular aside from sleeping and doomscrolling and some hobbies that you find comforting; to get between your home and your office and your hangouts you take public transit, or uber, and you avoid talking to the other transit passengers and to your uber driver, or better yet you live in a city with Waymo and can avoid interacting with anyone at all

to me this lifestyle feels a little bit like death. a kind of death that is not immediately obvious, because it is possible to sustain the illusion of forward progress for months or years; you can be doing a great job at work and spending time with amazing friends whom you feel deeply understood by and so on. there is no fundamental contradiction between the following two statements:

1. the vibes in your life are great

2. you are slowly becoming incapable of receiving genuinely new information

of course you are still always encountering new content, but very little of the kind that is capable of changing your mind on anything substantial: you scroll through feeds that have already learned your preferences; you do not push hard enough on your hobbies for anything new to emerge; the ways in which you interact with your friends have crystallized and you almost never talk to outsiders. perhaps you are continuing to grow at work, but unless you’ve found good alignment most of the lessons you learn from work probably have little bearing on the rest of your life. you are secretly transforming into a closed system

this is the same death that dfw addressed in his commencement speech this is water (which you should read if you haven’t! it’s one of my favorites): “how to keep from going through your comfortable, prosperous, respectable adult life dead, unconscious, a slave to your head and to your natural default setting of being uniquely, completely, imperially alone day in and day out”. dfw made the observation that “You graduating seniors do not yet have any clue what “day in day out” really means. There happen to be whole, large parts of adult American life that nobody talks about in commencement speeches. One such part involves boredom, routine, and petty frustration” and indeed, this is my first time reading the speech where i’ve been old enough to really comprehend that

where i disagree with dfw is that he frames the problem as being about empathy and awareness - one can avoid this kind of death if they have more humility and are more willing to consider the possibility that their perspective is incorrect. i believe this solution is insufficient and that the problem is fundamentally also about lifestyle construction

ii. ideas

i spent much of early january thinking about how to adjust my life to increase my exposure to new ideas. this was not supposed to be one of my new years resolutions; however, i did set a goal of writing more fiction, and when i actually tried to start i realized i did not have many compelling leads. similarly, i wanted to build writing tools and interp assistants, but when i actually confronted the problem of what should the goals of ai tools and assistants be? i ran into a blank wall

it quickly became apparent that i was not encountering new ideas at a rate fast enough to sustain the kind of work i wanted to do; i needed to consume more and consume better. younger versions of me (as recently as 2y ago?) believed that engaging with the world too much would bias you towards thinking like everyone else and creativity was about thinking harder until you came up with something brilliant, but i no longer believe in this perspective. there is a very special kind of person who can pull that off successfully (eg. ramanujan comes to mind), but most people thinking in isolation become trapped and ideate in circles. looking back it’s clear i was just coping with being afraid of new things

some of the changes i’ve made in the last few weeks include: reading a new sci-fi short story every day (i’m not attached to sci-fi as a genre but it’s good that the variance is high), browsing the hci section of arxiv instead of the machine learning section (which unfortunately is not really innovating anymore), meeting lots of grad students, diversifying my instagram and pinterest feeds. tickling my brain in new ways has been very refreshing! i’ve also had a longstanding goal of being less judgemental towards my own ideas and it’s hard to measure but i think i’m filtering less now?

i don’t know how this actually translates into output or results. what i do know is that there was a time where i despised my brain and was terribly sick of the thoughts i was having, and i no longer feel that way

iii. life

one difficulty with trying to rearchitect your life is that ideas come from unpredictable sources. for instance, a few weeks ago i was leaving the office to go to a dance class and had to decide whether to uber or take the bus. i thought uber would be more efficient and give me more time to read sci-fi or think about new things or whatever. i ended up taking the bus anyway, and it was extremely packed so i was standing inches apart from a bunch of strangers, and then i started daydreaming about hugging strangers and various other topics. it was fun, and i’m very confident that my brain entered a more vibrant state than it would have if i’d chosen the uber option which was supposedly more aligned with my goals

that is to say, i don’t really believe in a central planning / optimization approach to being more generative. i hope that’s not the message people took away from the previous section. maybe a more nuanced framing is that if you notice a part of your life which is consistently dead then you should think about how to revive it. sometimes that entails tearing everything down and sometimes that entails making small adjustments

two related thoughts:

• one reason i find love difficult is that almost everyone i meet (with the exception of 1-2 new people a year) reminds me of death, and as a result i can’t be attracted to them. this is definitely a skill issue that i need to resolve by finding better ways to relate to people

• i keep thinking i should move to new york or something. i don’t actually think i’d enjoy it more than my current life, but i do think spending your entire 20s immersed in one culture is probably not healthy

lately when stuck i’ve been telling myself that there are no skill issues, there are only love issues. as in, do i love the ideas i’m thinking about enough to give them the attention they require? to be sure they’re worth pursuing? to overcome any obstacle in the way of bringing them to fruition?

if the answer is yes, then i trust the love will carry me through whatever i’m currently struggling with. for instance, right now i am writing a story that includes a kissing scene. writing about physical affection terrifies me so i have never done it. but this scene is crucial to the central metaphor of the story, which i am deeply excited for, so i have no doubt that i will figure it out

and if the answer is no, then i probably need to re-evaluate why i am working on the project. sometimes the motivation behind the project is genuinely misguided and it’s a good idea to stop. for instance, my most recent paper began as an attempt to beat an interp technique known as SAEs, and upon further investigation this goal turned out to be ill-defined because there was no meaningful way to determine which technique was “better”. meanwhile for creative work it’s more likely the project has been sidetracked by some kind of external pressure, and reconnecting with my feelings allows me to pivot the project in a more meaningful direction, eg. what happened to this substack last year

(of course, there are also truly intractable research problems where this framing does not help. loving mathematics very deeply does not enable one to solve the riemann hypothesis. but i think the vast majority of people are not working on this kind of problem, even if they think they are)

sometimes love is present but difficult to connect with. the most common example in my life is barriers in activation energy. it was difficult to work at my job for 8-10 hours and then come home and be excited to build a ddr pad or write anything serious. i interpret this as fatigue clouding one’s judgement, rather than a lack of love. i’m not sure how other people manage fatigue, but my main strategies are:

• acquiring energy. for me this often means talking to housemates, playing piano, or dance

• finding a low-energy entrypoint. for instance, i will go to the basement and touch my ddr tiles without building anything or i will read through my draft without making edits. usually by the time i’m done touching my tiles or reading my draft i’ve reconnected with the love enough to want to work

• sometimes i just need to quit and sleep early so i have more energy the next day

i. annual recap

i made end-of-year websites in 2023 and 2024 but this year i decided i’d rather start on upcoming projects so you’re getting this written summary instead. here’s a brief work-only recap of 2025:

1. i spent jan-march stuck in emotional paralysis, meeting lots of cool faculty and being indecisive about whether to go to grad school. also i was doing research part-time (including pre-deployment testing of o3 that went viral) and writing mediocre blog posts

2. in april-july i realized i was tired of who i was and that i’d trapped myself in a specific genre of writing that was far too limiting. so i tried a lot of new things, like drawing a comic about declining grad school (which i realized would be fun socially but would probably result in stagnation / feeling trapped, mostly due to the current state of nlp), building a ddr pad, writing a short story that many people told me they still think about half a year later, wild. relatedly i met some new art friends like J and T, who were very good influences and helped me realize that i could / should branch out into new mediums

3. in aug-oct i did self-directed research for the first time (ie. in all my previous research a lot of the good ideas came from other people, whereas this time they almost all came from me), traveled a bit, and tried some more writing experiments. also there was a stroke which made me re-evaluate my priorities

4. ai took over my life in nov-dec as i went through 2 research releases and also NeurIPS occurred. the deadlines destroyed my soul in that i had so many ideas i wanted to explore but they all got backlogged to focus on finishing projects and my personality became extremely one-dimensional, sorry if you had to talk to me during this time

i think 2. and 3. were important phases to go through. trying new kinds of projects and seeing them turn out well was helpful both for confidence and for becoming more accepting of new ideas (i’d observed in may that blogging too much biases you towards only thinking about ideas that fit in blog posts); i was able to connect with a lot of new people who liked stuff i worked on this year but would not have found my pre-2025 work interesting; there were also just a lot of cool moments, eg. one time i was talking to my friend alicia and she said something like “i love how much you’ve been experimenting recently, it’s inspiring” and i wanted to scream: no?? literally YOU are MY inspiration i love all of YOUR experiments what are you talking about?!?

