Assorted Media

compilation of learning resources & nice reads (mostly classics)

I’ve compiled some of my favorite media for learning the things I’ve found most relevant to the nice life and the nice discourse. A lot of stuff is missing because I made this in an afternoon, and because I don’t remember many things from a long time ago.

If you have similar tastes, please leave recommendations in the comments!

ML/deep learning

1. Top-down, FastAI book https://course.fast.ai/Resources/book.html

2. (But don’t actually use FastAI) https://lightning.ai/docs/pytorch/stable/

3. Foundational theory, 3B1B https://www.youtube.com/playlist?list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi

4. Bottom-up, Dive Into Deep Learning https://d2l.ai/chapter_preliminaries/calculus.html

5. Modern & more thorough, UvA course https://uvadlc-notebooks.readthedocs.io/en/latest/index.html

6. NLP, HuggingFace course https://huggingface.co/learn/nlp-course/chapter0/1

7. RL theory, Decision Making Under Uncertainty https://github.com/JuliaAcademy/Decision-Making-Under-Uncertainty

8. RL practice, HuggingFace course https://huggingface.co/learn/deep-rl-course/unit0/introduction

9. (Idk about audio & vision)

10. Interpretability

    1. “Circuits” https://distill.pub/2020/circuits/

    2. “Superposition” https://transformer-circuits.pub/2022/toy_model/index.html

11. Research collections

    1. Anthropic https://www.anthropic.com/research

    2. GI https://imbue.com/our-work/

    3. Deepmind https://deepmind.google/technologies/

    4. OpenAI https://openai.com/research

12. Kochendorfer’s duology

    1. Algorithms for Optimization https://mitpress.mit.edu/9780262039420/algorithms-for-optimization/

    2. Algorithms for Decision Making https://algorithmsbook.com/

13. Some of my personal interests that aren’t that hot

    1. Deep implicit layers, http://implicit-layers-tutorial.org/introduction/

    2. GNNs, https://distill.pub/2021/gnn-intro/

    3. Differentiable programming https://arxiv.org/abs/1907.07587

Programming

• I think learning new programming languages very different in paradigm (not just syntax) is a better way to improve at programming in general than deep dives into a particular language. Do focus on one or few languages to be productive in though. Probably TS + Python + Rust

• Clojure standard library + APL/K primitives (which have a lot of intersection) are basically all you need, and indeed are pretty standard across stdlibs/big utility libraries, so old lisp tricks books are mostly distractions imo.

• I don’t have books to recommend for the popular languages like TS/Python/Rust. Reading docs+code is probably faster if you can program in general.

• Basics of computational complexity is important but idk a good resource; maybe Cormen et al.: Introduction to Algorithms

1. The C Programming Language (super well written and organized; helps for reading & writing any C-style syntax language including JS/TS and Rust; pascal/fortran/ruby/python style is basically just pseudocode and doesn’t need a primer)

2. Haskell https://www.haskell.org/tutorial/intro.html (as a note, language books & docs written by creators tend to be good in general)

3. Typescript

4. Type-level programming in TS https://type-level-typescript.com

5. (and then don’t get tempted by Idris/Coq/Lean/Agda etc unless you’re a mathematician; also don’t pay for the course; you can figure out how to do the stuff in the paid chapters from other free resources; maybe read Maguire: Thinking with Types too though, for type-level programming in Haskell)

6. K https://xpqz.github.io/kbook/Introduction.html (learn *before* numpy, since numpy is an Iverson ghost, i.e. derived from APL, and k is like consistent simplified apl with dictionaries)

   1. Alternatively, https://xpqz.github.io/learnapl/intro.html for Dyalog APL

7. Python (+ pydata; not polars bc RAPIDS is still faster than polars and chose numpy API, unless you want to use nodejs-polars too or work mostly without a GPU I guess)

8. Lisp (SICP, probably, or Clojure docs + Rich Hickey conference talks; learn *before* JAX for the same reasons as K/APL before numpy; see code as syntax trees)

