✳flâneur — a map of the web's best reading

Crossing the ocean of my ignorance

gleech.org · 2,388 words · saved by 1 readers

What do you need, to do new things? Imagine you’re a junior researcher; a scientist; a dry-lab scientist; a Machine Learning person. For good and bad reasons you want to publish in Deep Learning, a decade-old bandwagon which continues to steamroll your field. You’re rolling in the deep. How do you get to work? A natural answer is to start at the beginning: go read the underlying mathematics. OK, say you go off and do that. You’re not happy with your understanding: you can feel the aching gap in your knowledge of say linear algebra - that your looking at all those matrices actively concealed something important - but you figure it’s enough for now. It takes a month or six. Can you do new things now? No: you have to learn how to actually implement things. Brilliant people have built easy tools for you, so you learn one of those and reimplement some big papers. This is harder than it sounds, and you actually don’t manage to reproduce half of the results. You add 3d6 unease and self-doubt.

...postpone reading Nietzsche for the time being; first study Aristotle for ten to fifteen years. – Martin Heidegger 1 As a researcher, either you won’t understand something and you will feel stupid and like a worm, or you will understand something and think it’s too trivial and hence still feel like a worm. – Simon Peyton Jones I was much further out than you thought And not waving but drowning. – Stevie Smith What do you need, to do new things? Imagine you’re a junior researcher; a scientist; a dry-lab scientist; a Machine Learning person. For good and bad reasons you want to publish in D ee

Explore this link on the map →

related reading