ii. ai research

NeurIPS did not go the way i expected it to. twice a day i would walk through massive poster sessions, each exhibiting hundreds of machine learning papers, and i would be bored out of my mind; then i would walk through the 15-20 creative ai demos and consistently come away with new human-ai-interaction ideas to think about. after this occurred for the fourth consecutive session i thought hm, maybe this is a sign that i shouldn’t be doing machine learning

my current research is on language model interp. to clarify a common misconception i do not work on mechanistic interp (which tries to reverse-engineer the exact mechanisms used by models, and which i believe is impractical); instead i mostly use black-box methods (eg. training an assistant to do the interpreting). many people who work on interp are motivated by concerns about existential risk, so people often ask me why i (as someone generally unconcerned with existential risk) am doing interp. for the longest time i haven’t been able to give an answer beyond i think it’s cool to be able to understand language models better, and when people ask why i say something like i enjoy discovering new things about complex systems

which is definitely a bullshit answer, because i wouldn’t want to spend my life discovering new things about eg. zebra colonies. the discovered knowledge must have some use case that excites me, and if the use case for interp isn’t monitoring for safety / risk then what is it? i’m not entirely sure, but i think the interp applications i find most interesting might be spiritually closer to hci - how would we change the way we interact with language models if we could probe and steer their internal thoughts / hidden associations / conceptions of themselves while chatting with them? i will try to push my team in this direction, and if it doesn’t work out i think i should switch fields because i am quite tired of being far from applications

in november my housemates asked me to list examples of other peoples’ projects that i would’ve been happy working on, and i struggled to do so. probably most of my favorite books / art / media? the pinterest platform? the general user models paper? one of the common themes here is that i want to create things that help people better understand or express themselves

i think there are not many companies i would thrive at, eg. places like anthropic and cursor build great products that (in my view) do enable self-understanding and self-expression, and i could’ve joined either org but i find it hard to care when many other orgs are building the same thing. unfortunately everything with a large market incentive will have many copycats, so maybe i’ll never be wildly rich… i am still coming to terms with the fact that the things people are most eager to give me money for are not really the things i want to do

iii. reinvention

there’s a george bernard shaw quote i like that goes: “This is the true joy in life, being used for a purpose recognized by yourself as a mighty one. Being a force of nature instead of a feverish, selfish little clod of ailments and grievances, complaining that the world will not devote itself to making you happy.”

it’s a little harsh, but i think a feverish, selfish little clod of ailments and grievances, complaining that the world will not devote itself to making you happy accurately describes a lot of internet content, including much of my writing when i was depressed. i don’t mean to suggest this genre of writing is bad - it’s often relatable, beautiful, and genuinely valuable - but producing it feels like being trapped in a cycle of “have feelings -> write -> catharsis -> have more feelings” where you’re incentivized to keep having similar feelings so that you continue having easy material to write about, and i’ve been trying to instead move towards being used for a purpose recognized by yourself as a mighty one. (this is related to the observation that enneagram 4s grow by becoming more principled)

when i think about artistic reinvention one figure that often comes to mind is billie eilish. partially because she’s a year younger than me and watching her evolve in real time has been inspiring, partially because the journey she’s been on feels similar to what i’m looking to do. in particular: when we all fall asleep perfected a specific formula (bedroom pop, breathy vocals recorded close to the mic, minimalist bass-heavy production, mostly writing about teens being sad in various forms), but the commitment to the formula was often done out of fear (fear of belting and imperfection, fear of production that was too noticeable, fear of discussing topics where her opinion could be challenged; here’s a breakdown if you’re curious); happier than ever exhibited more diverse writing and vocal range (especially on the title track) and elements of rock / jazz; hit me hard and soft is still very obviously billie but large portions of it sound nothing like her first album with how much rock / jazz / edm / retro / other genres have been integrated

how do you go about completely altering your style, content, and relationship with your work, while keeping intact the emotional core of your output? it’s time for me to make the analogous transition from bedroom pop to rock. this year was a decent year for trying new things, but i didn’t explore nearly enough; much of my time was consumed by various self-created distractions and realistically i should be able to do 2-3x more in 2026. just gotta iterate faster, love harder, put more ideas out, fail earlier, make deeper contact with reality, train the idea muscles, compound the exponentials

this is the first time in at least four years where i’ve had resolutions entering the new year. in past years i was too confused to know what to do, and additionally i didn’t realize the antidote to confusion is rapid exploration rather than thinking hard and waiting for the confusion to resolve. i guess subconsciously i was also expecting the right people to appear and tell me what to do and save my life. those people actually did show up in 2023 and 2024 and 2025; and they were remarkably kind and insightful and helpful; and i’ll always be grateful to them for shaping me into who i am today; and they did tell me what to do with my life; but this year i realized everything they told me ended up being incorrect anyway, so i guess now i finally understand: you can’t wait for someone else to change your life, even if you’re surrounded by the best people ever. it doesn’t matter how good your taste in mentors is because even the most incredible person does not have enough information to know what your future should look like. your life is beyond the reach of prophetic instruction

i finally read the alchemist last week. most of my friends who have read the book say it’s overrated but i kind of disagree? i think it’s really interesting that the popular conception of alchemy is about doing bullshit pseudoscience to turn ordinary metals into gold and then the alchemist instead argues that alchemy is all about self-transformation and once you’ve made enough progress on yourself life will feel as if ordinary metals were gold. that to me is a very beautiful perspective. anyway here are a few relevant passages i liked:

“To realize one’s Personal Legend is a person’s only real obligation.”

“You must understand that love never keeps a man from pursuing his Personal Legend. If he abandons that pursuit, it’s because it wasn’t true love.”

“That’s what alchemists do. They show us that, when we strive to become better than we are, everything around us becomes better, too.”

iv. a few notes

on the theme of things that help people better understand or express themselves, here are some initial areas i am exploring (reach out if you want to chat or collaborate or do coworking / accountability!), and we’ll see where things go from there:

writing tools. language models have been decent at everyday writing for a while, but i think they’re finally good enough to be useful for the specific kinds of writing i care about, which look more like vibe curation. for instance, it should be possible to sample and remix text as easily as musicians sample and remix audio. i’ve been trying this out with poetry over the holidays, here’s a prototype that should be done soon:

optimistic stories about technology. many of my friends work at agi labs but none of them have actually been able to outline a scenario where agi occurs and the world is in a good state. there are absurdly few compelling visions of techno-optimist near-futures! i can think of solarpunk from the chobani commercial, the social-first app from better living through algorithms, the serendipity maximizer from all the birds in the sky, and that’s basically it. most techno-optimist narratives (eg. “solve science”) skip the near future and degenerate into high sci-fi / fantasy, or invoke concepts like “ubi” that don’t actually make any sense when you examine them critically. there is a clear vacuum here, and we should probably change that since narratives about technology are often self-fulfilling. right now we have no idea what to build towards other than that it should be intelligent

i don’t know what stories to tell yet, but as a starting point i really like this tweet from humans&, which is such a refreshing break from modern agi lab rhetoric about scaling / reasoning:

we think humanity’s biggest challenges won’t be solved by AI thinking for 1000 hours and coming back with an answer.
they’ll be solved by many collaborating humans, and AI that understands them and their different skills, goals, values, etc to empower them to do more together

feelings interventions. emotions are complex but when i look back on the last few years i find that something like 90% of how i feel is determined by two axes: hope / despair, conviction / uncertainty. a digital assistant or tamagotchi that just tracks these two axes and occasionally asks simple probing questions like: what would you do right now if you were less pessimistic? are you procrastinating out of despair or uncertainty? what are you currently confused about and how can you explore it? etc. would go a long way. also i wrote about friendship batteries two years ago and never figured out how to build them, i should do that

optimistic stories about people. this is pretty self-explanatory? uplifting narratives about other people. this can be nonfiction, like the work my friends donald and aadil are doing to capture our peers’ stories, or fiction, where i have some ongoing drafts

normal stuff. after writing the above list i reread it and thought why is it that i can only come up with useless ideas? i find it very suspicious that there is no traditional tech here (eg. “cursor for X”, “gadget for Y”); for some reason those ideas have been evading me. it might be an attention issue where if an idea is a) not specific to a niche interest i’ve thought a lot about b) obviously useful and feasible to build, then it’s likely to already exist? but also this could be an imagination issue. i can’t tell if lack of traditional tech is a feature or a bug, let me know if you have any thoughts

unbelievably excited for 2026! thanks to AA AG AG AY CW DJ DY EZ GZ JS JS JX SZ TW for relevant discussions

all the ways in which you are not free, all your fears, all your anxieties, all your guilts, all your shames - these are also part of you. they cannot be separated from you. in seeking to free yourself, you will inevitably change yourself. your way of thinking, your likes and dislikes, your hopes and dreams, your way of being “you”, your “raison d’etre” - all of these things may change even without you intending them to. holding onto them too hard will keep you from being free.

be wary of this thought: “i want to become more free so that i can better achieve my concrete goals.” goals are often just the chains we have bound ourselves with as we wonder why it’s become so hard to move.

you may find new things to love altogether. are you prepared for this?

do you value freedom, joy, and love more than anything in your entire life?

yes? good.

then we can begin.

a couple months ago i realized i was trapped in a Box. the Box was a set of expectations i’d imposed on myself without ever realizing it, and none of them made any sense! some examples:

• i thought piano was about playing the notes on the page correctly, so i would download sheet music for songs i liked that were too hard for me and then struggle very slowly to read through and play all the notes. it was not fun and i sounded awful. this continued until one day i remembered a) i will never be classically trained, nor do i want to b) why do i need to play all these notes someone else wrote when i’ve spent 2 years writing a cappella arrangements and learning music theory? c) my goal is to play music i like, not to play correctly! i should just read the score and play the version of it that best suits me

• for most of my life i thought running was about suffering. i would move my legs until i reached the familiar pace where the weight of my legs started to tire me out, and then i would think “well, looks like i’ve reached my running state” and maintain that pace moving forward. i equated running with that feeling of exertion and literally did not realize i could run faster until one day my friend pia bought me some very lightweight korean pants and the exertion diminished and i realized running is just moving your legs and you can tell your legs to move faster even if they feel heavy

• i also equated being productive in my research with sitting in front of my monitor and writing code, so i would spend a lot of time sitting in front of my monitor and trying to write code even when there wasn’t actually any high-priority code to write. it took me a while to realize i was wasting hours each week trying to roleplay productivity, and also that monitors (for me) are more of a cage than a productivity boost. nowadays i spend a lot less time trying to write code and a lot more time walking around thinking about research prioritization, and i think it’s resulted in faster progress

more generally i think the Box looks something like: you have some big-picture goal for how you want to be (eg. making good music) but the big-picture goal remains in the background while some proxy goal (eg. reading and playing sheet music) occupies your immediate attention. the proxy and big-picture goals are correlated but misaligned - taking steps towards the proxy gets you closer to the big-picture goal (especially when you’re just getting started), but optimizing too hard for the proxy can take you further from where you actually want to be. failure to realize your goals are proxies can lead to burnout as you think i’m trying so hard on things that i said were good for me, why is it not making me happier?