9. JAX (or Julia; but learn LISP first in either case)

10. Optionally BQN https://mlochbaum.github.io/BQN/doc/index.html, to write subtle AF get-fired-immediately pytorch/jax code with combinator calculus

11. Optionally Scala ZIO or https://effect.website (I think TS Effect has better docs. Scala Cats https://typelevel.org/cats/typeclasses/alternative.html has really good docs but Cats Effect less so last I checked; and ZIO is basically 1-1 with TS Effect and Cats Effect isn’t)

12. Optionally Rust if you actually need a low-level/systems programming language to use

13. Nice to read

    1. JSoftware APL paper collection; especially Notation as a Tool of Thought https://www.jsoftware.com/papers/tot.htm

    2. The AWK Programming Language; classic and well-written and actually useful on remote terminals. MAWK is often faster than Hadoop. https://adamdrake.com/command-line-tools-can-be-235x-faster-than-your-hadoop-cluster.html

    3. suckless http://suckless.org/philosophy/

    4. No Silver Bullet http://worrydream.com/refs/Brooks-NoSilverBullet.pdf

    5. Out of the Tar Pit https://curtclifton.net/papers/MoseleyMarks06a.pdf

    6. The Bitter Lesson http://www.incompleteideas.net/IncIdeas/BitterLesson.html

Math

Useful

1. Calculus, 3B1B https://www.youtube.com/playlist?list=PLZHQObOWTQDMsr9K-rj53DwVRMYO3t5Yr

2. Linear algebra, 3B1B https://www.youtube.com/playlist?list=PLZHQObOWTQDPD3MizzM2xVFitgF8hE_ab

3. Differential equations, 3B1B https://www.youtube.com/playlist?list=PLZHQObOWTQDNPOjrT6KVlfJuKtYTftqH6

4. Strang: Introduction to Linear Algebra, maybe + the MIT OCW course https://ocw.mit.edu/courses/18-06-linear-algebra-spring-2010/

5. Cover and Thomas: Elements of Information Theory

   1. Or Stone: Information Theory: A Tutorial Introduction

6. Ross: A First/Second course in Probability

7. Barabasi: Network Science

8. Number theory (for cryptography) https://crypto.stanford.edu/pbc/notes/numbertheory/

9. Skim the Princeton Companion to Applied Mathematics

Classic

1. Skim the Princeton Companion to Mathematics

   1. Or Bourbaki: A Panorama of Pure Mathematics

2. Tao: Analysis I&II

   1. Or Rudin: Principles of Mathematical analysis & Real and Complex Analysis

3. Herstein: Topics in Algebra

4. Group theory might be better from videos than books; e.g. https://www.youtube.com/playlist?list=PLwV-9DG53NDxU337smpTwm6sef4x-SCLv or https://www.youtube.com/playlist?list=PL8yHsr3EFj51pjBvvCPipgAT3SYpIiIsJ

5. Bartle: Elements of Integration and Lebesgue Measure

6. Nielson/Chuang: Quantum Computing and Quantum Information

7. Peliti: Statistical Mechanics in a Nutshell

8. Smullyan: To Mock a Mockingbird

Less Classic

1. Goldblatt: Topoi

2. Stein and Newman: Spin Glasses and Complexity

3. Brualdi & Cvetkovic: A Combinatorial Approach to Matrix Theory and Its Applications

4. Priest: An Introduction to Non-Classical Logic

Recent (post-Enlightenment) philosophy/psychology

1. 1000 Plateaus

2. Mipham, Beacon of Certainty

3. Meditations on the Tarot

4. Haley: Strategies of Psychotherapy

5. Reich: Character Analysis (don’t accept too directly)

6. Taleb: Incerto (one book is probably enough, they’re kind of redundant; idk which one though)

7. Becker: Denial of Death

8. Alexander: Notes on the Synthesis of Form

9. Bateson: Steps to an Ecology of Mind (don’t accept too directly)

10. Weiner: Cybernetics

11. Shinzen Young: Science of Enlightenment (buddhist practice for nerds)

12. If you have a lot of time

    1. von Balthasar: Herrlichkeit, Theodramatik, Theologik

    2. Sloterdijk: Bubbles, Globes, Foams

    3. Jung: complete works (don’t read Jung out of context); The Psychology of the Transference is a good peek at some content that hasn’t leaked far into pop culture

