No single big topic today — I've been jotting thoughts that floated up lately, they piled up, so I'm putting them in a blog.
- Last time I wrote about making your own variables or models for decisions. So if I were to build a model myself, what would that look like? Off the top of my head: something like using a general-purpose ML model in Python. You need a fair amount of data, but for small analyses Google Colaboratory is enough — playing with a reference-level model sounds fun. Treat me as someone who dabbled in Python briefly when it was trendy (I can't really program in any language). I asked ChatGPT for a step-by-step if I were to build a model. Classic ChatGPT — very clear. I understood the flow, of course, and also more essential things that sit a bit off-axis: what you'd gain if you could do it, and where the gap would still be versus people called ML engineers even if you finished everything they listed. Commercial programming books often feel verbose in places, or skip "why do I need this?" and just say memorize it — so this went much deeper. Digging through code, I eventually realized — oh, if you just prepare the data and instruct, it'll do preprocessing and modeling for you. I didn't need to learn to write the code myself. Someday I want to try. Though I also suspect there's fun in writing the code.
- I often remember old things. I've written memories on the blog a few times, but I more often write recent thinking. Still, lots of memories surface now and then, and I think I'll start putting those on the blog. Whether that changes how I hold them, deepens attachment, or sublimates the feeling — I don't know. But every time one comes up I often feel "why did I remember this, what a nuisance," so ideally I'd sublimate it and, when it returns, switch immediately: "I already wrote this on the blog and sorted it — no need to think about it. Memory, go away."
- Lately I've had time, so I've been drawing. I bought a fun anatomy book at a bookstore and following it to draw faces and bodies was really enjoyable. Adults who draw are a minority, I think — but whether you're good or not, just trying the act of drawing is something I'd recommend. Related: I happened to find Hiroshi Mori's essay collection "How Should Humans Think About Various Problems" at a bookstore and bought it. It starts from interesting opinions and questions — why do people constantly ask "put it more concretely" but never "put it more abstractly"? Why is abstraction sometimes treated as bad? There's too much concrete information in the world — and somewhere in it he says art appreciation lets you get that abstract sense of "somehow good." When I'm drawing, setting aside skill, I often hit moments of "somehow good." The reason I dragged Mori in at all: because of that, try drawing.
Honestly — enough with drinking outings. On days off I want play like drawing, making manga, shooting film, building a chair from leftover wood. Fine if it's a bit messy. It sticks in memory.
P.S. Mori also wrote "Anti-Organizing Method" — humans naturally live in chaos, so you don't need to organize — a good kind of contrarian fun. It cheers you up: you don't have to be put-together. Every time I recall the title I search Mori on Amazon; he has way more books than I thought — like ten times as many.
- There's a word "attention economy," apparently. Attention is scarce, as on SNS, so there's an economy of fighting over it. Whoever keeps inventing these tidy new words — scholars? — I thought again how shallow that whole buzzword genre is. There was "token economy" too, right? I'm not looking up definitions anymore, but maybe it means an economy where instead of yen for an hour of work you get 50 tokens, and 50 tokens buy a meal? If a community prices a meal at 50pt and Umaibo at 1,000 tokens, then in that community people eating Umaibo are the rich. Writing this myself feels ridiculous. If there are creepy people who'd high-five each other calling this a cool idea, I'd keep my distance.
- The best affiliates are influencers. Meaning: if you want to sell a product, become famous yourself. Meta and Tesla too. That makes the business personal — I think competing on technology and service value is more beautiful, and I prefer that side. That's all. I've been thinking that again lately.
That's it for today. I also have notes that are just words — "valuing highly reproducible know-how," "considering means putting something in the option set" — but I forgot the mental state when I wrote them and don't feel like expanding them, so this is enough for today.
Tweet© 2024 ZUUHE