(some other proxies i’ve been fooled by over the past year include “meeting new people at parties” “getting a phd” “being in a relationship” “writing Important Blog Posts”)

i think the Box is especially problematic for me because i often feel the need to prove to myself that i’m working hard and spending my time well, and legible proxies are an easy way to convince myself of that. this also results in being Serious and suppressing the sillier / more child-like parts of myself. people who feel less of a need to prove themselves may not experience as much of the Box; i’m not sure

if you’re here, you probably want to change your internal model. you may have tried to do this many times by suppressing the feelings you don’t like, by changing through internal pressure, by pushing back against yourself.

you may have noticed that this doesn’t work.

there is another approach. pay more attention to everything in the world around you except yourself. when you go for a walk look at the trees. look at the buildings. look at the sky. find something to appreciate about all of it.

it’s no secret that one of my favorite people in the world, and also the individual who’s changed my life the most in the last five years, is my housemate marley. one thing she’s exceptionally good at is finding a way to enjoy situations that are objectively terrible, eg. going through three simultaneous (non-romantic) breakups while also thinking life is great

i did not understand this mentality until a recent two-week stretch when one of my Big Dreams was dashed and i also had to work nonstop for a paper deadline. i was heartbroken and stressed and exhausted all at once, and i didn’t have time to process any of the feelings, but i also found that part of me was deeply excited about it all? excited about heartbreak because it was great to have found something i really wanted, excited about the stress and exhaustion too because it meant things were happening. at some point i was complaining to marley and she replied this is the best part of life

a different perspective on this is that first-person-vincent can be having a terrible time, and third-person-vincent narrating the situation can decide to have a terrible time as well, but they can also decide to be happy or curious or excited about the plot. you can’t fight or suppress feelings, you can only decide whether or not you’ll appreciate them. i think people who fail to do this are simply not taking their lives seriously enough. the stroke episode in august was helpful for understanding that the alternative to all this isn’t feeling happy, it’s feeling nothing at all

our lives are filled with all kinds of beautiful things made by all kinds of beautiful people. many of us have been taught to hide those feelings. or to cover them in layers of irony. or to find criticism to share instead. to lift ourselves up by putting down others. whenever you have a chance, practice the option to share your love. earnestly.

do not try to get anything in return. tell people how much you appreciate the things they do. how much you appreciate them being who they are.

you may not always get the response you want. that’s okay. because you’re not doing it in order to get a good response. you are doing it to become a more free person.

sharing your love is not for them. it is for you.

a few weeks ago someone asked me if i find it unpleasant to hang out with people post-rejection, to which i have 3 responses:

• why is your pain tolerance so low?

• do you actually like people, or do you just like the idea of being together?

• (more seriously) the point of life is to spend time with people who make your life full! people i’m deeply attracted to consistently bring out the best possible version of me, so i will always want to keep spending time with them

i also realized that i’m something like 5x funnier around my housemates than around strangers. everything feels so much freer and more playful! which is actually a little annoying because i don’t need to be funny around my housemates (we already live together) while i would benefit from being funnier around new people (it would help with making friends). life would be very different if the best versions of me showed up when i most needed them instead of when i was most happy, but i think there’s not much i can do about that other than learn to be happier

earlier this year i wrote about how the generator-discriminator dynamic from machine learning also applies to humans - as in, some people are naturally generative (come up with lots of ideas, often low-quality, but can see the goodness lying dormant in ideas that look bad), others are naturally discriminative (can quickly point out flaws and assess which ideas are better than others), and you need both halves to consistently produce great work

i’m a very good discriminator and have been trying to become a better generator, mostly by hanging out with generative people and trying more new things. i think it’s working? and along the way i’ve realized that some of the reasons i became such a good discriminator are rooted in: fear of the unknown, fear of embracing something different, fear of finding new things i love so much that they uproot my entire life. part of me is excited about all the unknown unknowns i’ve yet to encounter, but another part of me feels threatened by them. of course all this fear is silly and i need all the great ideas i can get my hands on and i should love the world harder and be less critical!! i am trying my best

do you want to learn to play an instrument? it’s easy to imagine how nice it would be to play it well some day.

but you must also be able to enjoy today, a day when you cannot yet play it well.

if you spend your entire life doing nothing but chasing dreams you will never be happy.

sometimes i enter a state where i’m extremely self-critical and aware of my deficiencies. i feel like i’ve squandered my entire life and am totally unprepared for the future, and then i spiral for a bit, and then i try to do some planning to remedy things. unfortunately the plans are almost never helpful

as a small example - it used to be the case that if i really wanted a social interaction to go well i would plan out some questions to ask, anecdotes to share, and so on. after trying this for some months i realized that there was basically no correlation between how much i planned and how well the social interaction went. in fact i experimented with many things and the only thing that seemed to have any correlation with social interaction quality was how much i smiled. the part of me that wanted to control the conversation was not actually capable of getting the outcomes it wanted, so i’ve been trying to let go more

more generally i think the only real way to make progress on your problems is by doing things you love and losing yourself in action. it must be done through joy, not grief. i am stuck grieving in my head far too often and need to get out more! simple procedures that collect lots of feedback do much better than complicated procedures that collect little feedback (in machine learning at least…). one time i was fixing a sewing machine with my mech-e friend brian and i watched him spool threads and disassemble components, making decisions faster than i could process what was happening and getting more done every ten seconds than i would in a minute (and it was also the most attractive thing i’ve ever seen a guy do), and i thought holy shit, this is what it means to truly iterate and make progress

• better living through algorithms, the first depiction of agi i’ve encountered that feels both plausible and optimistic. it was such a breath of fresh air from all the superintelligence / transhumanist / existential risk / disempowerment / automation narratives of the current moment, i read it and then just sobbed for thirty minutes straight

• all the wrong moves, a chess memoir by sasha chapin (better known for his substack). very short and pleasant read about addiction and confronting the fact that you’ll never be great at an activity you love. some allegories about chess and life

  • “Badly played chess is kind of like badly played life. Real problems are dealt with poorly or not at all, while much effort is expended on avoiding imaginary danger. Rather than dealing with the reality of the situation, you act as if you were playing the game you wish you were”

  • “In life, and in chess, people make terrible decisions just because they’re impatient. They want things to end, right now, on their terms. They just want a reckoning, whether or not it’s actually good. So they play f4, or they play bishop takes h7, and they just tear everything apart. How beautiful it would be, I realized, to be playing a game without ever wanting it to be over”

• east of eden, an american retelling of the biblical cain/abel story. when i first read the bible as a kid i thought “these characters seem pretty unmotivated, feels like whoever wrote the book of genesis just needed an origin story for murder and made one up arbitrarily”. steinbeck does a good job of crafting deep characters to answer questions like: what does rejection feel like? what are its effects? what would actually cause cain to murder abel in real life? where does sin come from and how much control do we have over it? also steinbeck is good at writing female antagonists and asian people. very pleasantly surprised

• learning to work, a short text about someone going from only being capable of working for 5-15 minute intervals at a time to being satisfied with their work output. from 1977 surprisingly, probably a lot more relevant to people today. many reflections on anxiety / delusion / attitude towards work

• twilight | contrapoints - a 3-hour video essay nominally about twilight but mostly about female desire / how depictions of female desire have changed over time / where limerence and submissiveness come from. really interesting, and i don’t even know anything about the twilight series

• the hardest problem evolution ever solved - explaining various features that we had to evolve to transition from living in the ocean to living on land

• jasmine sun had an incredible content streak. reflections on chinese attitudes towards tech and competition, what the ai art debate gets wrong (ugh it’s paywalled now), how democrats lost the support of tech

• read moby dick and did not like it unfortunately, far too many whaling expositions for my taste

things watched

• look back, one of my new favorite films, less than 1 hour long, about 2 kids who grow up drawing manga together. really beautiful depiction of friendship, and made me deeply envious of people with longtime creative collaborators. i only had competitors growing up and would’ve done anything to have that not be the case :’)

• the room, also one of my new favorite films, probably made me laugh more than any other content i’ve consumed in the last few years. the script, acting, delivery are all absurd

