2026 Q1 update

fav quarterly post to date

since the last update my reading diet has shifted heavily towards papers and short stories as opposed to books. lots to recommend!

• 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: