Reading Path

A starting map for the site author and for readers who want to follow the sources behind the argument.

Site frame

Little Worlds
The local framing essay: bounded forms, microcosm and macrocosm, and the LLM reversal from small units to large-scale behavior.
Little Worlds and Large Language Models
The first public post that ties the domain name to tokens, equations, scale, and the theological movement from small signs to large orders.

Eugene Wigner, The Unreasonable Effectiveness of Mathematics
Use this for the central analogy: mathematics fits the physical world, and LLMs invite wonder that mathematics also fits the linguistic world.
John Paul II, Letter to Artists
Use this for human creativity, beauty, and the idea that human making participates in a created order.
Antiqua et Nova
Use selectively for AI as a product of human intelligence and for distinctions between human intelligence and AI.

Stanford CS324: Introduction
Defines a language model as a probability distribution over token sequences.
Stanford CS324: Modeling
Explains autoregressive modeling and generation as a sequence of probabilistic choices.
Jay Alammar, The Illustrated Transformer
An accessible bridge into transformer architecture and attention.

Mikolov et al., Efficient Estimation of Word Representations in Vector Space
A key source for Word2Vec and the idea that words can be represented as learned coordinates.
Stanford NLP: GloVe
Shows how co-occurrence statistics leave semantic traces in vector space.

Vaswani et al., Attention Is All You Need
The transformer paper that makes attention the central mechanism for modern language models.
Kaplan et al., Scaling Laws for Neural Language Models
Use this for the mathematical regularity of model improvement across data, model size, and compute.
A Mathematical Framework for Transformer Circuits
A route into mechanistic interpretability: complexity invites investigation, not superstition.