• nausicaa of the valley of the wind felt like the anime version of dune in many ways (giant desert creatures that seem harmful but aren’t, tribes with drastically different levels of technology, etc) and clearly inspired attack on titan. lots of discussion of environmentalism and arms races. surprisingly good for being miyazaki’s second movie

• finally saw les mis live! stage limitations made me appreciate the visuals in the (2012) movie more, they actually did a great job with setting even if the singing was bad. i think the songs which benefited most from being onstage were drink with me, dog eats dog, empty chairs at empty tables

• loved fleabag but to be honest i cannot articulate what was good about it. sharp character writing? the lead actress also being the writer and producer? discussion of celibacy and repentance? all of the above but i feel like there was something else as well that i can’t articulate

• saw two polar opposite movies about kpop in the same weekend, kpop demon hunters (mid) and time to be strong (great). the former had good songs but did not utilize them properly and butchered the climax in my opinion. the latter is about three unsuccessful kpop stars confronting debt, trauma, and the dark side of the industry. first time seeing these topics explored in film

other

• for each of the last eight weeks i thought at the beginning of the week i was going to make significant progress on interpretability and then proceeded not to 🙁 in a big time crunch to get results out before thanksgiving, quite stressful

• went to my first wedding! very inspiring. first time in my life where i didn’t feel awkward while dancing, it turns out a year of practice helped or something. also realized toronto is nyc except canada

• visited boston and saw lots of friends. i really miss cambridge, especially the foot traffic. gave an interpretability talk, my first one in 2 years, turns out i miss giving talks too

• got some very lightweight pants from my korean friend which helped me discover that i can run way faster than i previously believed (more on this in a later post)

• (also more on this in a later post) realized i need to be happier. far too much of my happiness is tied to high-variance things outside my control, like other people and research results. i can’t always be waiting for good fortune to save me!! some new hobbies: piano (need more non-writing art in my life; it’s also way more enjoyable that i’m allowing myself to diverge from sheet music), e-biking (love getting slammed by 15mph winds), pinterest (what a great app! scrolling actually inspires me so much)

this post discusses two unrelated topics: my experience using chatbots for therapy, as well as why i have long ai timelines. (Q: why these specific issues? A: they happen to be top-of-mind subjects where i believe mainstream discussion is lacking, Q: why are these in the same post? A: mostly because i’m not sufficiently motivated to write two separate posts)

i. chatbot therapy

there have been many news articles on the risks of using chatbots for emotional regulation (dependency, sycophancy, psychosis). i think these are largely correct, but i also think it’s possible to have extremely healthy therapy-adjacent interactions with chatbots

to me the main difference between healthy and unhealthy interactions is whether the user is looking for understanding or validation. any interaction where the user is looking for validation has the potential to become dangerous via sycophancy and repetition. it is not sufficient to prompt the language model and tell it to avoid validating you because chatbot dependency is, for now at least, primarily a demand-side problem (users have holes in their lives that need filling) rather than a supply-side problem (chatbots “trying” to make users dependent). this makes the problem difficult to resolve with weak supply-side interventions like prompting

(to rephrase this without the econ framing: as long as the desire for validation exists within yourself, you will latch onto statements made by the language model that feel affirming to you, regardless of the model’s intent. because you have unlimited access to these models, you will eventually encounter these statements if you chat for long enough, and once you encounter some you may decide to look for more of them)

next i will briefly document one of my conversations with chatgpt and claude which is representative of how i typically use these services. the context here is that last week i was talking to a friend about self-worth; after a few minutes she asked me “what even is self-worth anyway?” and i realized i couldn’t actually provide a coherent definition (and neither could anyone i asked)

so i start by asking each model to explain self-worth:

they’re both helpful, although in the early stages of exploration i find chatgpt tends to be more helpful than claude because it naturally suggests lots of follow-ups:

after some rounds of chatting i feel like i have a good understanding of self-worth, self-esteem, how they manifest, etc. i make note of which pieces of the chatbot explanations do / don’t resonate with my personal experiences, and then ask the chatbot to clarify:

i asked a similar question to chatgpt, but at this stage i find that it tends to respond with prescriptive cliches which i dislike, so i stop talking to it:

my experience is that neither chatbot is very good at walking you through your own feelings / asking the right questions / doing so in any thorough or systematic manner. it’s simply missing too much context on your inner state to do this effectively. you need to be self-aware enough to explore your own feelings and know when you’ve done enough, and then compile and present all the evidence to the model

once the crux has been identified, the chatbots are pretty good at brainstorming next steps:

one thing the models are quite bad at is pulling in content from domains unrelated to the current discussion. for example, in this case, i had some tangential experiences that i thought were relevant, but the chatbots have no way of figuring that out on their own, so i have to make the connection myself and ask about it:

again, claude’s responses tend to be pretty helpful, but the crux needs to be supplied to the model first

i thought this interaction went well, though importantly: 1) at no point was i looking for emotional relief; i’d already processed all the feelings and was only interested in understanding where they came from 2) i had already collected lots of relevant context to supply to the models before starting

don’t forget to thank your chatbots!

ii. long timelines

many people assume that i expect “agi” soon because i do language model research at a safety-motivated org. unfortunately i actually have fairly long ai timelines, probably longer than almost all of my friends in sf, and i’ll try to clarify why

ai tools often feel incredible to use. you can crank out way more lines of code per day than you used to while also scrolling twitter as you wait for your agents to finish running. unless you track your time carefully it’s easy to feel like ai is helping you do everything way faster when in reality it’s probably greatly accelerating some subtasks while barely helping on other subtasks. empirical studies show that people drastically overestimate how much time ai tools are saving them

a simple statement about ai capabilities that i think most people agree with is: for any task of interest, ai systems can become arbitrarily good at that task when trained with sufficient amounts of high-quality training data. this has proven to be the case in a wide variety of domains - driving, video generation, writing code, clicking and scrolling in documents, strategy games - and there is no obvious reason to expect otherwise

so we generally agree that ai will be useful on tasks we’ve trained it to be good at. however, collecting training data for tasks can be very labor-intensive. when people talk about agi, they usually have some notion of generalization in mind, where the ai is capable of doing well on tasks it was not explicitly trained for

i think it’s worth noting that there are few (i can only think of obscure in-context learning examples?) instances where ai was initially bad at a task we cared about and improved steadily without being trained on lots of data very relevant to the task. furthermore, many leading figures at ai labs believe this will continue to be the case - it’s why so many respected researchers joined thinking machines to help companies unlock enterprise-specific applications, why openai’s top new grads left to focus on rl for businesses, etc. these startups, whose plan seems to be to work with companies on a case-by-case basis to create ai systems specialized for individual use cases, would not exist if there were signs of a generalization boom in the near future

when i say i have long timelines maybe what i really mean is that i expect ai capabilities to continue advancing one task at a time as we identify valuable tasks and then train on them. progress may accelerate as we become more efficient at collecting task-specific data but it will still occur in a deliberate manner. generating synthetic data can sometimes provide a shortcut for data collection (eg. reasoning rollouts on math problems) but appears to still be quite resource-intensive

to be fair there are a few legitimate arguments for generalization

• one argument goes along the lines of “we only need to train ai to be very good at software engineering, once ai is good at that it can write software / use digital tools to complete a wide variety of other tasks” (and there’s a more narrow version of this argument where “software engineering” is replaced with “ai research”). i am still undecided on the extent to which i believe this (it’s partially a question of how much breadth you believe software and tools unlock), but currently am skeptical because it seems to suggest software engineers should’ve already taken over the world

• another argument goes along the lines of “capabilities might advance one task at a time, but because of moore’s law / increasing compute supply + synthetic data we’ll get exponentially faster at gathering data and training on new tasks over time, so at some point in the near future ai will feel generally capable”. to me this seems like a question of what you believe the future of data collection looks like. i’m currently of the view that as we exhaust tasks for which data is easy to collect, the effort required to perform data collection / evaluate data quality will also increase exponentially, similarly to how scale.ai used to be able to sell data very cheaply via filipino data labelers but nowadays labs spend hundreds of dollars collecting individual datapoints because we’ve exhausted gains from low-quality data

anyway all of this is merely to suggest that the “g” in agi is difficult to attain, rather than that ai will not be tremendously useful

some final thoughts:

• i’m very sympathetic to the arguments in ai as normal technology and why transformative ai is really, really hard to achieve, both of which argue that very smart ai will not feel transformative due to structural bottlenecks and diffusion costs

• i am fairly concerned that the us economy is becoming a bet on ai (see: gdp growth may be primarily driven by ai capex, lack of job creation, extreme concentration of s&p in ai companies) which may not pay off. i have moved half of my savings to crypto, foreign stocks, commodities as a result

(While we’re on the subject, this is a sick Celeste tattoo. Almost makes me want to do a road trip to LA. Also, sorry for spoilers. But this post has spoilers for so many things that I’ve given up on including warnings.)

This might be my first properly-capitalized non-academic writing in like ten years or something. Mostly because the vast majority of the text is copied from other places, but also for legibility reasons as this one is especially hard to read. Don’t worry (or sorry), we will return to lowercase shortly.