    4. Land: Fanged Noumena + xenosystems.net on archives; Against Orthogonality and https://etscrivner.github.io/cryptocurrent/ are nice short reads

    5. Treasury of Knowledge

Other nerd classics that I liked

1. Quantum Computing Since Democritus

2. The Art of Doing Science and Engineering

3. Norvig: Artificial Intelligence Programming

4. Godel Escher Bach

5. On Lisp

6. Jaynes: The Origin of Consciousness in the Breakdown of the Bicameral Mind (bc it’s interesting, not bc I think it’s super convincing)

Classics (both old and recent)

1. I Ching (Thomas Cleary’s translations are ok, but read the original with a word-by-word translation https://www.yellowbridge.com/onlinelit/yijing.php even if you only know modern Chinese, even though a lot of the word translations are wrong)

2. Dao De Jing (Every translation I know about sucks. Don’t read it if you can’t read Chinese. https://www.yellowbridge.com/onlinelit/daodejing.php)

3. Philokalia

4. Zhuangzi (Burton Watson’s translation is good)

5. Faust (I recommend watching the opera instead of reading the script https://youtube.com/watch?v=Q7oVDMVXyqw )

6. Joyce: Ulysses

7. Lolita

8. Journey East and Red Chamber Dream (only if you can read Chinese; I don’t like the other 2)

9. Six records of a floating life

10. Paradise Lost

11. The plum in the golden vase

    Why not

    1. I hate the Illiad/Odyssey

    2. Plato seems too outdated to be a good read unless you want to cite/dialogue with him

    3. I’m too dumb and unenlightened for the Upanishads; also the translations probably suck

    4. Gilgamesh translations probably suck suck suck; interesting as contrast against modern consciousness though

    5. Same for Beowulf

    6. Tale of Genji is Red Chamber Dream but worse (maybe), & about Japan, and translations suck

    7. I remember Divine Comedy being mediocre but I read it when I was pretty little so maybe I just didn’t understand it

Other mentions

1. I like Umberto Eco’s novels but they’re neither “classic” nor “philosophy/psychology”

2. Same with Nisio Isin’s books

3. Same with Don Norman’s books

4. Neal Stephenson books basically capture a lot of tech zeitgeist, so they’re useful, but I don’t especially like them

5. I like Marquis de Sade’s books but a lot of people that like the other books I like won’t

6. Descartes, Kant, Hegel, Nietsche, etc., including Guenon, Bataille, Girard seem too long to be worth reading completely. Wikipedia or plato.stanford or conversations with people who’ve actually read them are nice though.

7. I’ve liked pretty much everything I’ve encountered that mentions Peer Gynt (especially Schnittke’s ballet), but haven’t watched the play.

8. I suspect Don Quixote is pretty good but I haven’t read it

9. Voltaire is funny but his philosophy is pretty reddit-tier by now

10. I feel like you don’t actually need to read Freud or Marx since everybody else talks about them so much. People talk about Jung a lot too but it’s too often too distorted.

11. A lot of anthropology field studies are nice and “different”

    1. Notes on the Balinese Cockfight is super classic

    2. Also https://ranprieur.com/readings/preconquest.html

Other lists

• Patrick Kidger https://kidger.site/thoughts/just-know-stuff/

• Patrick Collison https://patrickcollison.com/bookshelf

• Self-teach pure math https://marktomforde.com/academic/mathmajors/textbook-suggestions.html

• 0xparc zero-knowledge cryptography https://learn.0xparc.org

• awesome https://github.com/bayandin/awesome-awesomeness