poast office i

intro

there’s really no time like the present to start writing. i’ve been itching to do this for soo long (i bought myself a new domain handle on christmas day 2025 and still haven’t posted). so here we go

what is this?

i feel actually quite shy! but have a huge amount of pent up creative energy. i’m excited to use this as a creative outlet. social media (mostly twitter + linkedin for me) is a liiittle bit too public for me to feel quite free to experiment and express myself, so i wanted to come back to my newsletter

i want this to be lighthearted, fun, experimental, silly, and free-flowing. this is not meant to be particularly finely-crafted, so i will make mistakes. but i think practice is the only way you can get really good at something

because of the nature of my work, my interests, my current surroundings, i’ll probably post about ai, tech, and memes

but other things i’m interested in include films, comedy, philosophy, literature, politics, welfare, and more. i’ll be balancing questions i’m genuinely puzzling over with unserious memes to fuel my creative excitement here lol

and at the end of the day, i’m interested in making friends on the internet, so i’m going to try this and see what happens

food for thought

this section is going to be about random thoughts that pop into my head over the week. think of these like a list of questions i’d like to think about more. talk to me about these if you feel so inspired!

• i’m kinda interested in the idea of if chatgpt/claude will be able to replicate my writing style after it has a few newsletter posts (in this style) for context. so far, when i’ve paired with llms on creative writing style (or whatever writing style you call my shit-posting on the internet as), they haven’t really nailed it. it’s still somewhat to formal, or doesn’t flow the way i want, or they replace too many words or too few works (as phrases or via full sentence redesign)

• i think packy mccormick maybe tried this in 2023/2024 when models were just starting to really write creatively. curious to know if this has worked for any writers
  

• are we playing enough? by play, i mean unstructured recess time kids have. as adults, we are so focused on structured hangout time. go for drinks, grab a coffee, share a meal, go for a walk, do a thing. what about games of hide & seek, make believe / role playing, blanket forts, paper plane competitions, friendship bracelets, mock wrestling?

• is this part of the reason why people like having kids? it encourages adults to step into their child-like playful sides?
  

• on perfectionism: does someone get better at something by

  • thinking a lot about it and crafting the perfect single output?

  • or do you get better with constant creation and practice? practice iterating on that output? on the creative process?

• i go back and forth. sometimes i think the editorial process is so important for actual learning and taste-making. i am pulled by the idea of thoughtful craft, intentionality, and serious study of the greats

• but generally, i think the latter is more correct, at least maybe for me. which is why i’m experimenting with posting more regularly.

tidbits // poasts

this is random list of (mostly non-serious) thoughts i’ve had this week. some reflections on fractals of life i see throughout the week

• i’m interested in understanding how sudoku puzzles can increase in difficulty without being larger in size (2-star to 4-star puzzles are still 3 by 3)

• is it that some of them only have one solution while others have several solutions?

• i wonder what the strategies are for solving puzzles? how do people train for this? just like there are different strategies for doing mental math, some being more intuitive to some people than others, what are the ways to practice sudoku strategy?
  

• dang i remember when image models used to be so bad. and now they are so, so good it’s insane to believe! this was the image of my first newsletter post and now my whole newsletter rebrand is model-generated:

• as a small appreciation of technological progress in one of the most pedestrian ways (i told you this would be a random poast section), i’m so pleased that starbucks <insert any hot drinks company> serves drinks that are the perfect drinking temperature. it used to be the case that drinks were too hot and you had to artificially wait for them to cool down before drinking, subverting our instinct for instant gratification. even the wework coffee dispensers are the perfect, instant drinking-temp. sign of technological progress

more to come, maybe i’ll double post this week

on my reading list

articles

• machines of loving grace (finally) - dario amodei

  • dario’s essay on how ai could benefit humanity in 5 buckets: bio/med, neuro, global welfare, governance, personal productivity

  • some interesting insights:

    • great discoveries usually come from a handful of exceptional researchers

    • with ai, discovery rates could increase dramatically as o(1000s) researchers are deployed in parallel

  • he believes in a rapid but physically grounded acceleration (i.e. not ‘singularity’ intelligence explosion)

  • progress timelines in these domains are 5-10 years from the point of agi/asi

• on sincerity - jcs

  • tldr: sincerity is about orienting yourself towards doing what you genuinely consider is right, instead of acting performatively and explaining your actions through post-hoc rationalization

  • instead of “How can I justify what I already want to do?”, sincere people ask “What should I actually do?”

  • sincerity is about being honest with yourself, which is not the same as being honest with other people. the former is more important than the latter

  • sincerity is a way of engaging with life in which your beliefs, motivations, and actions are genuinely aimed at responding to reality rather than protecting your ego, identity, or social performance. a sincere person honestly asks what they should do, allows themselves to be changed by the answer, and integrates competing motives instead of hiding from them. sincerity requires seriousness, intellectual honesty, and self-awareness, but it does not guarantee moral correctness. its purpose is not a way to justify your actions, but instead a way to attend honestly to the real world and your true beliefs about what it means to be a good person

• harness engineering for self-improvement - thinky machines

• claude’s new constitution - anthropic

• alignment remains a hard, unsolved problem - lesswrong

• training llms for honesty via confessions - openai

• will it become harder to oversee ai systems - aisi

• meditations on moloch - scott alexander

books

• seeing like a state - james scott

• picture of dorian gray - oscar wilde

• the infinity machine - demis hassabis

• a thousand brains - jeff hawkins

• how minds change - david mcraney

• a brief history of intelligence - max bennett

• the social leap - william von hippel

• moral tribes - joshua greene