I wake up & it breaks my heart. My tea’s gone cold, I’m wondering why I got out of bed at all. I draw the blinds & the thrill of rain breaks my heart. The morning rain clouds up my window, and I go outside. personally idk how u can walk through downtown SF late at night and I ride the train, walk among the buildings, men in Monday suits. [MARLEY] There’s a sense in which everyone is pathetic. The flight of doves, the city of tents beneath the underpass, the huddled mass, old women hawking roses, & children all of them, break my heart. Pathetic on the surface, but also pathetic deep-down. see people asleep on the sidewalk and feel anything other than There’s a dream I have in which I love the world. a deep tenderness And you can choose to see everyone as pathetic if you want, or you can choose to not. I run from end to end like fingers through her hair. I can’t see at all, but even if I could it would all be gray [VINCENT] I think that’s one of my problems. There are no borders, only wind. Like you, I was born. I grew up not really respecting other people. Like you, I was raised in the institution of dreaming. and a wish for a better world. Hand on my heart. And when you’re in the habit of not respecting other people it’s also really easy to not respect yourself. Hand on my stupid heart.

If you want to be loved, find something you love. People can sense it when you have something you’re dedicated to. [CLEMENTINE] Too many guys think I’m a concept or I complete them or I’m going to make them alive, but I’m just a fucked-up girl who is looking for my own peace of mind. Don’t assign me yours. No one wants the burden of being the answer to your dissatisfaction. [JOEL] I remember that speech really well. [CLEMENTINE] I had you pegged, didn’t I? [JOEL] You had the whole human race pegged. [CLEMENTINE] Probably. When you’re unsure of yourself, it’s easy to be obsessed with the idea of love—the idea that happiness will arrive when someone else loves you. [JOEL] I still thought you were going to save my life. Even after that.

[VINCENT] I have this behavior where I obsess over projects and people and place a lot of happiness in the distant future. [BOJACK] Everything comes so easy for you. Like I tell myself I’ll be happy in a few months when this project is done, or I’ll fantasize about someone. The fourth rule was to ignore thoughts about the end product and how the end product would be received. I could too easily find myself inhabiting a fantasy world [MR PEANUTBUTTER] Oh, and it doesn’t for you? [CLAUDE] The obsessive quality you mention might be serving as emotional anchors or sources of dopamine that keep you engaged and hopeful, but potentially at the expense of feeling satisfied with life as it is right now. in which my thesis led to fame and renown. You’re a millionaire movie star Not only was this eventuality extremely unlikely, but it led me further away from, rather than closer to, This could relate to a few different things: a tendency toward fantasy as emotional regulation, with a girlfriend who loves you, difficulty with present-moment fulfillment or contentment, my goal of discovering the pleasures of the process of work. possibly using these obsessions to avoid uncomfortable feelings or emptiness. acting in your dream movie. I wanted to work not because of the supposed effect of my work on others, but for the gratifications, to me, of working. What more do you want? [VINCENT] Recently I did IFS and noticed that there was a part of me buried in my childhood which was harshly criticized by my parents / felt unloved / needed external validation. What else could the universe possibly owe you? I wonder how this relates to self-worth and my obsession over future outcomes. [CLAUDE] Here’s how it likely connects: That younger part who was harshly criticized and competed for validation learned that love and safety came through achieving and performing. [BOJACK] I want to feel good Even though your adult self has built healthy self-worth, that part is still running in the background. about myself. So your current obsession with future outcomes might be that part’s survival strategy still operating—except now it’s internalized. My fantasies made the reality of my barely begun thesis look so shabby Your adult self developed good boundaries and independence (probably as protection), but the underlying mechanism of “happiness = accomplishment” remained. The way you do. Romantic fantasies might be that younger part’s way of accessing the unconditional love and acceptance it never got. And I don’t know how. In fantasy, you don’t have to achieve anything—you’re just loved for existing. I didn’t want to have anything to do with it. But because it stays in fantasy rather than present reality, it maintains that “happiness is always just ahead” pattern. I don’t know if I can.

I want someone to tell me what to wear in the morning. No, I want someone to tell me what to wear every morning. [TIFF] I would put you as more serious than silly. I want someone to tell me what to eat. What to like. [VINCENT] I think I want to go ebiking regularly. It seems fun and will help me get out of my head more. What to hate. Lord, I confess I want the clarity of What to rage about. What to listen to. [MARLEY] You’re trying to make all these top-down changes to your life and I’m not sure if that actually works. What band to like. What to buy tickets for. [VINCENT] Yeah I know, it’s a problem, I’m trying to be less serious. What to joke about. What not to joke about. [TIFF] I have a problem where I’ve been doomscrolling a lot on TikTok. I want someone to tell me what to believe in. [VINCENT] Oh, I hope you do less of that soon? Who to vote for It feels like you’re trying to plan solutions because you want to have already solved the problems, catastrophe but not the catastrophe. and who to love and how to tell them. when instead you should be having fun in the process. The point of life isn’t to have lived it. I just think I want someone to tell me how to live my life, Father, because so far, I think I’ve been getting it wrong. [VINCENT] I see what you mean. It’s like how I would never go on TikTok because it seems like a bad use of time, [TIFF] It’s kind of nice though. The silly part of me thrives on TikTok. but actually some friends I really respect use it a lot and it seems quite valuable for them. And I know that’s why people want someone like you in their lives, because you just tell them how to do it. [MARLEY] Yeah, it seems like there’s a lot of exploration you don’t let yourself do You just tell them what to do and what they’ll get out at the end of it, even though I don’t believe your bullshit because you’ve told yourself it seems wrong. and I know that scientifically nothing that I do makes any difference in the end, anyway, Like everyone else, I want a storm I can dance in. I’m still scared. Why am I still scared? So just tell me what to do. Just fucking tell me what to do, Father. I want an excuse to change my life.

Reed College at that time offered perhaps the best calligraphy instruction in the country. Throughout the campus every poster, every label on every drawer, was beautifully hand calligraphed. Because I had dropped out [KEVIN] What are your goals right now? and didn’t have to take the normal classes, Slow down, you’re doin’ fine, I decided to take a calligraphy class to learn how to do this. [VINCENT] I think my goal right now is just to make a lot of stuff. I learned about serif and san serif typefaces, about varying the amount of space between different letter combinations, about what makes great typography great. [KEVIN] Does it matter what kind of stuff? you can’t be everything you wanna be before your time. It was beautiful, historical, artistically subtle in a way that science can’t capture, and I found it fascinating. [VINCENT] I don’t think it matters what the stuff is as long as it’s aligned with my values and I’m pushing myself. Although it’s so romantic on the borderline tonight, tonight. None of this had even a hope of any practical application in my life. Too bad, but it’s the life you lead, [KEVIN] Are you worried about being too unfocused? But ten years later, when we were designing the first Macintosh computer, it all came back to me. you’re so ahead of yourself, that you forgot what you need. And we designed it all into the Mac. It was the first computer with beautiful typography. If I had never dropped in on that single course in college, the Mac would have never had multiple typefaces or proportionally spaced fonts, and personal computers might not have the wonderful typography that they do. [VINCENT] I used to worry about this a lot. I thought I needed to have a grand vision for how everything fit together. Of course it was impossible to connect the dots looking forward when I was in college. Though you can see when you’re wrong, But it was very, very clear looking backwards ten years later. Again, you can’t connect the dots looking forward; you can only connect them looking backwards. I guess I no longer believe that’s a realistic goal though, so I’ve stopped caring about it. So you have to trust that the dots will somehow connect in your future. you know you can’t always see when you’re right, You have to trust in something — your gut, destiny, life, karma, whatever, you’re right.

(Words borrowed from: Meditations in an Emergency, Thank You / Stan, a ctbeiser tweet, many conversations with friends and Claude, how to avoid half-heartedness, an Eternal Sunshine of the Spotless Mind scene, a Bojack Horseman scene, Learning to Work, a Fleabag scene, CATASTROPHE IS NEXT TO GODLINESS, Steve Jobs’ 2005 commencement speech, Vienna.)

i rode the (46-hour) empire builder from portland to chicago! having taken the (52-hour) zephyr from sf to chicago last year i thought i knew what to expect, but this train experience was significantly better for a number of reasons:

• the route is more scenic. montana is probably the most gorgeous american state i’ve seen outside alaska (and i have visited 44 states), especially the parts near glacier national park which are just endless mountains / valleys / streams / forests / fog. the washington portion of the trip is pretty too as you follow the columbia river for a few hours, and same with the minnesota / wisconsin portion which follows the upper mississippi. (the zephyr route has some nice sections too, especially portions of utah and colorado along the colorado river, but large sections of it are quite boring, namely nevada / nebraska / iowa)

• another consequence of the route is that there are less delays. in fact we were so far ahead of schedule that i spent an hour walking around minot, north dakota. i think it’s largely a function of 1) not sharing as many tracks with freight trains (which have priority over amtrak) 2) the route is quite flat. the zephyr route has a lot of elevation gain due to crossing the rockies, which often leads to engine failures, eg. i had an eight-hour delay when taking the zephyr. but also, that delay resulted in them giving me a voucher which made my empire builder ride free, so maybe it’s worth it…

• (small sample size, and it’s possible i just present as more sociable now) the passengers seemed friendlier. i had very pleasant conversations with strangers about life, whereas last year people only talked to me to say 你好. also significantly more young people on the train after passing through university of minnesota

• better prep on my part - i knew there would be no signal and bad food so i downloaded moby dick and brought vitamin water and a lot of harvest pea snaps + vegan protein bars + other snacks (the common theme being high fiber, high protein, no sugar)

why take the train at all? apart from the nice scenery and interesting strangers, i appreciate the occasional act of stripping life down to its bare necessities and being reminded of what that feels like + seeing what emerges, eg. there was one morning where i woke up and noticed my phone was dead, and then i realized there was no point in charging it because there was literally nothing i would do with said phone anyway. it’s like a meditation retreat with sightseeing. and i haven’t experienced anything quite like the pure joy and gratitude that comes from walking around the chicago riverwalk after two days stuck on the train

here is my somewhat optimized amtrak meta:

• your options are to buy a sleeper car ticket (private room with beds) or a coach ticket (normal seat in an aisle, similar to buses and planes). do not get a room unless you’re old or bringing kids. it’s 3-5x the cost and you should not spend much time in the room anyway. instead you should be spending your time in the observation car (see top left photo), which has tables + very large windows on both sides and is the best place for scenery / work / socializing. literally just wake up early, find a seat in the observation car, and spend all day there

• the coach seats are split into upper vs lower levels. you should get an upper level seat. the lower level seats are less spacious and usually filled with old people who don’t want to walk up and down stairs

• how to sleep without a bed or private room?

  • the coach seats are pretty large, have reclining + footrest features, and few people are insane enough to plan on sleeping in them overnight (on both of my rides they were <20% capacity at night) so you will have a row to yourself, meaning you can lie down on 2 adjacent seats. the air conditioning and engine provide loud sources of white noise so you will not be bothered by sounds from other passengers

  • some people pack lots of pillows, blankets, etc. as someone who likes traveling on a single backpack, i don’t have space for these items. instead i wear warm pants and use a jacket as a blanket (wear it backwards and put your arms in the sleeves). i also prefer to drape a soft black shirt over my head instead of using eye masks, i find it’s more comfortable and versatile

  • you might have some concern about people taking your stuff while you’re asleep? most people seem to leave their bags unprotected and nothing bad happens but i tend to be paranoid about these things. i like to sleep with my backpack straps wrapped around one of my legs and it works quite well without interfering with my sleeping position

• never buy drinks on the train. largely because 1) they’re overpriced and sugary 2) many of them do not come in resealable containers, which i find extremely annoying because it means you have to finish them once you open them. you can buy beverages at vending machines in intermediate train stops. i also like to dehydrate myself so i don’t need to use the bathroom as much but i’m told this is an unpopular opinion

• pack all your stuff and wait downstairs 10 minutes before the last train stop to skip a long line

(O is my research advisor / manager)

O: did you ever figure out the cause of the stroke symptoms?
V: not really… my blood sugar and cholesterol are a bit high but not dangerous. it could’ve been caused by holes in my heart? or just general stress
O: oh, have you been stressed recently? let me know if there’s anything i can do to help
V: yeah i have but it’s unrelated to research, so not really stuff you can help with
O: you know i care about you outside of research, right?
O: like, we met ten years ago and have only been working together for the last year
V: oh
V: a little stressed about love
V: but mostly stressed because i feel like i’m running out of time to do good work
V: which i know is silly. but like, my housemates are my age and working on all these really cool companies. and i wish i had evidence that i was capable of doing something like that, because so far i haven’t worked on anything great in my life?
O: didn’t you get multiple gold medals at the IMO? is that not great?
V: i don’t think so
V: i started learning math really early on, but i don’t think my results conditioned on starting early were that good
V: it feels like i often underperformed when it mattered most? like, in 2019 i had four hours to solve a single problem and didn’t. i’ve always thought someone in that position should’ve been able to do better
O: this is a very interesting standard to have
V: i guess gold medals are also just not something i value? they’re not actually useful
O: what about the work you did for the Transluce launch? i thought that was pretty great and useful
V: i think the launch went well but would’ve gone fine without me? worst case it would’ve been delayed a few weeks
V: maybe it’s because i wasn’t driving anything, i mostly had a supporting role in D’s project and K’s project. every great project i’ve been a part of has been driven by someone else
O: well, you’re driving your current project. what would it take for you to feel happy with that in, say, a couple months?
V: i’m not sure… i think i’ll know when i see it?
O: i’m worried that if you don’t define what success looks like now then in a few months you’ll look back and find something to be unhappy with even if things went well
V: that’s fair. but i think ultimately it is just about whether i feel happy with the work or not?
V: like, i used to care about how many likes i got on my blog posts, or how viral they went. and then i wrote a bunch of posts that never went viral but it turned out i was happy with them anyway because i decided they were good. so nowadays i feel like that’s basically all there is
O: sure, i still think it would be helpful for you to have intermediate goals you’re happy with
O: i believe in your ability to do good work, and i think we can work together to get you to a state you’re happy with. but it seems like you have a really sparse reward signal right now, which makes it difficult to optimize against

(T is a friend)

T: have you been meeting new people recently?
V: kind of… but i feel like i have a problem where the type of person i attract is disjoint from the type of person i’m looking for
T: ooh, what are the two types?
V: okay this is going to be a really inaccurate oversimplification
V: i’m looking for people who are very generative and know what they want
V: but i don’t have those qualities myself, and i think people with those qualities are usually interested in other people who also have those qualities
V: and i think i attract people who score even worse on these axes than me
V: for example, sometimes i write about being confused and that’s validating to read for people who are more confused than me, so i often attract confused readers
T: i see
T: i don’t really understand why you say you’re not generative or don’t know what you want though
T: from my perspective it seems like you create lots of stuff and know what to do?
T: you blog pretty consistently and people like your posts. when you decided you wanted to build a ddr pad you did that. right afterwards you decided you wanted to write a short story and did that too. and now you’re working on [REDACTED]
V: hmm, maybe i’m looking for a bigger-picture direction and view all of those as pretty small-scale projects?
T: maybe
T: but i wonder if you actually already have all the qualities you’re looking for and are just misrepresenting them for some reason

(B, C, and D are three of my housemates)

V: i feel like there hasn’t been enough conflict in my life recently
V: all my friendships are super chill, which is very pleasant but makes it harder to identify growth directions
V: like how friction helps you learn more about yourself?
B: we can fight right now if you want
C: ooh vincent roast session!
V: ok sure
B: how honest do you want me to be?
V: whatever you want, i think it’s quite hard to hurt my feelings
…
B: your jokes aren’t funny, sometimes you’re loud and annoying, you neg people too much
D: here are specific instances X Y Z of those behaviors
B: also you should go to therapy
V: ok thanks, this is all helpful to know
C: you’re more open on your blog than in-person
V: i am working on this but yes i agree it’s an issue
C: you don’t believe in yourself enough
V: yeah it’s hard
C: that’s weird though? i find it easy to believe in you
B: they probably have some unresolved trauma to work through
V: probably
B: is there a reason you don’t have a therapist?
V: i think i can solve my problems without a therapist? because i’m very self-aware and good at working through feelings
V: when i need outside perspectives my mentors usually give very helpful non-technical feedback. also i only really became aware of my feelings last year and have already managed to solve so many problems since then
B: have you read the IFS self-therapy book?
V: no but i’m familiar with IFS and people have done it on me
B: if you’re serious about doing things on your own then you should read it, i don’t think having experienced IFS is a substitute for the book

(an attempt at IFS that turned into something else)

V: you have some really weird self-talk going on
V: your friends and housemates believe in you. you work with the goat research advisor and he believes in you. people who have every right to be upset at you, like your ex and your former best friend, say you’re the most thoughtful person they’ve ever met
V: so why the fuck do you still not believe in yourself?
V1: i don’t know
V1: i really appreciate that people believe in me
V1: but i don’t view that as evidence for being competent?
V1: i guess in my mind the only real evidence that you’re capable of doing hard things is if you’ve already done them
V: yeah but then as soon as you’ve done the hard thing you stop counting it as hard
V1: true
V1: i’m not doing that on purpose though, things genuinely stop feeling hard the moment i’ve finished them
V: i understand, it just sounds like with this approach you’ll never feel capable of doing hard things
V: though i’m starting to think evidence of competence isn’t the real thing you’re looking for
V: that framing doesn’t really make sense? eg. see the line about you don’t become capable of doing hard things before doing them; you become capable of doing hard things by doing them
V1: yeah i know. i think deep down it’s not really about competence. maybe it has more to do with feeling good enough in some broader sense?
V: yeah, you also do this thing where you keep telling yourself that you’re uncreative, unfocused, uninteresting. what’s the point of doing that?
V1: i agree there’s no point, in the sense that telling myself this is not helpful. it’s just a reflex at this point
V1: or unreasonable standards maybe, since i’m always comparing against my most creative / focused / interesting friends
V: okay i’m going to focus on the feeling of inadequacy until i find the child part
…
V: i’m coming up on a bunch of memories of my parents criticizing me as a kid. how did that make you feel?
V2: upset? uncared for? confused? incompetent?
V1: wait i don’t care about my parents’ approval though
V1: they have really dumb values
V1: like if i listened to their preferences i would be living in nyc and working in finance and avoiding {ai, writing on the internet, coliving houses} and spending my free time going on hinge dates
V: i agree that you truly, genuinely do not care about your parents’ approval anymore
V: and i think that’s a great thing
V: quite literally all the best parts of your current life have come downstream of realizing you can stop valuing what your parents think
V: but that wasn’t always the case. the child version of you did need their parents’ approval, even if you no longer do, and we never actually acknowledged that until now
V: those things can linger and continue affecting you for a long time
V: i’m sorry you had to feel that, V2. i think i understand though
V2: the criticism was really overwhelming
V2: i think the actual content was reasonable, but it was delivered in a way that made me think i was doing a bad job and should feel bad about it
V2: and it was always framed as “here’s a thing you’re bad at” instead of “here’s how your life would be better if you improved at this thing” so the incentive to improve wasn’t clear, it was all just problems i had
V1: right
V1: at some point i remember thinking to myself - i’m never going to be perfect; if i fix this thing my parents are complaining about they’ll just find another thing to start complaining about instead, so i might as well just not fix anything
V1: actually i remember having that thought many times throughout childhood
V1: maybe that’s where some of the self-talk comes from - i’m not good enough and never will be, and fixing any individual problem won’t change that so i shouldn’t bother
V: do you guys still believe that?
V2: a lot less than i used to. i feel more cared for now
V1: and i think i’m learning to see flaws in terms of improvements instead of problems
V: i think you guys are doing great!
V: and i will keep checking in on you to see how you’re feeling
V1: do you ever worry that we’re going in circles without actually making progress though?
V: a little bit
V: i’m aware that last fall i wrote about escaping the scarcity mindset and this spring i wrote about uncovering my internalized pessimism, both of which are pointing at the same issue
V1: it seems like this comes up almost exactly once every six months
V: it’s not the same thing every six months though. each time it happens we understand the problem at a deeper level and begin addressing it closer to the origin
V: to me it feels like my heart thrashing against the walls of a cage it’s trapped in. it started off not even knowing there was a cage, and now every once in a while it comes into contact with a wall and learns a bit more about the shape of the cage, places where it’s weak, places where it won’t budge
V1: and?
V: i don’t know, maybe one day we’ll punch a hole in the wall. or we’ll see the cage clearly enough to realize the door was unlocked the whole time and we can just leave
V1: how long will that take?
V1: i’m worried it’s not happening quickly enough
V: me too
V: but look how far we’ve come in a year!

thank you to a large number of friends for helping me organize my thoughts on this topic, including AG AH AY FX GZ JS JX JZ SZ TW and possibly others i forgot

one (very coarse) metric i use to determine if i’m spending time in good ways is to ask myself if succeeding at the thing i’m spending time on would feel life-changing along some dimension i care about? and if the answer is no, success wouldn’t feel that different, the options are either 1) stop spending time on that thing 2) be more ambitious / reimagine the scope of the thing / find the best possible version of it. like, this dance class feels like it’s no longer leading to changes in my physical awareness → i need to be in a higher-intensity environment or my social interactions feel too frictionless right now → i need to hang out with people in less chill ways (working on stuff together, digging into and resolving conflicts, etc)

there are currently four ongoing endeavors that all feel potentially life-changing if they go well (they roughly fall into these categories: research, art, social, health). it feels great to have concrete goals that i actually care about! unfortunately this was not true for most of the last six months

i’ve been thinking about the importance of deeply wanting things. to this day it’s the only condition i’ve found which consistently leads to growth? new environments and friends are great too, but i’ve been in situations where i had both of those and didn’t change much because i didn’t want anything. crushes are really useful because they can provide clear signalling about desires that most people otherwise wouldn’t realize they had

i spent too much time in college trying to be detached / zen / stoic before concluding it was simply not the right approach to life if you care about growth. desires come with a lot of potentially destructive feelings - anxiety, disappointment, grief, and so on - and i think many people (eg. me 3 years ago) don’t have the tools to properly deal with those feelings so they turn to detachment to avoid pain? but you could also simply keep your desires and develop better emotional tools

more generally i think Feelings Are For Doing Stuff, by which i mean that feelings point you towards actions you should take to change your life. unfortunately it’s easy to lose sight of the “taking action” part and instead focus on Experiencing Better Feelings in a way that doesn’t actually acknowledge the changes you could’ve made. many strategies for avoiding pain fall under this category for me - like, the pain might’ve been telling you something extremely useful and subtle but instead of figuring out what that was you just disengaged from it

i feel similarly about many actions that try to manufacture positive emotion out of nothing? the most obvious example is drugs like heroin. some other examples include using “authentic relating events” to feel connection without having a solid basis for it, playing games solely to experience excitement, thinking very hard about your feelings until you’ve convinced yourself that you’re happy. this is basically goodhart’s law for emotional state - feelings are helpful indicators of reality, but only until you start optimizing them to the exclusion of everything else

most of my friends know that i’m a big proponent of the enneagram personality test. like many personality tests, it’s useful as a classification tool when applied properly. however, what really elevates enneagrams is that it provides explanations of each type’s fears / weaknesses / growth directions which i’ve found to be extremely accurate in practice

for instance, i’m type 4, which is described as: self-aware, sensitive, and reserved. they are emotionally honest, creative, and personal, but can also be moody and self-conscious… typically have problems with melancholy, self-indulgence, and self-pity… can look at their motives, contradictions, and emotional conflicts without denying or whitewashing them… willing to reveal highly personal things because they are determined to understand the truth of their experience… makes it easier for them to process painful experiences that might overwhelm other types

and my core problem is correctly identified as: do not pay so much attention to your feelings; they are not a true source of support… one of the most important mistakes Fours make is to equate themselves with their feelings. the fallacy is that to understand themselves they must understand their feelings, particularly their negative ones, before acting

over the last few years i became much more emotionally aware. unlocking new parts of my mind felt so satisfying that i assumed exploring my inner world must be the most important thing in life? but at some point that exploration began to feel oddly stagnant and i realized oh, it’s time to get back to doing stuff

you google what does a stroke feel like reddit and start reading firsthand accounts of strokes. it feels a bit absurd to be reading about these things as they happen to you instead of actually doing something about it, but you’re curious, and you figure that a few minutes won’t make a difference here

your vision darkens and your hearing becomes muffled. or have they really? are these things happening because your brain is losing circulation, or simply because you expect them to happen and have an overactive imagination, or is it some kind of anxiety response? too bad you can never uncover the answer to that question

your housemates call an uber to the er. on the way there they fill you in on transient ischemic attacks and stroke risk factors. you check in and lie on the hospital bed as they do blood work, chest x-rays, ekg, all of which come back inconclusive. you go home and return the next day for a ct scan, again inconclusive. your brain is unremarkable, the radiologist says, how dare they

at last, after twenty-four years, you have received your first real glimpse of the view from halfway down. it’s a little jarring, the possibility that at any moment half your body could collapse. you decide you’ll keep wearing your hospital wristband indefinitely - it’s waterproof and a good reminder that 1) you will die 2) you need to take better care of yourself. you’ve spent a dozen hours over the past year trying on rings without any success; they always felt clunky, impractical, a distraction from putting your hands to work (which you believe is the whole point of having hands); but now you finally possess an item which belongs there

you remember a story L told you - whenever i’m bored in a conversation, i remind myself that i’m going to die and i am choosing to spend my time in this conversation, and then i change the subject to something more interesting. everything you are doing is a choice, so stop choosing things you don’t want! figure out what you really want to do and then go do it. you do not have time to worry about your inhibitions, about whether or not you’ll succeed, about if people will think you’re weird

but back to the emergency room: you’d forgotten how much of it is just boredom, empty space, waiting an hour or two for the next hospital staff to see you. you’d forgotten how doctors are often oddly resistant to conducting follow-up tests or getting to the root of a diagnosis. the effective altruists famously declared that not saving a drowning child was equivalent to shooting the child yourself; what moral framework do these doctors operate by, you wonder, because it’s certainly not that one

and yet you also discover the emergency room can be a place full of joy. getting carted across the hospital is really fun. you have wonderful friends who collectively spend thirty hours waiting with you, doing research and planning, talking to hospital staff (huge thank you to AY MX SZ RH AN <3). when you moved to sf two years ago none of that was true, and for the longest time you complained that all your favorite people were far away; look how far you’ve come. you take bizarre er selfies with your friends and joke about how you might have holes in your heart, which would explain the health situation as well as why you’re currently single

and the ct scan - you don’t know if words can adequately describe the ct scan. iodine entering your veins, ice-cold at first; every blood vessel dilating at the same time; your whole body warming up from the increased blood flow; wondering am i supposed to feel this way? nobody warned me about it; the heat becoming unbearable in your arms, your head, your chest as you become convinced you will actually burn to death; all of it dissipating within seconds of exiting the machine. it is the most incredible physical sensation you have ever experienced, your first time truly feeling how your body is connected via blood and a circulatory system

you could have a stroke at any moment. you are at elevated risk, after all, the bayesian evidence has never been stronger, like the fist of god always hovering over you. but what is there to do, other than to continue onwards, to take care of yourself, to keep doing all the things you like doing with the people you like being around?

lately i’ve been confused about what value i am providing to the world. it seems that i am in possession of a rare set of legos, and i am just a few steps away from building something shiny, but for one reason or another i keep assembling the legos in incoherent arrangements and nothing beautiful ever gets created?

what i mean is something like - i have a lot of useful skills and strengths, enough to be a decent engineer or researcher or coworker, but empirically not enough to be a great one; people generally enjoy spending time with me, enough to want to be friends, but not enough to want to indulge in my craziest ideas or spend the rest of our lives together. when i think about people my age whom i deeply respect, it doesn’t seem like they have better or more numerous tools than me, but somehow they’re able to put everything to use in a way that i haven’t been able to yet. it feels like the difference between linear and compounding returns

sometimes i wonder if i will ever (by my standards) do truly good work or create anything of value? that’s quite important to me. everything i’ve worked on so far has been an unserious demo or a one-off research prototype; nothing with staying power. i thought one of my projects earlier this year might get there, but i was quite ineffective while working on it, to the point that someone had to tell me banging your head against a wall doesn’t help anyone and make me relocate to something else

multiple events occurred recently, each of which upset me deeply, that i’m unable to elaborate on. they have been top of mind for a few days now, and i have become quite anxious; my stomach is constantly in a state of churn and it is taking me from 12:30 until 2:30 to finish lunch each day

yesterday one of my friends casually mentioned things are going really well for T right now. and then i went home and thought for a bit and realized that the most recent time i would’ve agreed things were “going really well” for me was… spring 2022? i’ve been happy many times since then of course, but that was the last time i felt like nothing important was missing

today i attained a full eight hours of sleep for the first time in weeks. i got up a few times in the middle of the night to use the bathroom; each time i woke up there would be a few seconds of bliss - before i regained full consciousness, before i could recall all the events that had just taken place, before the aching in my heart resumed. i slept all night just to experience those sweet twenty seconds where everything was normal again

yeah, tell me again / how it fills the chest, fills the head, fills the lungs

• non-books

  • two articles on economic imbalances (no country ever got rich from tourism, will datacenters crash the economy?)

  • a neuroscience paper on how humans operate at 10 bits per second (the unbearable slowness of being)

  • john green’s personal interpretation of christianity (my religion), which manages to feel more grounded and practical than most atheists, it’s quite remarkable

  • the case for ai standards + insurance + audits (underwriting superintelligence hey look mom i’m in the news)

  • also one of my new favorite poems, i think love is something that happens to other people :’)

• crime and punishment was very slow and verbose but quite good. tldr protagonist commits a crime at the start and is wracked by nerves / guilt / paranoia for the remainder of the story. halfway through the book we get an intro to the best character, police investigator porfiry petrovich, who alternates between psychoanalyzing / confusing / torturing / giving genuinely helpful advice to the protagonist. it's some of my favorite character writing of all time

• cloud atlas was interesting. it’s basically 6 slightly related narratives bound in a single book. pretty impressive how well the author switches between very distinct styles / narrators / historical eras

• did not really understand why letters to a young poet is so well-known? i thought the letters were okay; it was cool to see rilke’s correspondence with someone he’s never met, but ultimately it felt very repetitive (i think almost all the advice can be distilled down to trusting your voice and worrying less about perception?)

• enjoyed parts of anna karenina, especially the first ~100 pages. the writing about marriage problems / jealousy / boredom / cheating is extremely clear and you get a great understanding of why the characters are behaving the way they are. unfortunately there are also some long digressions about the state of russian politics + farming that i did not enjoy. also has some hilarious passages about people roleplaying as artists:

  • They talked about Mikhailof and his paintings. The word talent was often heard as they talked; they meant by it an innate gift, almost physical, independent of intellect and heart, and they tried to express by it all that had been experienced by the artist. It seemed as if they needed to have a term which should express something of which they had not the slightest comprehension, but yet wanted to talk about

  • Vronsky's enthusiasm for painting and the Middle Ages was, however, of short duration; his art instinct was strong enough to prevent him from finishing his painting. His work came to a standstill. He had a dim consciousness that his faults, at first apparently trifling, would grow more and more grievous if he went on

notable things i watched

• bubble was basically a shinkai film but with parkour and bubbles everywhere. really pretty, like look at these frames (and the second one feels like it was ripped directly out of weathering with you)

• cloud atlas was objectively not a well-constructed film (very scattered between 6 plotlines with too much time on exposition, which was kind of inevitable), soundtrack was good though and there are some cool moments where multiple plotlines came together

• watched rick and morty season 8 with my housemates! feels like the show ran out of innovative ideas and has as a result pivoted from hard sci-fi to being more character-driven / focusing on rick’s growth. i like the new focus so i still find the show quite enjoyable, but the shift is not for everyone

• spirit: stallion of the cimarron plot was okay, the horse animations were incredible. during production they sent ~300 people to learn about horses and nail down all the body language + facial expressions? some of the most natural-feeling animal movements i’ve ever seen

other

• finally found a good dentist in sf! first time meeting someone who seemed like they actually wanted to deeply understand the science of oral health - they gave me a lot of helpful things i’d never heard of like saliva ph tests, mouth plaque stains, regenerative peptides for tooth decay, etc

• visited seattle, a bit boring, best part was bainbridge island + museum of art. also uw has the best-dressed undergrads of any school i’ve visited

• redesigned my personal website which also meant reading through all my public writing from the past 6 years to decide what to put on the site. i was surprised by the number of posts from ~2020-2022 which 1) i had no recollection of writing 2) i mostly still agree with and find reasonably coherent. young-vincent was ignorant of many things but also more self-aware than i can remember. also fun to see the focus evolve from being depressed (~2020) to being stressed about post-grad life for all the wrong reasons (~2022) to friends and feelings (~2024)

• played celeste, my new favorite game! best portrayal of mental health and wrestling with different parts of yourself since bojack. music, difficulty curve, storyline were all incredible. protagonist is adorable. extremely cathartic final boss fight

i. strong vs weak links

(this terminology comes from science is a strong-link problem)

early on in most projects i assess the quality of my work based on potential value, and towards the end there’s a shift to assessing quality based on actualized value. this roughly corresponds to the difference between exploring and finishing. one consequence of this shift is that, in the beginning, the quality assessment is dominated by my strongest skills (which tend to be the reason i started the project), while towards the end the quality assessment is dominated by my weakest skills (which are usually the bottleneck to being done)

as a concrete example - i think with ALIGNMENT it was pretty clear that my strengths were prose / narration and my weaknesses were pacing / worldbuilding. my earliest drafts had no semblance of story and were basically just a series of metaphors i was excited to write, so that was what i spent all my time developing and there was nothing else to evaluate. meanwhile, during my last few days of editing, the plot and characterization issues felt so jarring to me that i didn’t have time to worry about anything else. doing a single thing badly can ruin enjoyment (eg. a badly-dubbed tv show) even if everything else is done well

or for a more technical example - in my current work i’m much better at ml engineering than frontend development, so when i’m building features this manifests as having a lot of promising-looking data early on and then spending a long time struggling to integrate that data into good interfaces (in this case the right thing to do may be to hire more people and specialize)

ii. decisionmaking

i roughly bucket skills as either a) domain-specific knowledge (execution) or b) higher-level, like planning and decisionmaking (deciding what to execute on). a) is usually a prerequisite to b) and many people get stuck on a) so they never have a chance to really understand b)

it’s become very clear in the past year that i am a competent executor but rarely pick the correct thing to execute on. once i noticed this i also started noticing that almost everyone spends the majority of their time executing on obviously incorrect things. common examples include: prioritizing smaller tasks to feel productive, distracting oneself with low-level execution to avoid feeling big-picture confusion, copying design patterns without thinking through how applicable they are

(to elaborate on the last point: one pitfall i ran into was designing websites based on how i thought websites “should look” instead of thinking about the exact experience and reaction i wanted a user to have; another was structuring meetings / 1:1s based on meeting structures i was accustomed to instead of working backwards from the state i wanted each person to be in at the end of the meeting)

one thing i appreciate about some of my current collaborators is that they’re quite good at constructively questioning why we are doing something and using that questioning to clarify our plans. that was ultimately one of the main reasons i decided not to go to grad school - with few exceptions the people i met there seemed like they were gaining lots of technical experience and execution ability, but not that much feedback on how to really decide what to do. unfortunate because the latter was the only thing i really cared about anyway

sometimes people ask me how committed i am to working on ai in 5+ years and if i’m worried about skills not transferring. to which my answers are: not committed at all (i’m happy to stay if it continues feeling interesting and important, also happy to jump ship if the field hits a definite wall), not really worried either since i am usually good at acquiring execution skills and expect higher-level skills to transfer. from a growth perspective i don’t think it really matters what you work on as long as it’s aligned with your values and you’re really pushing your boundaries (though of course growth should not be the only thing you care about)

iii. focus

life has been much better ever since i wrote what is the point of blogging? and then proceeded to quit blogging. the (self-imposed) expectation to post regularly was always lurking in the back of my mind, and whenever i had free time or new thoughts i would wonder if i should be writing something. i didn’t realize the extent to which this kind of background process was harming my ability to focus until i cleared it, and i think it’s safe to say most of the things i finished recently would not have gotten done if i’d continued trying to write in the same way

i also developed a few other strategies for focusing. two of my favorites:

• only keep 1 browser window with a small number of tabs (say, 5-10) open, one of which was required to be whatever project i was working on at the time. the project should always be clearly visible every time i open my computer

• i only ever had time to work on ddr during weekday evenings when i was already pretty tired. instead of committing to working on ddr every day, i committed to going to the garage and touching all my wooden tiles every day. some days i would be too tired afterwards to do real work, but most days once i started touching the tiles i would decide to fix something and then an hour would go by